Skip to content

feat(agent-server): projects API, coordinator tools, board groups and artifacts - #6602

Open
enyst wants to merge 244 commits into
mainfrom
claude/project-thread-xwkk9b
Open

enyst wants to merge 244 commits into
mainfrom
claude/project-thread-xwkk9b

Conversation

@enyst

@enyst enyst commented Nov 7, 2026 •

Copy link
Copy Markdown
Member

HUMAN:

Many harnesses struggle with this: a way to move up from “chat UI” to a real UX/UI for agents. Here is an example screenshot.

Also, you can search on GH a set of PRs talking about Insider Cat, which is a different take on something similar. (they’re drafts in openhands repo, agent-sdk repo)


AGENT:

Live verification evidence — 2026-11-16

Strong local source-level evidence for the Projects core, not proof of every feature in this large PR pair. The SDK and Canvas #19495 were exercised together with real DeepSeek models and a real browser in Paper. Canvas used the locally built TypeScript client from the SDK branch, not published 1.62.0 packages.

  • Core Projects: live API/browser checks covered CRUD, membership and board transitions, coordinator/worker tasks and follow-ups, approval refusal, independent git worktrees, artifacts, memory revision conflicts, and persistence across an owned restart. [Full feature-by-feature ledger, exact revisions, failures, controls and logs](https://github.com/enyst/openhands-evidence/blob/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/README.md).
  • The two initial recipe failures are investigated, not hidden: F36.mention lacked the coordinator prompt preset and later rejected valid dashed UUID links; F37.ts-client assumed reactions were the only experimental flag. Corrected complete families passed F36 34/34 and F37 12/12 at SDK 2a4b4e20c1d253bc56df2591e16c9166d5dd99d7, plus 3/3 fresh mention and 2/2 client repetitions. The final merge-resolution source b9bb0e962c6f3c3f01a37f7bccb62677da8cdeb4 passed F37 12/12, two fresh mention repetitions and 121 targeted tests, retaining upstream's independently fixed F37 unchanged. The remaining F36 recipe correction was merged into this branch via #6908. [Before/after runs, negative controls and exact source provenance](https://github.com/enyst/openhands-evidence/blob/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/README.md).
  • Real browser mention proof: the coordinator answered the unchanged question with a worker link; clicking it opened the exact worker and its real done reply. [Observed link target](https://github.com/enyst/openhands-evidence/blob/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/checks/gap-browser-link.json), [executed click/resulting URL](https://github.com/enyst/openhands-evidence/blob/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/checks/gap-browser-click.json), [worker answer](https://github.com/enyst/openhands-evidence/blob/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/checks/gap-browser-worker-answer.json). Screenshots illustrate the states; these executed checks establish navigation.
  • Limits remain explicit: PDF preview stayed blank; Voice needs its separate compatible realtime backend/provider/device setup; long-history restart replay and several lifecycle/adversarial cases remain unproven. No claim of published-package, Docker/binary, deployed Cloud, cross-platform or whole-PR-head CI validation. The ledger is a dated snapshot of the named commits, not a fresh run of every later commit.

Selected Paper screenshots

Projects board: untitled conversation identifiers fit their columns after the separately delivered Canvas wrapping fix.

[Paper Projects board with wrapped conversation identifiers](https://raw.githubusercontent.com/enyst/openhands-evidence/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/paper-wrapped-board.png)

Coordinator mention: the real model names its worker with a clickable conversation link.

[Paper coordinator answer with its clickable worker mention](https://raw.githubusercontent.com/enyst/openhands-evidence/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/paper-coordinator-mention.png)

After clicking: the linked worker opens with its real answer. Browser snapshots use Canvas 6a8331920af5b4c505d42084764eaa33ba420c7f and SDK launcher source 2258f2e38fe1317d19bc80389c5aacf6d2bc9733; detailed build provenance is in the follow-up ledger, separate from the later direct API recipe runs above.

[Paper linked worker with its actual done reply](https://raw.githubusercontent.com/enyst/openhands-evidence/fe6606c8bebd962267c993d4c82d783da87c3a27/.pr/OpenHands/software-agent-sdk/pr-6602/2026-11-16/gap-followup/paper-worker-opened.png)

Why

Once a user runs several agents at once, a chat window stops being the right shape: what they need is one conversation to talk to (a coordinator) and a view of every other conversation grouped by what it needs from them. Insider Cat (OpenHands#19495, #6605) already makes a saved, tool-using conversation the user's companion, but it has no project to belong to, and it starts and steers workers by calling the API from its terminal, which no UI can render or check.

This PR gives the Agent Server a Project resource and gives a coordinator typed tools to follow and steer its conversations, so every front end (Canvas, the Insider App, CLI) can build the same board on one open API.

