Skip to content

Add aitools mcp login, logout and list --check - #7551

Merged
simonfaltum merged 1 commit into
mainfrom
aitools-mcp/07-login
Nov 19, 2026
Merged

simonfaltum merged 1 commit into
mainfrom
aitools-mcp/07-login

Conversation

@simonfaltum

@simonfaltum simonfaltum commented Nov 12, 2026 •

Copy link
Copy Markdown
Member

Part of a stack that adds the hidden databricks aitools mcp command group, on #7550 (aitools mcp serve). It adds provider sign-in for external MCP services such as GitHub, which each user signs in to before the service's tools work.

Changes

  • mcp login <service> prints the sign-in URL, opens it through libs/browser, and waits up to 5 minutes. Server and network errors while polling are retried until the timeout, which names the last one. An already signed-in caller gets "Already signed in" and no URL.
  • mcp logout <service> deletes the caller's credential, even when its service is no longer visible. login and logout refuse a changed profile before any request.
  • A FAILED or DEGRADED credential is deleted before the sign-in page opens.
  • Service names are refused only when they contain / or a control character. Other URL syntax, such as ?, # and %, is escaped in the request and is allowed.
  • mcp add --login signs in to each sign-in-required server that reached an agent. Its JSON gains a per-server sign_in object, given only to servers that reached an agent; a server that does not need sign-in gets NOT_REQUIRED. A failed sign-in is reported in both servers[].sign_in.error and the top-level error, which is intended, so consumers can read either. The command exits 1 and keeps the installs.
  • mcp list --check checks each profile's token, then each server. A drifted server or one that cannot be read is skipped, and says why. Each check request ends after 5 seconds. A needed sign-in names the login command.
  • Hints for a server pinned to a config file set DATABRICKS_CONFIG_FILE. libs/browser gains OpenOutputToStderr, and its default opener no longer waits for a browser its launcher leaves running.

Why

A sign-in is judged on the attempt that runs, so a FAILED or DEGRADED credential left by an earlier attempt is deleted before the page opens. A failed poll does not mean the user has left the provider's page, so it does not end the wait. One unreachable workspace held list --check for the SDK's five-minute retry budget.

Decisions to review

  • SignIn returns (alreadySignedIn, urlShown bool, err error), not a state. The command derives the JSON state and URL from them.
  • Polls retry 5xx and network errors. Cost: a longer wait while the error persists, up to the timeout.
  • Service names with ?, # or % are accepted, rather than refused locally. Cost: such names reach the API.
  • list --check sets DATABRICKS_CONFIG_FILE in the process environment while it checks a profile, because the token store reads the process environment.
  • logout sends the DELETE before it looks up the service, so a credential for a deleted service can be removed. Cost: a mistyped name costs one request.

Tests

  • TestOpenDoesNotWaitForLauncherChild in libs/browser failed before the fix. cmd/aitools: TestAddChecksBoundsAProfileThatNeverAnswers hung past five minutes before the fix.
  • Polls through server errors and the timeout's last error are covered in libs/aitools/mcp.
  • TestSignInDeletesDegradedCredential covers the DEGRADED deletion. TestPathSafeName and TestServiceNameAllowsEscapedURLSyntax cover the name check. TestSignInInstalledRecordsTheSignInOfEachServer covers sign_in for a server that reached no agent.
  • Acceptance list-check and login-flips-active pass.
  • go test -race ./cmd/aitools/ ./libs/aitools/mcp/... and GOOS=windows go vet on the same packages pass.
  • Not tested: the 5-minute timeout through the CLI (unit tests use milliseconds), and a real provider sign-in.

How to review

Refactors that keep behavior: the sign-in helpers shared by login, logout and add --login (libs/aitools/mcp/login.go, cmd/aitools/mcp_login.go). Read closely: the poll loop in libs/aitools/mcp/login.go, the bounded checks in cmd/aitools/mcp_list.go, and the default opener in libs/browser/browser.go, which changes how browser.Open waits.

This pull request and its description were written by Isaac.