Summary

  • Projects is an experimental feature (projects in /api/settings/experimental, off by default). While it is off, every /api/projects route answers 404 (following the switch at once), new conversations get no project tools even when their agent spec lists them (finalize gains withheld_tools), a member's child is an ordinary child (no project tag, worker tools, guidance or memory), and no update or copy of the user's messages reaches a coordinator. Projects on disk are kept. A coordinator started while it was on keeps its tools, and after the switch each project tool call tells it Projects is turned off and not to retry. Agent-server tests run with it on unless marked projects_off.
  • Projects API under /api/projects: create, list, get, patch, delete; GET /conversations lists members grouped by attention; PUT/DELETE /conversations/{conversation_id} change membership. Records live in projects.json, with the same atomic-write and file-lock store as workspaces.json; a corrupted or newer-schema file answers 409 and is never overwritten.
  • A project can start its own coordinator: POST /api/projects takes an optional coordinator: {working_dir, agent_profile_id?, title?}, plus the start-request fields a client such as Insider needs for its own coordinator: conversation_id, tags, conversation_runtime, agent_launch_additions, confirmation_policy, security_analyzer and max_iterations (the server still adds the project tag and the tools; a reserved id that already exists is 422). The server starts the coordinator in that workspace from the given profile (or the user's active one) with the coordinator and worker tools, tags it with the project, leaves it idle with no first message, and returns the project with coordinator_conversation_id set (201). If the coordinator cannot start (no profile, a launch failure), no project is kept and the request answers 422; giving both coordinator and coordinator_conversation_id is a 422 too. Passing an existing coordinator_conversation_id, or none, works as before.
  • Membership is the project conversation tag (PROJECT_TAG_KEY in the SDK). The conversations a project member starts with start_child_conversation get the project tag, so they join the project with no extra call. They get no other tag of the parent's, such as Insider's insidercat/insiderrole.
  • Confirmation-safe messages on the existing send route: POST /api/conversations/{conversation_id}/events takes an optional refuse_pending_confirmation. With run, it never runs a conversation that is waiting for confirmation, since LocalConversation.run() treats that run as approval: a conversation already waiting answers 409 and nothing is saved. A conversation the user paused, or one stopped on an error or stuck, is held for the user the same way (409 with its own detail and the status in a status field, which the project tool reads), so another conversation's message cannot resume it, and the check itself is a new option on EventService.run()/send_message(), decided under the run lock. With run, the response now also says what happened: run is started, queued (a run was already active), waiting_for_confirmation (message kept, not run) paused, error, stuck or not_run. There is no project-only message route; Engel asked for one route every client already uses. Refusals are small typed exceptions, ConversationAlreadyRunning, GoalAlreadyRunning and HeldForUser (with the status), still ValueErrors with the old messages, so callers catch a type instead of comparing strings. Which conversations may run is one rule, admission_refusal() in event_service.py, which returns the refusal to raise: for another conversation, only an idle or finished one; actions left unconfirmed count as waiting for confirmation whatever the status says; and its from_user case (Voice's send_message_if_idle in feat(agent-server): realtime voice broker for Insider controller conversations #5180) may also resume a pause. A stuck conversation stays stuck when a guarded message is saved, a run that never started is reported not_run, and the run replacing an interrupted ACP prompt may resume the server's own pause.
  • Workers are normal conversations: a conversation a project member starts does not copy its parent's agent. It starts from the project's worker_agent_profile_id (new optional field on project create/patch, 404 for an unknown profile), or else the user's active agent profile. It keeps the parent link, the parent as sender of its task, worktree isolation, and the parent's confirmation policy, plugins, hooks and secrets. It does not take the parent's agent, launch additions or client tools. The server adds write_artifact, list_artifacts, read_artifact, read_project_memory and update_project_memory to it. Children outside projects still copy their parent, as before.
  • Coordinator tools in openhands.tools.project, registered by the agent-server like start_child_conversation: list_project_conversations, message_project_conversation (a thin tool over the send route) and read_project_conversation. They read the project from the caller's tags at call time, so a conversation only reaches its own project's conversations.
  • Coordinator and worker roles are prompt presets. The SDK prompt registry has two more compositions, project_coordinator and project_worker (context/prompts/sections/project.py, pinned by snapshots). The coordinator keeps no coding-workflow sections; it gets ROUTING, BRIEFS, COORDINATING_WORKFLOW, WHAT_YOU_HEAR and TALKING_TO_THE_USER, and a persona replaces only its SOUL. A worker is the default coding agent plus WHO_IS_SPEAKING and YOUR_REPORT: it asks in its report instead of "consulting the user", treats a request whose origin a summary lost as the user's for what to do but never as approval, and never uses git stash in a shared worktree. prompt_preset is a field of Agent, OpenHandsAgentSettings and OpenHandsAgentProfile. The server assigns the role at launch (AgentLaunchAdditions.prompt_preset) to every coordinator and worker it starts, whatever profile it runs, and renders the project (name, goal, who started the worker, the MEMORY.md index capped at 6,000 characters) in a dynamic <PROJECT> block (AgentLaunchAdditions.project_context). On first use the server creates two predefined profiles from the user's active one, project-coordinator (terminal, file editor, task tracker, plus the project tools) and project-worker (the user's tools without start_child_conversation), used when a project names no profile. The section-by-section design is at https://enyst.github.io/arch/project-agent-profiles.html.
  • Workers know their coordinator: a message sent with sender_conversation_id (which message_project_conversation fills in) records sender: "conversation:<id hex>" on the saved MessageEvent and opens with a line telling the receiving agent who sent it (the coordinator, another conversation of the project, or another conversation) and that it approves nothing on the user's behalf. A sender in a project can only message its own project's conversations (422). A run asked for by a sender is always guarded. Both fields live on ConversationMessageRequest, the events route's body, not on the SDK SendMessageRequest that is also initial_message. UIs can then fold such messages as instructions instead of showing them as the user's words. Every child's task message records its parent as sender, and a project worker's brief opens with the same [Sent by ...] line.
  • Conversation mentions: tool results and server headers name a conversation as a Markdown link, [Title](conversation:<id hex>) (conversation_mention() in the SDK). The title is cut to one line with no brackets, parentheses, quotes or backslashes, so it cannot end a link or a header early. The coordinator can copy these into what it tells the user, and UIs render them as links. The headers' opening phrases ([Sent by , [Update from , [Reply in , [Message from the user in ) are unchanged. The conversation_mentions experimental feature (the agent's read-only find_conversations and read_conversation tools, their guidance, verify family F38) came from #6851, now merged into main.
  • Message edits (#6876) are merged into main now. The reaction lookup fix (#6901) is merged too, so this PR carries no copies of other PRs anymore. Decision cards (ask_user, #6852) are merged into main now; this PR only adds the Projects path for workers' questions, and its closing rule (the message saved after a tap's answer closes the question, whoever delivers it).
  • A worker's question card reaches the user through the coordinator (with projects and ask_user on): when a member's turn that the coordinator started ends on an ask_user card, the board counts the card as the turn's output (turn=asked, ask is the question, no judge), and the coordinator gets an update with the question and options. A coordinator started while ask_user is on gets ask_user with a for_conversation field; when the user answers that card, by a tap or in the chat, the server records the same answer on the member's card (AskUserAnswerEvent) and runs the member, with the coordinator as sender, through the same confirmation guard as other server deliveries. Only the user's own answer in the coordinator does this; the coordinator's messages cannot answer a member's card. Tests: tests/agent_server/test_project_cards.py. Tool-call approvals are not part of this.
  • The coordinator hears back and stays informed (project_wake.py): when a member finishes a turn that the coordinator started (its task message or a follow-up), the coordinator gets one message with the member's final message, an "Update from" line and the member as sender, and runs on it, without approving a pending confirmation. Each message the user writes in a member ([Message from the user in ...) and the member's final reply to it ([Reply in ...) are copied to the coordinator too, without running it: they keep it aware of everything said in the project, and running it on each would tempt it to forward them back. The board ignores these copies, so they neither reopen the coordinator nor start a turn. The coordinator does not see members' events; read_project_conversation (over GET /api/projects/{id}/conversations/{conversation_id}/messages, the latest messages and finish texts) reads more on demand. When a member the coordinator started stops on the user instead (waiting for approval, an error, stuck), the coordinator is woken with a line saying what it waits on, once per stop; a pause the user made is saved for its awareness without a run. A delivery that finds every run slot taken waits for a free one in the background instead of leaving the update unread, and a failed save is retried. When the server starts, it delivers any update a member's coordinator-started turn should have sent but did not (the server stopped first, or every attempt failed), from the member's event log; turns the user started are not replayed. A run the server's own shutdown cut off is marked as such: it comes back paused rather than errored (the interrupted tool's error says the server was shut down, not that it crashed), the coordinator hears nothing while the server goes down, and after the start it is told a restart interrupted the member, not the user. The board writes a turn's outcome as soon as the run finishes, and the resolution judge's verdict replaces it when it differs.
  • Emoji reactions and experimental features, from #6609 and #6759, both merged into main now: ReactionEvent, react/unreact tools, POST /api/conversations/{id}/reactions, and the conversation_reactions experimental feature under /api/settings/experimental. With the feature on, every conversation the server starts gets the react tools, project workers included.
  • Board groups (project_board.py): when a project member's run finishes, the server tags it turn=asked|answered (did the agent's last message ask the user something) and, when the critic scored the task at least 0.9, resolved=auto. The user resolves a conversation with resolved=user through the conversation tag PATCH; any new user message clears both, in a project or not. GET /conversations reports resolved, turn and automation_id, and attention is needs_you, needs_input, working, finished, idle or resolved (resolved never hides a conversation that is running or blocked on the user). Once the agent stops on a question, turn stays asked until the user answers, even when a Stop hook makes it go on to a finish that does not repeat the question. Children do not copy resolved/turn. The messages a /goal loop sends its own conversation now carry sender: "goal_loop", so the board does not take them for the user's words, and messages the server delivers to a coordinator leave its /goal loop running. POST /api/projects/{id}/conversations/{conversation_id}/resolve and /reopen (204) set and clear resolved=user, and both clear resolvenote. The server owns project, turn, ask, resolved and resolvenote: a client's PATCH /api/conversations/{id} merges its tags under the server's tag lock and keeps those as they are, so a full set read a moment earlier cannot undo a board update or change membership. Those routes are the only way to resolve or reopen.
  • Project artifacts (project_artifacts.py): files a project produces or is given that do not belong in a repository. Routes under /api/projects/{id}/artifacts upload (multipart), list, download and delete them, and POST .../artifacts/from-conversation copies a file from a member's workspace; that file is opened without following a final symlink or blocking on a FIFO and must stay inside the working directory. The write_artifact tool lets an agent save its deliverables, and list_artifacts and read_artifact let it see the project's files and read one: small UTF-8 text comes back inline, anything else is copied into its workspace (.agent_tmp/artifacts/<name>, or a path it names inside the workspace). The tools only work in a project. A download is sent like a workspace file or trajectory download, application/octet-stream as an attachment, plus X-Content-Type-Options: nosniff, so an uploaded HTML or SVG file is saved, never rendered from the server's origin. Deleting a project deletes its artifacts. A project's artifacts and memory share one lock, a thread lock plus a file lock like projects.json's, so server processes sharing a persistence folder take turns.
  • Asks: when a conversation stops on a question, the server also saves that question, in one line of at most 200 characters, as the ask tag, so the board can show what a conversation needs from the user, not only that it needs something. GET /conversations reports ask while turn is asked; a new user message clears it, and children do not copy it.
  • Project memory (project_memory.py): MEMORY.md (an index of one-line rules and links) and lowercase .md topic files under <persistence dir>/projects/<id>/memory/, beside the project's artifacts/. GET /api/projects/{id}/memory lists them; GET/PUT/DELETE .../memory/{name} read, write and delete one. Every write names the revision it was based on, and a stale one answers 409 with the current file to merge, so two conversations cannot silently lose each other's change. The read_project_memory and update_project_memory tools let agents keep it, and a conversation a project member starts gets MEMORY.md in its first message, marked as data, not instructions.
  • Typed TypeScript client: ProjectsClient in clients/typescript (exported from @openhands/typescript-client/clients) covers every /api/projects route: projects, membership, the transcript route, artifacts (upload, copy from a conversation, download, read as text, delete) and memory with revisions, with the wire types (Project, ProjectConversation, ProjectArtifact, ProjectMemoryFile, …) and projectIdKey. The coordinator's start fields reuse the generated StartConversationRequest types, so the public weak-type budget is unchanged. The routes are client-ahead entries in endpoint-audit.config.json until a release carries them. Canvas can then drop its hand-written Projects service.
  • Pull request status (GET /api/git/repositories/{owner}/{repo}/pull-requests/{number}?provider=github), the HTTPException headers fix, and reading session-socket messages while a step runs are now on main (feat(agent-server): read a pull request's state and CI result #6704, fix(agent-server): keep HTTPException headers in error responses #6706, fix(agent-server): read session-socket messages while a step runs #6709), so this PR no longer carries them. Projects UIs use the status route for each conversation's PR card.
  • Fresh tags on read: GET /api/conversations/{id} now reads tags from meta.json, which a tag PATCH writes first, instead of base_state.json, which a running conversation saves only when its run releases the state lock. Before, resolving a running conversation stayed invisible until its run ended.

REST API contract changes

Compared with base OpenAPI ee062ba88704 for public /api/** paths.

--- base public OpenAPI
+++ head public OpenAPI
@@ -724,0 +725 @@
+response POST /api/conversations/{conversation_id}/children 409 no-content
@@ -1063 +1064 @@
-schema Action oneOf=[MCPToolAction,ClassifyAndSwitchLLMAction,FinishAction,InvokeSkillAction,SwitchLLMAction,ThinkAction,VisionInspectAction,AskOracleAction,AskUserAction,BrowserAction,BrowserClickAction,BrowserCloseTabAction,BrowserGetContentAction,BrowserGetStateAction,BrowserGetStorageAction,BrowserGoBackAction,BrowserListTabsAction,BrowserNavigateAction,BrowserScrollAction,BrowserSetStorageAction,BrowserStartRecordingAction,BrowserStopRecordingAction,BrowserSwitchTabAction,BrowserTypeAction,StartChildConversationAction,FindConversationsAction,ReadConversationAction,DelegateAction,FileEditorAction,EditAction,ListDirectoryAction,ReadFileAction,WriteFileAction,GlobAction,GrepAction,EditMessageAction,PlanningFileEditorAction,ReactAction,UnreactAction,TaskAction,TaskTrackerAction,TerminalAction,WorkflowAction]
+schema Action oneOf=[MCPToolAction,ClassifyAndSwitchLLMAction,FinishAction,InvokeSkillAction,SwitchLLMAction,ThinkAction,VisionInspectAction,AskOracleAction,AskUserAction,AskUserForConversationAction,BrowserAction,BrowserClickAction,BrowserCloseTabAction,BrowserGetContentAction,BrowserGetStateAction,BrowserGetStorageAction,BrowserGoBackAction,BrowserListTabsAction,BrowserNavigateAction,BrowserScrollAction,BrowserSetStorageAction,BrowserStartRecordingAction,BrowserStopRecordingAction,BrowserSwitchTabAction,BrowserTypeAction,StartChildConversationAction,FindConversationsAction,ReadConversationAction,DelegateAction,FileEditorAction,EditAction,ListDirectoryAction,ReadFileAction,WriteFileAction,GlobAction,GrepAction,EditMessageAction,PlanningFileEditorAction,ListArtifactsAction,ListWorkersAction,MessageWorkerAction,ReadArtifactAction,ReadProjectMemoryAction,ReadWorkerAction,ResolveConversationAction,UpdateProjectMemoryAction,WriteArtifactAction,ReactAction,UnreactAction,TaskAction,TaskTrackerAction,TerminalAction,WorkflowAction]
@@ -1145,0 +1147 @@
+schema AgentContext-Input property project_context optional schema=anyOf=[type="string",type="null"]
@@ -1158,0 +1161 @@
+schema AgentContext-Output property project_context optional schema=anyOf=[type="string",type="null"]
@@ -1291,0 +1295,6 @@
+schema AskUserForConversationAction property for_conversation optional schema=anyOf=[type="string",type="null"]
+schema AskUserForConversationAction property kind required schema=type="string" const="AskUserForConversationAction"
+schema AskUserForConversationAction property options required schema=type="array" items=AskUserOption minItems=2 maxItems=4
+schema AskUserForConversationAction property question required schema=type="string" minLength=1 maxLength=300
+schema AskUserForConversationAction property recommended optional schema=anyOf=[type="integer" minimum=0.0,type="null"]
+schema AskUserForConversationAction type="object" additionalProperties=false
@@ -2357,0 +2367,15 @@
+schema ListArtifactsAction property kind required schema=type="string" const="ListArtifactsAction"
+schema ListArtifactsAction type="object" additionalProperties=false
+schema ListArtifactsObservation property content optional schema=type="array" items=anyOf=[TextContent,ImageContent]
+schema ListArtifactsObservation property is_error optional schema=type="boolean" default=false
+schema ListArtifactsObservation property kind required schema=type="string" const="ListArtifactsObservation"
+schema ListArtifactsObservation property project_id optional schema=anyOf=[type="string",type="null"]
+schema ListArtifactsObservation type="object" additionalProperties=false
+schema ListArtifactsTool property action_type required schema=type="string"
+schema ListArtifactsTool property annotations optional schema=anyOf=[openhands__sdk__tool__tool__ToolAnnotations,type="null"]
+schema ListArtifactsTool property description required schema=type="string"
+schema ListArtifactsTool property kind required schema=type="string" const="ListArtifactsTool"
+schema ListArtifactsTool property meta optional schema=anyOf=[type="object" additionalProperties=true,type="null"]
+schema ListArtifactsTool property observation_type optional schema=anyOf=[type="string",type="null"]
+schema ListArtifactsTool property title required schema=type="string"
+schema ListArtifactsTool type="object"
@@ -2377,0 +2402,16 @@
+schema ListWorkersAction property kind required schema=type="string" const="ListWorkersAction"
+schema ListWorkersAction type="object" additionalProperties=false
+schema ListWorkersObservation property content optional schema=type="array" items=anyOf=[TextContent,ImageContent]
+schema ListWorkersObservation property conversations optional schema=type="array" items=ProjectConversationSummary
+schema ListWorkersObservation property is_error optional schema=type="boolean" default=false
+schema ListWorkersObservation property kind required schema=type="string" const="ListWorkersObservation"
+schema ListWorkersObservation property project_id optional schema=anyOf=[type="string",type="null"]
+schema ListWorkersObservation type="object" additionalProperties=false
+schema ListWorkersTool property action_type required schema=type="string"
+schema ListWorkersTool property annotations optional schema=anyOf=[openhands__sdk__tool__tool__ToolAnnotations,type="null"]
+schema ListWorkersTool property description required schema=type="string"
+schema ListWorkersTool property kind required schema=type="string" const="ListWorkersTool"
+schema ListWorkersTool property meta optional schema=anyOf=[type="object" additionalProperties=true,type="null"]
+schema ListWorkersTool property observation_type optional schema=anyOf=[type="string",type="null"]
+schema ListWorkersTool property title required schema=type="string"
+schema ListWorkersTool type="object"
@@ -2643,0 +2684,17 @@
+schema MessageWorkerAction property conversation_id required schema=type="string"
+schema MessageWorkerAction property kind required schema=type="string" const="MessageWorkerAction"
+schema MessageWorkerAction property message required schema=type="string" minLength=1
+schema MessageWorkerAction type="object" additionalProperties=false
+schema MessageWorkerObservation property content optional schema=type="array" items=anyOf=[TextContent,ImageContent]
+schema MessageWorkerObservation property conversation_id optional schema=anyOf=[type="string",type="null"]
+schema MessageWorkerObservation property is_error optional schema=type="boolean" default=false
+schema MessageWorkerObservation property kind required schema=type="string" const="MessageWorkerObservation"
+schema MessageWorkerObservation type="object" additionalProperties=false
+schema MessageWorkerTool property action_type required schema=type="string"
+schema MessageWorkerTool property annotations optional schema=anyOf=[openhands__sdk__tool__tool__ToolAnnotations,type="null"]
+schema MessageWorkerTool property description required schema=type="string"
+schema MessageWorkerTool property kind required schema=type="string" const="MessageWorkerTool"
+schema MessageWorkerTool property meta optional schema=anyOf=[type="object" additionalProperties=true,type="null"]
+schema MessageWorkerTool property observation_type optional schema=anyOf=[type="string",type="null"]
+schema MessageWorkerTool property title required schema=type="string"
+schema MessageWorkerTool type="object"
@@ -2699 +2756 @@
-schema Observation oneOf=[MCPToolObservation,ClassifyAndSwitchLLMObservation,FinishObservation,InvokeSkillObservation,SwitchLLMObservation,ThinkObservation,VisionInspectObservation,ClientToolObservation,AskOracleObservation,AskUserObservation,BrowserObservation,StartChildConversationObservation,ConversationMentionsObservation,DelegateObservation,FileEditorObservation,EditObservation,ListDirectoryObservation,ReadFileObservation,WriteFileObservation,GlobObservation,GrepObservation,EditMessageObservation,PlanningFileEditorObservation,ReactionObservation,TaskObservation,TaskTrackerObservation,TerminalObservation,WorkflowObservation]
+schema Observation oneOf=[MCPToolObservation,ClassifyAndSwitchLLMObservation,FinishObservation,InvokeSkillObservation,SwitchLLMObservation,ThinkObservation,VisionInspectObservation,ClientToolObservation,AskOracleObservation,AskUserObservation,BrowserObservation,StartChildConversationObservation,ConversationMentionsObservation,DelegateObservation,FileEditorObservation,EditObservation,ListDirectoryObservation,ReadFileObservation,WriteFileObservation,GlobObservation,GrepObservation,EditMessageObservation,PlanningFileEditorObservation,ListArtifactsObservation,ListWorkersObservation,MessageWorkerObservation,ReadArtifactObservation,ReadProjectMemoryObservation,ReadWorkerObservation,ResolveConversationObservation,UpdateProjectMemoryObservation,WriteArtifactObservation,ReactionObservation,TaskObservation,TaskTrackerObservation,TerminalObservation,WorkflowObservation]
@@ -2849,0 +2907,11 @@
+schema ProjectConversationSummary property ask optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property attention required schema=type="string"
+schema ProjectConversationSummary property automation_id optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property conversation_id required schema=type="string"
+schema ProjectConversationSummary property execution_status required schema=type="string"
+schema ProjectConversationSummary property parent_conversation_id optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property resolved optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property title optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property turn optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary property updated_at optional schema=anyOf=[type="string",type="null"]
+schema ProjectConversationSummary type="object"
@@ -2903,0 +2972,19 @@
+schema ReadArtifactAction property kind required schema=type="string" const="ReadArtifactAction"
+schema ReadArtifactAction property name required schema=type="string"
+schema ReadArtifactAction property path optional schema=anyOf=[type="string",type="null"]
+schema ReadArtifactAction type="object" additionalProperties=false
+schema ReadArtifactObservation property content optional schema=type="array" items=anyOf=[TextContent,ImageContent]
+schema ReadArtifactObservation property is_error optional schema=type="boolean" default=false
+schema ReadArtifactObservation property kind required schema=type="string" const="ReadArtifactObservation"
+schema ReadArtifactObservation property name optional schema=anyOf=[type="string",type="null"]
+schema ReadArtifactObservation property path optional schema=anyOf=[type="string",type="null"]
+schema ReadArtifactObservation property project_id optional schema=anyOf=[type="string",type="null"]
+schema ReadArtifactObservation type="object" additionalProperties=false
+schema ReadArtifactTool property action_type required schema=type="string"
+schema ReadArtifactTool property annotations optional schema=anyOf=[openhands__sdk__tool__tool__ToolAnnotations,type="null"]
+schema ReadArtifactTool property description required schema=type="string"
+schema ReadArtifactTool property kind required schema=type="string" const="ReadArtifactTool"
+schema ReadArtifactTool property meta optional schema=anyOf=[type="object" additionalProperties=true,type="null"]
+schema ReadArtifactTool property observation_type optional schema=anyOf=[type="string",type="null"]
+schema ReadArtifactTool property title required schema=type="string"
+schema ReadArtifactTool type="object"
@@ -2937,0 +3025,34 @@
+schema ReadProjectMemoryAction property kind required schema=type="string" const="ReadProjectMemoryAction"
+schema ReadProjectMemoryAction property name optional schema=type="string" default="MEMORY.md"
+schema ReadProjectMemoryAction type="object" additionalProperties=false
+schema ReadProjectMemoryObservation property content optional schema=type="array" items=anyOf=[TextContent,ImageContent]
+schema ReadProjectMemoryObservation property is_error optional schema=type="boolean" default=false
+schema ReadProjectMemoryObservation property kind required schema=type="string" const="ReadProjectMemoryObservation"
+schema ReadProjectMemoryObservation property name optional schema=anyOf=[type="string",type="null"]
+schema ReadProjectMemoryObservation property revision optional schema=anyOf=[type="string",type="null"]
+schema ReadProjectMemoryObservation type="object" additionalProperties=false
+schema ReadProjectMemoryTool property action_type required schema=type="string"
+schema ReadProjectMemoryTool property annotations optional schema=anyOf=[openhands__sdk__tool__tool__ToolAnnotations,type="null"]
+schema ReadProjectMemoryTool property description required schema=type="string"
+schema ReadProjectMemoryTool property kind required schema=type="string" const="ReadProjectMemoryTool"
+schema ReadProjectMemoryTool property meta optional schema=anyOf=[type="object" additionalProperties=true,type="null"]
+schema ReadProjectMemoryTool property observation_type optional schema=anyOf=[type="string",type="null"]
... diff truncated; 77 more line(s)