@simonfaltum
simonfaltum changed the base branch from aitools-mcp/06a-bridge to aitools-mcp/06b-serve November 12, 2026 12:42
@github-actions

github-actions Bot commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor

Commit: e6ae53a11 | Workflow run

Regression test report

Selected 6 changed tests (limit=20, 0 not selected)

6 tests: regression=6 unreleased=0 coverage=0 failing=0
test branch main (cb94103) latest (v1.27.0)
TestAccept/cmd/aitools/mcp/add-text ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/help ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/hidden ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/list-and-remove ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/list-check ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/login-flips-active ✅ ❌ ❌
TestAccept/cmd/aitools/mcp/add-text: branch ✅ | main (cb94103) ❌ | latest (v1.27.0) ❌

main (cb94103):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/add-text/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcpadd-text243762036/002/output.txt
        @@ -12,7 +12,7 @@
         github-logfood  system.ai.github
           claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
           cursor       installed  [TEST_TMP_DIR]/home/.cursor/mcp.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        +  system.ai.github needs sign-in. Run: databricks aitools mcp login system.ai.github --profile logfood
           Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
         
         Cursor may ask you to approve genie_one_mcp-logfood, github-logfood. Run cursor-agent mcp enable <name> for each (`agent` in the Cursor docs), or approve them in Cursor's MCP settings.
        @@ -30,7 +30,7 @@
         github-logfood  system.ai.github
           claude-code  already installed, re-asserted  [TEST_TMP_DIR]/home/.claude.json
           cursor       already installed, re-asserted  [TEST_TMP_DIR]/home/.cursor/mcp.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        +  system.ai.github needs sign-in. Run: databricks aitools mcp login system.ai.github --profile logfood
           Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
         
         === Cursor config
        
    acceptance_test.go:1166: 
        LOG.create-genie_one_mcp-logfood: {
        LOG.create-genie_one_mcp-logfood:   "comment": "Ask questions about your data",
        LOG.create-genie_one_mcp-logfood:   "create_time": "2026-11-19T15:18:40.505339712Z",
        LOG.create-genie_one_mcp-logfood:   "created_by": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "metastore_id": "7a0e7a3c-0537-fabd-7a0e-7a3c0537fbb2",
        LOG.create-genie_one_mcp-logfood:   "name": "mcp-services/system.ai.genie_one_mcp",
        LOG.create-genie_one_mcp-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-genie_one_mcp-logfood: }
    acceptance_test.go:1166: 
        LOG.create-github-logfood: {
        LOG.create-github-logfood:   "comment": "Repositories, issues, pull requests",
        LOG.create-github-logfood:   "config": {
        LOG.create-github-logfood:     "source_connection": {
        LOG.create-github-logfood:       "name": "connections/github",
        LOG.create-github-logfood:       "options": {
        LOG.create-github-logfood:         "oauth_provider": "OAUTH_PROVIDER_GITHUB_MCP"
        LOG.create-github-logfood:       }
        LOG.create-github-logfood:     }
        LOG.create-github-logfood:   },
        LOG.create-github-logfood:   "create_time": "2026-11-19T15:18:40.533537376Z",
        LOG.create-github-logfood:   "created_by": "tester@databricks.com",
        LOG.create-github-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-github-logfood:   "metastore_id": "7a0e7a3c-06e6-3e74-7a0e-7a3c06e63f49",
        LOG.create-github-logfood:   "name": "mcp-services/system.ai.github",
        LOG.create-github-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-github-logfood: }

latest (v1.27.0):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/add-text/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcpadd-text3383035403/002/output.txt
        @@ -2,75 +2,29 @@
         === Text output: Cursor approval hint, and the sign-in step for github
         
         >>> [CLI] aitools mcp add --profile logfood --names system.ai.genie_one_mcp,system.ai.github --agents claude-code,cursor
        -Profile logfood ([DATABRICKS_URL])
        -Agents (named with --agents): claude-code, cursor
        +Error: unknown flag: --names
         
        -genie_one_mcp-logfood  system.ai.genie_one_mcp
        -  claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        -  cursor       installed  [TEST_TMP_DIR]/home/.cursor/mcp.json
        +Usage:
        +  databricks aitools [command]
         
        -github-logfood  system.ai.github
        -  claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        -  cursor       installed  [TEST_TMP_DIR]/home/.cursor/mcp.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        -  Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
        +Available Commands:
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
        -Cursor may ask you to approve genie_one_mcp-logfood, github-logfood. Run cursor-agent mcp enable <name> for each (`agent` in the Cursor docs), or approve them in Cursor's MCP settings.
        +Flags:
        +  -h, --help[=true|false]   help for aitools
         
        -=== Running it again re-asserts the entries, and nothing is written twice
        +Global Flags:
        +      --debug[=true|false]   enable debug logging
        +  -o, --output type          output type: text or json (default text)
        +  -p, --profile string       ~/.databrickscfg profile
        +  -t, --target string        bundle target to use (if applicable)
         
        ->>> [CLI] aitools mcp add --profile logfood --names system.ai.genie_one_mcp,system.ai.github --agents claude-code,cursor
        -Profile logfood ([DATABRICKS_URL])
        -Agents (named with --agents): claude-code, cursor
        +Use "databricks aitools [command] --help" for more information about a command.
         