Issue Number

Fixes #6637. Follows the Insider Cat drafts above and OpenHands#17557.

How to Test

Unit and route tests (real ConversationService, no mocks for tag persistence):

uv run pytest tests/agent_server/test_projects_router.py tests/agent_server/test_project_board.py tests/agent_server/test_parent_child_conversations.py tests/tools/project -q

They pass, including the race the send guard handles (a conversation already waiting when the guarded send runs keeps the message, is not run, and stays waiting_for_confirmation), a subscriber registered by the real ConversationService, Stop hook bounces, a FIFO in the workspace, the sender and its header line (and the 422 for a sender outside the target's project), the coordinator update (run only for turns it started), the copies of the user's messages and replies (in order, without a run, ignored by the coordinator's board), /goal nudges and deliveries that leave a coordinator's /goal loop running, a worker started from its profile with only the project tag and the worker tools, hostile titles in mentions, the transcript route, asks (including list questions and a Stop hook bounce), memory revisions and conflicts, the worker note (once, even for a worker's own children), and tags read before the state is saved (that test fails without the fix). Across tests/agent_server, the one failure is test_exited_leader_descendant_does_not_block_recovery, which also fails on main in this container (it needs a reaping PID 1, per the verify-agent-server skill).

TypeScript client (in clients/typescript): npm test (416 passed, including the ProjectsClient test of every route's method, path, query and body), npm run lint, npm run format:check, npm run check:public-type-budget (unchanged, 103 sites), npm run check:agent-server-api and npm run audit:endpoints pass. Canvas #19495 with this client packed locally and its Projects service moved onto ProjectsClient passes npm run typecheck and its project tests.

Live, against a real server and a real model (DeepSeek), with the new feature-map family:

export PATH="$PWD/.agents/skills/verify-agent-server/scripts:$PATH"
control-agent-server map run --file F36 --fresh
control-agent-server map run --file F37 --fresh

All 34 F36 recipes pass on 255c665, including F36.mention, where the coordinator answered with a mention of the conversation it started. All 7 F37 (reactions) recipes pass on d24b2ad, after #6609 merged into main. Among them, a coordinator with start_child_conversation, list_project_conversations and message_project_conversation starts a child that joins the project, lists it, and sends it a follow-up that makes it run again; the child's system prompt names the coordinator, and both its messages carry the coordinator as sender. When the child finishes, the coordinator gets its final message and reads its transcript with read_project_conversation. When the user talks to a member directly, the coordinator gets a copy of each message and reply and stays idle, and the member gets no coordinator messages. Asked to tell a conversation that is waiting for approval to "go ahead", the tool refuses, and the conversation keeps waiting with no new message. A conversation whose agent asks a question lists as needs_input and is reopened by the answer; an agent with write_artifact saves its report and is refused a path outside its workspace. A question saves its ask and the answer clears it; an agent adds a line to MEMORY.md with the revision it read, a blind overwrite is refused, and the child it starts sees the index in its first message.

In a live Insider run with dispatch-only Coordinator and Worker profiles, the Cat started two worker conversations from the Worker profile, both finished and merged their PRs, and a release-notes file was saved to the project. In a full OpenHands stack (verify-openhands, DeepSeek) built from an earlier revision of this branch, a message sent from the Canvas conversation chat cleared resolved=user and, in a second run, resolved=auto, keeping the other tags, and /api/projects answered.

Video/Screenshots

See Live verification evidence above for inline Paper screenshots and pinned links to the full evidence ledger, browser observations, before/after failures and corrected F36/F37 runs.

Type

  • Bug fix
  • Feature
  • Refactor
  • Breaking change
  • Docs / chore

Notes

  • Projects are local to the server that holds them: the record, its files and its memory live in that server's persistence folder, and a conversation in a Docker runtime's server is not a member. Project scoping in the tools is a guard rail for well-behaved tools, not a security boundary, because every caller holds the same session key.
  • Additive only: new routes, two optional fields on the events route's body (ConversationMessageRequest) (the send route answers {"success": true} as before unless run is set, when it adds run), new optional keywords on EventService.run()/send_message() (send_message now also returns its run outcome; run=False still returns None), a new only_if_absent keyword on set_conversation_tag, a new keep_goal_loop keyword on EventService.send_message(), a new optional worker_agent_profile_id on projects, and new SDK constants. Existing routes are unchanged, except that conversation reads now report the tags last written, a child's first message records its parent as sender, /goal messages record goal_loop as sender, and a project member's child starts from an agent profile instead of a copy of its parent. Older SDK clients cannot parse the new ReactionEvent, as with any new event type; it belongs in the release notes.
  • Naming: members are ordinary OpenHands conversations, so the API, tools and docs say "conversation" throughout (an earlier revision said "thread").
  • Deleting a project never touches conversations; their project tag stays. It deletes the project's folder, projects/<id>/ (artifacts and memory).
  • turn and ask are heuristics on the last paragraph of the agent's final message. automation_id comes from the automationid tag OpenHands automations already set; an automation's conversations join a project only when it starts them with the project's tag.
  • Not in this PR: task progress, listing or reading the user's uploaded project files from tools, and the Canvas board.
  • Five fresh-context reviews ran. The first found that the message tool checked a conversation's status on the client and then ran it, so a conversation reaching a confirmation in between would have been approved; the check now happens on the server under the run lock. It also found that a timed-out send could lead the agent to resend a delivered message; the tool now says not to. The second traced the guard through the ACP supersede branch, the re-arm after a run's tail, and goal loops, and found it holds; its reporting and PATCH-rollback findings are fixed. The third, on the board and artifacts, found nothing blocking; its artifact-safety findings (a FIFO swapped in, names with a trailing newline, case-only clashes, upload size) are fixed, and a running conversation is never listed as resolved. A fourth, on the sender and reactions, found a grandchild told the wrong parent, reaction targets taken from all events instead of the active branch, titles that could break the header line, and tool paths that dropped reactions; all are fixed. An outside review found a race where a reopen could land before a pending board write; it is fixed and tested. The fifth, after merging main, found no blockers. It found that the projects store could be set up with the default folder when a relay ran first, and that deliveries to a coordinator stopped its /goal loop while a member's /goal nudges were taken for the user's words; both are fixed and tested, and the sender header now promises a reply only to a coordinator. Still open and minor: a run-less copy turns a finished coordinator idle on the board, and two members messaging each other at once can wait out the 300 s tool timeout. The bot review approved the PR before the worker, coordinator-prompt and mentions changes. A later outside review found three more: a run-less copy turned a stuck coordinator idle, so the next update resumed it (fixed and tested); two project creations reserving one coordinator id at once both succeeded (the second is now a 422, tested); and restart reconciliation reads only the 50 newest coordinator messages, so an old update can be delivered again (#6817, out of scope here).

🤖 Generated with Claude Code

https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E


🐳 Agent Server images for this PR — GHCR package, pull/run commands, and all pushed tags (click to expand)

• GHCR package: https://github.com/OpenHands/agent-sdk/pkgs/container/agent-server

Variants & Base Images

Variant Architectures Base Image Docs / Tags
java amd64, arm64 eclipse-temurin:17-jdk Link
python-slim amd64, arm64 python-node-runtime Link
python-minimal amd64, arm64 python-node-runtime Link
python amd64, arm64 python-node-runtime Link
golang amd64, arm64 golang:1.21-bookworm Link

Pull (multi-arch manifest)

# Each variant is a multi-arch manifest supporting both amd64 and arm64
docker pull ghcr.io/openhands/agent-server:ecbb860-python

Run

docker run -it --rm \
  -p 8000:8000 \
  --name agent-server-ecbb860-python \
  ghcr.io/openhands/agent-server:ecbb860-python

All tags pushed for this build

ghcr.io/openhands/agent-server:ecbb860-golang-amd64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-golang-amd64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-golang-amd64
ghcr.io/openhands/agent-server:ecbb860-golang_tag_1.21-bookworm-amd64
ghcr.io/openhands/agent-server:ecbb860-golang-arm64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-golang-arm64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-golang-arm64
ghcr.io/openhands/agent-server:ecbb860-golang_tag_1.21-bookworm-arm64
ghcr.io/openhands/agent-server:ecbb860-java-amd64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-java-amd64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-java-amd64
ghcr.io/openhands/agent-server:ecbb860-eclipse-temurin_tag_17-jdk-amd64
ghcr.io/openhands/agent-server:ecbb860-java-arm64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-java-arm64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-java-arm64
ghcr.io/openhands/agent-server:ecbb860-eclipse-temurin_tag_17-jdk-arm64
ghcr.io/openhands/agent-server:ecbb860-python-amd64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-amd64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-amd64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-amd64
ghcr.io/openhands/agent-server:ecbb860-python-arm64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-arm64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-arm64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-arm64
ghcr.io/openhands/agent-server:ecbb860-python-minimal-amd64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-minimal-amd64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-minimal-amd64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-minimal-amd64
ghcr.io/openhands/agent-server:ecbb860-python-minimal-arm64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-minimal-arm64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-minimal-arm64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-minimal-arm64
ghcr.io/openhands/agent-server:ecbb860-python-slim-amd64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-slim-amd64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-slim-amd64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-slim-amd64
ghcr.io/openhands/agent-server:ecbb860-python-slim-arm64
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-slim-arm64
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-slim-arm64
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-slim-arm64
ghcr.io/openhands/agent-server:ecbb860-golang
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-golang
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-golang
ghcr.io/openhands/agent-server:ecbb860-golang_tag_1.21-bookworm
ghcr.io/openhands/agent-server:ecbb860-java
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-java
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-java
ghcr.io/openhands/agent-server:ecbb860-eclipse-temurin_tag_17-jdk
ghcr.io/openhands/agent-server:ecbb860-python-minimal
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-minimal
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-minimal
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-minimal
ghcr.io/openhands/agent-server:ecbb860-python-slim
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python-slim
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python-slim
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime-slim
ghcr.io/openhands/agent-server:ecbb860-python
ghcr.io/openhands/agent-server:ecbb860c392df103252a432f8878b1834285492a-python
ghcr.io/openhands/agent-server:claude-project-thread-xwkk9b-python
ghcr.io/openhands/agent-server:ecbb860-python-node-runtime

About Multi-Architecture Support

  • Each variant tag (e.g., ecbb860-python) is a multi-arch manifest supporting both amd64 and arm64
  • Docker automatically pulls the correct architecture for your platform
  • Individual architecture tags (e.g., ecbb860-python-amd64) are also available if needed

Generated by Claude Code

A project groups the conversations that work toward one goal: a
coordinator conversation the person mainly talks to, and its threads.

- Projects API under /api/projects: create, list, get, patch, delete,
  plus /threads to list, add and remove members. Records live in
  projects.json with the same atomic-write and file-lock store as
  workspaces.json; a corrupted or newer-schema file answers 409 and is
  never overwritten.
- Membership is the `project` conversation tag, so children started with
  start_child_conversation join their parent's project. The threads
  endpoint groups members by attention (needs_you, working, finished,
  idle) derived from execution_status.
- ConversationService gains list_tagged_conversations and
  set_conversation_tag (single-tag change that keeps the other tags).
- openhands.tools.project: list_project_threads and
  message_project_thread, server-registered tools that let a coordinator
  follow and steer its threads. The message tool refuses a thread that is
  waiting for confirmation (running it would approve the pending action)
  and sends to a running thread without requesting a new run.
- Feature map family F36 with 19 recipes, all passing on a fresh server
  with a real model.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
Review found that message_project_thread checked a thread's status and
then ran it in a separate request, so a thread that reached a
confirmation in between would have its pending action approved by the
run.