        -genie_one_mcp-logfood  system.ai.genie_one_mcp
        -  claude-code  already installed, re-asserted  [TEST_TMP_DIR]/home/.claude.json
        -  cursor       already installed, re-asserted  [TEST_TMP_DIR]/home/.cursor/mcp.json
         
        -github-logfood  system.ai.github
        -  claude-code  already installed, re-asserted  [TEST_TMP_DIR]/home/.claude.json
        -  cursor       already installed, re-asserted  [TEST_TMP_DIR]/home/.cursor/mcp.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        -  Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
        -
        -=== Cursor config
        -
        ->>> cat home/.cursor/mcp.json
        -{
        -  "mcpServers": {
        -    "genie_one_mcp-logfood": {
        -      "args": [
        -        "aitools",
        -        "mcp",
        -        "serve",
        -        "--profile",
        -        "logfood",
        -        "--service",
        -        "system.ai.genie_one_mcp",
        -        "--agent",
        -        "cursor"
        -      ],
        -      "command": "[CLI]",
        -      "env": {
        -        "DATABRICKS_CONFIG_FILE": "[TEST_TMP_DIR]/.databrickscfg"
        -      }
        -    },
        -    "github-logfood": {
        -      "args": [
        -        "aitools",
        -        "mcp",
        -        "serve",
        -        "--profile",
        -        "logfood",
        -        "--service",
        -        "system.ai.github",
        -        "--agent",
        -        "cursor"
        -      ],
        -      "command": "[CLI]",
        -      "env": {
        -        "DATABRICKS_CONFIG_FILE": "[TEST_TMP_DIR]/.databrickscfg"
        -      }
        -    }
        -  }
        -}
        +Exit code: 1
        
    acceptance_test.go:1166: 
        LOG.create-genie_one_mcp-logfood: {
        LOG.create-genie_one_mcp-logfood:   "comment": "Ask questions about your data",
        LOG.create-genie_one_mcp-logfood:   "create_time": "2026-11-19T15:19:13.185860337Z",
        LOG.create-genie_one_mcp-logfood:   "created_by": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "metastore_id": "7a0e7a43-a121-266d-7a0e-7a43a121279e",
        LOG.create-genie_one_mcp-logfood:   "name": "mcp-services/system.ai.genie_one_mcp",
        LOG.create-genie_one_mcp-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-genie_one_mcp-logfood: }
    acceptance_test.go:1166: 
        LOG.create-github-logfood: {
        LOG.create-github-logfood:   "comment": "Repositories, issues, pull requests",
        LOG.create-github-logfood:   "config": {
        LOG.create-github-logfood:     "source_connection": {
        LOG.create-github-logfood:       "name": "connections/github",
        LOG.create-github-logfood:       "options": {
        LOG.create-github-logfood:         "oauth_provider": "OAUTH_PROVIDER_GITHUB_MCP"
        LOG.create-github-logfood:       }
        LOG.create-github-logfood:     }
        LOG.create-github-logfood:   },
        LOG.create-github-logfood:   "create_time": "2026-11-19T15:19:13.218700210Z",
        LOG.create-github-logfood:   "created_by": "tester@databricks.com",
        LOG.create-github-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-github-logfood:   "metastore_id": "7a0e7a43-a316-3f48-7a0e-7a43a316408a",
        LOG.create-github-logfood:   "name": "mcp-services/system.ai.github",
        LOG.create-github-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-github-logfood: }
TestAccept/cmd/aitools/mcp/help: branch ✅ | main (cb94103) ❌ | latest (v1.27.0) ❌

main (cb94103):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/help/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcphelp2121814975/002/output.txt
        @@ -19,9 +19,8 @@
         --profile lists the servers of that profile only. Unmanaged entries have no profile, so they are not
         listed then.
         
        -Without --check this command makes no network calls. With --check, a token is requested for each profile,
        -the sign-in state of each sign-in-required server is read, and the recorded binary is checked. It always exits 0
        -when entries have drifted or cannot be checked; an entry or agent config that cannot be checked is a warning.
        +This command makes no network calls. It always exits 0 when entries have drifted or cannot be checked;
        +an entry or agent config that cannot be checked is a warning.
         