- EventService.run()/send_message() take refuse_pending_confirmation;
  the check is made under the run lock, where no other run can change
  the status. A refused send keeps the message and raises
  ValueError("waiting_for_confirmation").
- New POST /api/projects/{id}/threads/{conversation_id}/messages uses
  it: 409 for a thread already waiting (nothing saved), otherwise
  run_requested tells whether a run was started.
- The tool delivers through that route, waits up to 300 s (a running
  thread holds its state lock for a whole tool call) and, on a
  timeout, says the message may still arrive and must not be resent.
- Project writes re-check the coordinator under the file lock, tag the
  coordinator only after the record is saved (rolling back if it
  vanished), and run file I/O off the event loop; tag updates lock per
  conversation.
- F36 gains message-rest and message-refused recipes; 21 recipes pass
  on a fresh server with a real model.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
Second review confirmed the confirmation guard holds on every path, and
found that the route could still say a run was requested when the
message was parked behind a confirmation or a stop.

- EventService.send_message(run=True) now returns "started", "queued"
  (a run was already active) or "not_run"; run=False returns None, so
  existing callers are unaffected.
- The messages route answers run: started | queued |
  waiting_for_confirmation | not_run instead of run_requested, and the
  tool words each case without claiming more than happened.
- A failed coordinator tag on PATCH undoes the whole change, and only
  if no later write touched the project; it never restores a coordinator
  that now leads another project. Exceptions while tagging roll back too.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst enyst self-assigned this Nov 7, 2026
@github-actions

github-actions Bot commented Nov 7, 2026 •

Copy link
Copy Markdown
Contributor

REST API breakage checks (OpenAPI) — ✅ PASSED

Result: ✅ PASSED

Action log

@datadog-official

This comment has been minimized.

@github-actions

github-actions Bot commented Nov 7, 2026 •

Copy link
Copy Markdown
Contributor

Coverage

Coverage Report •
FileStmtsMissCoverMissing
openhands-agent-server/openhands/agent_server
   api.py4603493%147, 201–202, 211–212, 214, 217–218, 225–226, 240–241, 258, 264, 267–268, 365, 379, 385, 454, 459, 472–474, 507, 536, 542, 546, 567–568, 863, 866, 868, 874
   conversation_router.py3712294%134–135, 138–139, 364, 497–498, 545, 631, 712, 728, 785, 789–790, 890, 960–962, 1007, 1047–1048, 1079
   conversation_service.py133410592%218–219, 301, 322, 506–507, 654, 692, 699–700, 800, 851–852, 856, 861, 911, 963–964, 971, 1004–1005, 1021, 1053, 1057, 1070, 1090, 1111–1112, 1121, 1200, 1225, 1231–1232, 1236–1237, 1245, 1272, 1278, 1372, 1378, 1383, 1389, 1397–1398, 1407–1410, 1419, 1432, 1450, 1487, 1491–1495, 1518, 1567, 1596, 1607, 1893, 1922–1925, 1979, 2110, 2130, 2133, 2140–2142, 2145–2146, 2152, 2162, 2166, 2195, 2204, 2266–2268, 2303, 2442, 2445, 2474, 2484, 2489–2492, 2506, 2517, 2529–2530, 2563, 2664, 2733, 2791, 2806–2807, 3211, 3264, 3267
   dependencies.py93397%104, 156, 203
   errors.py32197%109
   event_router.py183597%298, 399, 410, 472–473
   event_service.py118912290%204, 210–211, 351–352, 476, 480, 485, 512, 519, 554, 557–558, 562–563, 574, 580, 651–652, 658, 714–715, 862, 881, 910, 920, 943–944, 948, 1012, 1047–1049, 1052, 1074–1075, 1086–1092, 1103, 1110, 1116, 1329, 1362, 1365, 1448, 1466, 1480, 1655–1656, 1712, 1746, 1900–1902, 1951, 1955, 1962, 1976, 1992, 2000, 2024, 2071–2072, 2163, 2185, 2188–2189, 2216, 2223, 2239, 2241, 2248, 2256, 2258, 2275, 2300, 2321, 2372, 2383–2386, 2395, 2400, 2402, 2407, 2416–2417, 2423, 2435, 2445, 2483–2485, 2570–2571, 2573–2580, 2582–2583, 2593, 2603, 2619, 2630, 2654, 2660, 2671, 2680–2681, 2703, 2713
   experimental_features.py67297%170, 176
   project_artifacts.py150696%113, 120, 136, 246–247, 255
   project_board.py198597%139–140, 285, 319, 326
   project_memory.py89397%92, 172–173
   project_profiles.py51394%100–102
   project_prompts.py60297%175, 178
   project_service.py3064187%147–148, 152–156, 211–213, 225–226, 289–290, 292, 313, 340–342, 390, 398, 421, 424, 455, 488, 519–520, 556, 602–605, 682–683, 704–705, 710, 775–778
   project_turns.py196299%61, 249
   project_wake.py3085183%92, 101–103, 105, 124, 144, 154, 188–194, 199, 221–222, 227–228, 315, 318, 356, 405, 411–413, 415, 445, 447, 482, 487, 495–496, 571–575, 589, 600, 608, 612, 626, 643, 666, 671, 685, 688, 701–702
   projects_router.py4222095%223, 360–362, 387–389, 400, 421, 537–539, 596, 599, 612, 759, 851, 856, 1002, 1084
   resolution_judge.py20290%50–51
openhands-agent-server/openhands/agent_server/persistence
   models.py2572989%58, 429, 434, 472, 521, 550–556, 558, 560, 563, 567, 593, 610–611, 656, 660, 662, 711, 714, 744–747, 750
   store.py5146987%52–53, 103, 107, 111, 115, 119, 125, 148–149, 171, 174–177, 179–180, 185, 222, 224–228, 230–234, 239–244, 267, 270–271, 389–390, 395, 397–398, 547, 580–581, 586, 588–589, 597, 674–675, 754, 756, 775, 779, 797–798, 804, 872–873, 877–879, 898, 947–948, 1141–1142
openhands-sdk/openhands/sdk/agent
   base.py3783292%103–104, 257, 337, 343, 554, 640, 650, 701, 703–704, 808–809, 819–820, 851, 855, 858, 933, 936, 948–950, 955, 957, 960, 967, 971, 980, 994, 997, 1013
openhands-sdk/openhands/sdk/context
   agent_context.py175498%467–468, 538, 561
openhands-sdk/openhands/sdk/context/prompts
   presets.py48296%211–212
openhands-sdk/openhands/sdk/context/prompts/sections
   project.py84199%281
openhands-sdk/openhands/sdk/conversation
   types.py111595%178, 197, 199, 201, 206
openhands-sdk/openhands/sdk/event
   ask_user.py832076%74–76, 80–82, 118, 123, 130, 146–148, 176, 182, 196–198, 201–203
openhands-sdk/openhands/sdk/launch
   finalize.py1421788%114, 120, 159, 248, 254–263, 275, 278, 288
openhands-tools/openhands/tools/ask_user
   definition.py49982%62, 70–74, 94, 139, 172
openhands-tools/openhands/tools/project
   definition.py26911159%52, 61–72, 117–118, 121, 123, 161–165, 173–182, 221–222, 225, 227, 265–270, 280–289, 321–322, 325, 327, 362–367, 379–388, 411–412, 415, 417, 441, 449–454, 475–476, 479, 481, 514, 522, 547–548, 551, 553, 593, 602, 628–629, 632, 634, 676, 685, 712–713, 716, 718, 742, 748, 777–778, 781, 783
   impl.py28822721%93–96, 100, 103–106, 109–110, 113, 119–125, 128–129, 132–134, 139–143, 147–148, 153–154, 167, 174–177, 179–180, 192, 197–198, 213, 220–224, 227–228, 232–233, 238–242, 245–248, 250, 254, 258–265, 267–268, 281–283, 306, 313–317, 320–321, 324–325, 329–331, 334, 344–346, 350, 357–359, 361, 386, 393–395, 399–401, 406–414, 428–431, 434, 438–442, 444, 448–453, 466, 473–476, 481, 484–486, 491–492, 495–501, 521, 528–530, 533–541, 543–544, 559, 566–569, 574–575, 578, 587–588, 590–591, 598–599, 602–605, 620, 627–631, 635–637, 642, 649–658, 680, 687–690, 698–700, 704, 707–713, 716, 726–731, 735
TOTAL60412980584% 

A finished run tags its conversation with how the agent's last message
ended (turn=asked|answered) and, when the critic scored it at least 0.9,
resolved=auto. A new user message removes both. The project threads list
uses them for needs_input and resolved groups and reports automation_id.
Children no longer copy their parent's resolved/turn tags.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
Projects keep artifacts, the files they produce or are given that do not
belong in a source repository, under the persistence dir. Routes list,
upload, download and delete them, and copy a file from a member
conversation's workspace (paths, symlinks included, must stay inside it).
The save_project_artifact tool lets an agent save its deliverables there.
Deleting a project deletes its artifacts.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
Open workspace files without following a final symlink or blocking on a
FIFO, in a worker thread, and check the opened file again. Match artifact
names in full (no trailing newline), refuse trailing dots or spaces,
Windows device names and names differing only in case, and serialize
saves and deletes per project. Refuse an oversized upload before copying
it. List a running thread as working even if it is still tagged resolved,
and skip tag writes a user message does not need.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
A Stop hook that denies stopping on a plain message makes the agent go
on to a finish that may not repeat its question. Once the agent stops on
a question, keep turn=asked until the next user message.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
The Resolve button can mark conversations outside a project too, so a
new user message clears resolved/turn wherever they are set. Recording
outcomes stays limited to project members.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst enyst changed the title feat(agent-server): projects API and coordinator tools feat(agent-server): projects API, coordinator tools, board groups and artifacts Nov 7, 2026
GET /api/conversations/{id} took tags from base_state.json, which a
running conversation saves only when its run releases the state lock, so
a tag PATCH (for example resolving a running thread) stayed invisible
until the run ended. Read tags from meta.json, which the PATCH writes
first and the catalog already uses.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
…-xwkk9b

# Conflicts:
#	.agents/skills/verify-agent-server/references/feature-map/README.md
@enyst

enyst commented Nov 7, 2026

Copy link
Copy Markdown
Member Author

sdk-tests is red on tests/sdk/conversation/test_local_conversation_mcp.py::test_failing_mcp_server_keeps_the_other_servers_tools[invalid-entry], and the failure isn't from this PR. The same test fails on plain main (origin/main at ff3a0b8). #5796 now rejects an MCP server entry with neither command nor url when the config is validated. That turns the test's {} case into a ValidationError before the conversation starts, while the test from #6480 still expects it to fail only at runtime. This PR doesn't touch MCP. No fix exists on main yet. One option is to have the invalid-entry case assert the new validation error instead of a dropped server, but that call belongs with #5796's owners, so I haven't changed the test here. I also merged main into the branch to resolve a feature-map count conflict.


Generated by Claude Code

OpenHands calls these conversations. Routes move to
/api/projects/{id}/conversations, the list field is `conversations`, the
models are ProjectConversation*, and the coordinator tools are
list_project_conversations and message_project_conversation.
THREAD_STATE_TAG_KEYS becomes BOARD_TAG_KEYS and ThreadOutcomeSubscriber
becomes ConversationOutcomeSubscriber.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
A message one project conversation sends another (the coordinator's
message_project_conversation, or POST .../messages with
sender_conversation_id) now records sender="conversation:<id hex>" and
opens with a line telling the receiving agent who sent it and that it
approves nothing on the person's behalf. A child started by a project
member gets a system-prompt note naming the conversation that started it,
and its task message records that parent as sender.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
A ReactionEvent (target_event_id, emoji, removed, source) records an emoji
on a message; it never reaches the LLM. Agents react with the new react and
unreact tools (openhands.tools.reaction), which default to the person's
latest message. The executor only reads the log: the observation carries
the reaction and the agent records the event right after it on its own
thread, through a new Observation.follow_up_events() hook, so it is safe
with parallel tool calls. People react through
POST /api/conversations/{id}/reactions (LocalConversation.react), which
defaults to the agent's latest message or finish and never runs the agent.
Emoji come from a fixed set of shortcodes. New feature-map family F37.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
- A user message that arrives while a finished run's outcome is still being
  written now queues a reopen behind it, so a stale turn/resolved=auto no
  longer lands after the person wrote again (reproduced by
  test_a_user_message_before_the_outcome_is_saved_still_reopens).
- A child gets guidance naming its own parent: inherited guidance is
  removed first, also when the parent has left its project.
- Titles in the sender header and guidance are flattened to one line
  without quotes or brackets.
- Reaction targets come from the active branch.
- execute_tool records an observation's follow-up events, and rerun_actions
  logs them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
A ReactionEvent (target_event_id, emoji, removed, source) records an emoji
on a message; it never reaches the LLM. Agents react with the new react and
unreact tools (openhands.tools.reaction), which default to the person's
latest message. The executor only reads the log: the observation carries
the reaction and the agent records the event right after it on its own
thread, through a new Observation.follow_up_events() hook, so it is safe
with parallel tool calls. People react through
POST /api/conversations/{id}/reactions (LocalConversation.react), which
defaults to the agent's latest message or finish and never runs the agent.
Emoji come from a fixed set of shortcodes. New feature-map family F37.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
… on every tool path

Reaction part of de324a9 from #6602.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
…s the first

A feature flag is a yes/no switch the user sets for themselves, stored in
PersistedSettings.features and read when a conversation starts.
GET /api/settings/features lists the known features with their state, and
PATCH /api/settings/features/{key} {enabled} turns one on or off; unknown
keys are 404. The registry is openhands.agent_server.features.

The first flag, conversation_reactions (off by default), gives every new
OpenHands-agent conversation the react and unreact tools through a new
finalize(feature_tools=...) parameter (ACP agents are left alone), and
lets POST /api/conversations/{id}/reactions through; while it is off that
route answers 409. Feature map F37 covers both.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJ5sQMXhwgSTG2Ztkt
ReactionEvent / ReactionEmoji types with isReactionEvent,
ConversationClient.react(), and SettingsClient.listFeatures() /
updateFeature() with the FeatureFlag type.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJ5sQMXhwgSTG2Ztkt
- Coordinator tools say to delegate by default. When a member finishes a
  turn the coordinator started, the coordinator gets the member's final
  message as one update (project_wake.notify_coordinator). Turns the person
  starts in a member do not wake it, and an answered update starts no round.
- read_project_conversation and GET /api/projects/{id}/conversations/{cid}/messages
  give the coordinator a member's latest messages on demand.
- POST /api/conversations/{id}/events takes optional sender_conversation_id
  (records the sender and a "Sent by" line; a sender in a project only reaches
  its project) and refuse_pending_confirmation (409 instead of approving).
  With run it reports the run outcome. The project-only messages route is
  gone; message_project_conversation now uses the generic route.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst

enyst commented Nov 19, 2026

Copy link
Copy Markdown
Member Author

Two CI failures on 4a4a34c (the merge of main) aren't from this PR:

  • tools-tests: tests/tools/terminal/test_pool_integration.py::TestPoolClose::test_closing_the_executor_releases_a_call_waiting_on_a_command got tmux's "no current target" recovery message instead of "terminal was closed".
  • agent-server-tests: tests/agent_server/test_conversation_service.py::test_prepare_for_sandbox_pause_blocks_new_hydration found a conversation loaded back after the pause.

This PR touches neither the terminal tools nor the sandbox-pause path or its test, and both tests pass 3/3 locally on this head. I'll rerun the failed jobs once the run finishes.


Generated by Claude Code

@enyst
enyst requested a review from all-hands-bot November 19, 2026 14:31

@all-hands-bot all-hands-bot left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This review was posted by an AI agent (OpenHands).

Re-review at head 4a4a34cc (delta since the approved f76a7784: a single merge from main; mergeStateStatus is CLEAN). Verified in this workspace:

  • The only change in the range is the merge of main, whose commits are dependency bumps (boto3 1.44.1, prettier 3.10.0 and @openrouter/sdk 1.5.91 in clients/typescript) plus the already-reviewed verify-scale tests (#6984). The merge diff touches only uv.lock, openhands-sdk/pyproject.toml and clients/typescript/package-lock.json — no Projects code, no route or SDK behavior change.
  • This matches the PR body's current contract diff, which is unchanged in shape from the previously approved head.

Test results at this exact head: Projects router + board + project tools 209 passed. CI on this exact head SHA: 54 checks success, 1 known skip, 0 failures.

No material findings on the current head.

✅ APPROVED

The client no longer reserves the coordinator's conversation id, so a
new coordinator can never collide with an existing conversation and
ConversationService needs no reuse_existing switch.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
…a profile

Drop create_conversation and the extra _create_conversation keywords:
start_conversation takes the server's tools, the first message carries
sender_conversation_id, and the title is saved after the start.

A project worker always runs an agent profile; with none to run, the
start is refused (422). start_child_conversation is back to main's
copied child for everyone else.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
A client no longer reserves the coordinator's id, so the server picks
it before the start and the <PROJECT> block says which conversation the
coordinator is: one that calls the API itself can send as itself.

F36 now activates an agent profile: workers start from the predefined
worker profile, built from the active one, and are refused without it.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
claude and others added 2 commits November 19, 2026 16:35
Remove the pending-confirmation guard from the shared send path:
refuse_pending_confirmation, the held/run outcomes on the send route,
HeldForUser, admission_refusal and the sender special case are gone, so
POST /events and EventService.send_message behave as on main.

Projects keeps two checks of its own: message_worker refuses a worker
waiting for confirmation, and project updates save without running a
coordinator that waits for one, so no agent approves another's actions.

The sender header stays in message_sender where main has it; Projects
adds project_sender_header beside it for the scope check and role.
ASK_USER moves back to main's place in experimental_features.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
… with the user

The live check no longer requires the coordinator to call message_worker:
it may decline on its own, as its prompt says only the user approves.
It still checks the worker keeps waiting and got no new message.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst
enyst requested a review from all-hands-bot November 19, 2026 17:08

@all-hands-bot all-hands-bot left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This review was posted by an AI agent (OpenHands).

Re-review at head aede9ccd (delta since the approved 4a4a34cc: five branch commits plus merges from main; mergeStateStatus is CLEAN). Verified in this workspace:

  • 4641eb341 — the shared send path returns to main's behavior. refuse_pending_confirmation, the SendMessageResponse run outcomes, admission_refusal() and HeldForUser are gone; POST /api/conversations/{id}/events answers Success again and behaves as on main. I traced where the Projects-specific protections now live and confirmed they are still enforced: the message_worker tool refuses a worker listed as waiting_for_confirmation before sending anything (impl.py status check), and every server delivery to a coordinator goes through _deliver(), which clears the run flag when _waits_for_confirmation() is true, so an update is saved without approving pending actions. The member-side answer path (pass_on_answer) applies the same check (run=not await _waits_for_confirmation(member)). The sender header lives in message_sender.py as on main, with project_sender_header layered beside it for project scoping and role.
  • 308bed138/3a45ea35c — project conversations start through start_conversation. Workers always run an agent profile (422 with none), start_child_conversation is back to main's copy for non-projects, and the coordinator's id is chosen by the server rather than client-reserved, removing the id-collision surface without losing the "one coordinator per project" behavior (concurrent duplicate coordinator starts now simply get distinct ids; the previously tested reserved-id 422 path was removed with the reservation mechanism itself).
  • e6352bdf8 names the coordinator's own id in its <PROJECT> block; aede9ccd0 regenerates the coordinator prompt snapshot and F36 recipe wording only.

Test results at this exact head: Projects router + board + cards + parent/child + project tools + run-admission + event-router suites 292 passed. CI on this exact head SHA: 54 checks success, 1 known skip, 0 failures.

No material findings on the current head.

✅ APPROVED

claude and others added 2 commits November 19, 2026 17:30
…nternally

A project never adopts an existing conversation: coordinator_conversation_id
is gone from POST and PATCH /api/projects. POST starts a new coordinator when
asked, as before.

The coordinator's and worker's prompt, project context and tools reach the
agent through finalize's project_role (a ProjectRole), set only by
ProjectService. AgentLaunchAdditions is main's again, so no client can ask
for a coordinator prompt, and start_conversation loses server_tools.

Restore needs nothing of its own: the finalized agent is saved with the
conversation, as on main. F36 now checks that a coordinator and a worker keep
their prompt, context and tools after a restart and use a project tool.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
…prompt it has not sent yet

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
enyst pushed a commit to OpenHands/OpenHands that referenced this pull request Nov 19, 2026
OpenHands/software-agent-sdk#6602 no longer takes coordinator_conversation_id
on POST or PATCH /api/projects; only "coordinator" starts the project's
coordinator. Drop it from the request type (Canvas never sent it).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwM9SZ2RcWff4zZivYR
Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst enyst mentioned this pull request Nov 19, 2026
1 of 5 tasks
claude and others added 4 commits November 19, 2026 17:50
tool_module_qualnames lists every tool module the server has registered,
so a worker's response names message_worker even though its agent does not
have it.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: openhands <openhands@all-hands.dev>
…-xwkk9b

# Conflicts:
#	openhands-agent-server/openhands/agent_server/event_service.py
#	openhands-sdk/openhands/sdk/conversation/types.py
@enyst
enyst requested a review from all-hands-bot November 19, 2026 19:47

@all-hands-bot all-hands-bot left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This review was posted by an AI agent (OpenHands).

Re-review at head 2aacbbd4 (delta since the approved aede9ccd: five branch commits, two merges from main, and one self-merge adopting #7012; mergeStateStatus is CLEAN). Verified in this workspace:

  • bd6c00c14 — a project never adopts an existing conversation. coordinator_conversation_id is removed from POST and PATCH /api/projects, so the id-collision surface from the previous reserved-id mechanism is gone entirely; a project's coordinator is always started by the server. I confirmed the role delivery is now closed: ProjectRole (prompt, <PROJECT> context, tools) is a finalize() parameter set only by ProjectService (coordinator and worker call sites), not readable from AgentLaunchAdditions or any client-supplied request field — no client can ask for a coordinator prompt or project tools. ACP agents are still refused for coordination (runs_in_a_project check).
  • 872916163 (self-merge) — sender/stop/update/paused/card "headers" renamed to "lines" per #7012, wording preserved; pure naming across project_wake.py, project_prompts.py, message_sender.py.
  • 3d952fc31/e9147274a/5ae3ad95a are F36 recipe corrections only.
  • Merges from main adopt #7007 (typed already-running errors) and #7010/#6925/#6811 upstream; no carried copies.

Test results at this exact head: Projects router + board + cards + parent/child + project tools + SDK launch suites 359 passed. CI on this exact head SHA: 54 checks success, 1 known skip, 0 failures.

No material findings on the current head.

✅ APPROVED

Only the server gives a conversation the Projects role and tools, so main's
launch path no longer strips Projects tools by class, and an ordinary child
is copied from its parent as on main. A Projects tool a conversation still
holds refuses to run when Projects is off or it is not in a project.

Co-authored-by: openhands <openhands@all-hands.dev>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ArwwJH94xGSYwfoyccT4E
@enyst

enyst commented Nov 19, 2026

Copy link
Copy Markdown
Member Author

The Persisted settings failure on 0fed774 is not from this PR. The job can't build its 1.63.0 baseline: the released openhands-sdk==1.63.0 imports litellm.llms.openai.chat.gpt_5_transformation, which the litellm version it now resolves no longer has. The same check fails on main (4b9a723, 19:12 UTC) and passed on this PR's previous head.


Generated by Claude Code

@enyst

enyst commented Nov 19, 2026

Copy link
Copy Markdown
Member Author

agent-server-tests on 0fed774 failed only test_conversation_service.py::test_prepare_for_sandbox_pause_blocks_new_hydration, a timing-sensitive test of main's sandbox-pause path that this PR doesn't touch. It fails the same way on other branches' runs of the same workflow, and it passes locally on this head (153/153 in that file, 10/10 repeated runs). The newer head ecbb860 runs it again.


Generated by Claude Code

@enyst
enyst requested a review from all-hands-bot November 19, 2026 22:02

@all-hands-bot all-hands-bot left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This review was posted by an AI agent (OpenHands).

Re-review at head ecbb860c (delta since the approved 2aacbbd4: one branch commit plus the merge of main/v1.63.0). Verified in this workspace:

  • 0fed774a3 — Projects: drop withheld tools and the child's role stripping. Since only the server assigns the Projects role and tools (via finalize(project_role=...), unchanged in scope), withheld_tools is removed from finalize(), and an ordinary child is copied from its parent as on main. The off-switch is still enforced where it matters: project tools resolve their project from the caller's tags and translate the API's 404 PROJECTS_OFF_DETAIL into an agent-visible refusal ("do not retry"), and a tool held by a conversation without a project tag refuses with _NOT_IN_PROJECT. Feature-flag tests and parent/child tests were trimmed to match; 245 tests (Projects router + board + cards + parent/child + project tools + feature flag) pass at this exact head.
  • The merge from main brings v1.63.0 and #6846 (verified-models additions); no carried copies of other PRs.
  • CI note: the Persisted settings check fails on this head (1 of 55) with ModuleNotFoundError: No module named 'litellm.llms.openai.chat.gpt_5_transformation' while generating baseline payloads from the published 1.63.0 baseline under --exclude-newer. I traced this to the main-side model-list commit ee062ba88 (#6846): the same check fails on that commit on main, succeeded at 98f0fe18d and 4b9a72375 on main, and the module imports fine in this workspace with litellm 1.108.3. The failure reproduces on main-line commits and is not introduced by this PR; the Projects code in this range is unaffected by it. Flagging as mergeStateStatus UNSTABLE context rather than a PR finding.

No material findings on the current head.

✅ APPROVED

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Feature]: Projects in the Agent Server, so a coordinator conversation can follow and steer its conversations

5 participants