         Usage:
           databricks aitools mcp list [flags]
        @@ -32,8 +31,7 @@
           databricks aitools mcp list --output json
         
         Flags:
        -      --check[=true|false]   Also check tokens, sign-in state and the recorded binary (uses the network)
        -  -h, --help[=true|false]    help for list
        +  -h, --help[=true|false]   help for list
         
         Global Flags:
               --debug[=true|false]   enable debug logging
        @@ -56,8 +54,7 @@
         this CLI is left in place and reported, and a --names run that names its server is refused. When an agent's config
         moved, the entry is removed from the file it was written to and from the file the agent reads now.
         Entries that run 'aitools mcp serve' without an install record are not chosen by any flag;
        -'aitools mcp list' shows them as unmanaged. Your sign-ins to services are kept; 'aitools mcp logout'
        -removes one.
        +'aitools mcp list' shows them as unmanaged. Provider credentials are not deleted.
         
         Usage:
           databricks aitools mcp remove [flags]

latest (v1.27.0):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/help/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcphelp2554969713/002/output.txt
        @@ -2,38 +2,27 @@
         === list and remove help show their examples
         
         >>> [CLI] aitools mcp list --help
        -List Databricks managed MCP servers installed by this CLI, grouped by profile, with the
        -coding agents each one is in. Entries that drifted from their profile or from the agent config are
        -reported. Entries in agent configs that this CLI did not write are listed as unmanaged and never changed.
        -Entries that run 'aitools mcp serve' but have no install record are listed as unmanaged too.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        -Each agent of a server has one of these statuses. The JSON output uses the same names, with underscores:
        -  installed        the entry is the one this CLI wrote
        -  changed          the entry still serves the server, but differs from what 'add' writes, for example in
        -                   an argument; running 'add' again repairs it
        -  modified         the entry was changed outside this CLI, so it no longer serves the server
        -  missing on disk  the entry is not in the agent config
        -  path changed     the agent now reads its config from another file
        -  unknown          the entry or the config could not be checked
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
         
        ---profile lists the servers of that profile only. Unmanaged entries have no profile, so they are not
        -listed then.
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
         
        -Without --check this command makes no network calls. With --check, a token is requested for each profile,
        -the sign-in state of each sign-in-required server is read, and the recorded binary is checked. It always exits 0
        -when entries have drifted or cannot be checked; an entry or agent config that cannot be checked is a warning.
        -
         Usage:
        -  databricks aitools mcp list [flags]
        +  databricks aitools [command]
         
        -Examples:
        -  databricks aitools mcp list
        -  databricks aitools mcp list --profile logfood
        -  databricks aitools mcp list --output json
        +Available Commands:
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
         Flags:
        -      --check[=true|false]   Also check tokens, sign-in state and the recorded binary (uses the network)
        -  -h, --help[=true|false]    help for list
        +  -h, --help[=true|false]   help for aitools
         
         Global Flags:
               --debug[=true|false]   enable debug logging
        @@ -41,40 +23,35 @@
           -p, --profile string       ~/.databrickscfg profile
           -t, --target string        bundle target to use (if applicable)
         
        +Use "databricks aitools [command] --help" for more information about a command.
        +
         >>> [CLI] aitools mcp remove --help
        -Remove Databricks managed MCP servers that this CLI installed.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        -Choose servers with --names (server names as listed by 'aitools mcp list', or service names; the
        -last part of a service name, such as github, is enough when only one schema has it), --profile
        -(every server of one profile), or --all. --profile also narrows --names or --all to one profile, and
        ---agents narrows any of them to some agents.
        -Interactive selection is not available yet, so one of --names, --profile or --all is required.
        -With --profile or --all and nothing installed to remove, the command succeeds and removes nothing.
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
         
        -Only entries this CLI wrote are removed. An entry that this CLI did not write, and that is not a Databricks
        -entry, is left in place, and the install record for its agent is dropped. A Databricks entry that was changed outside
        -this CLI is left in place and reported, and a --names run that names its server is refused. When an agent's config
        -moved, the entry is removed from the file it was written to and from the file the agent reads now.
        -Entries that run 'aitools mcp serve' without an install record are not chosen by any flag;
        -'aitools mcp list' shows them as unmanaged. Your sign-ins to services are kept; 'aitools mcp logout'
        -removes one.
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
         
         Usage:
        -  databricks aitools mcp remove [flags]
        +  databricks aitools [command]
         
        -Examples:
        -  databricks aitools mcp remove --names github-logfood
        -  databricks aitools mcp remove --profile logfood --agents cursor
        -  databricks aitools mcp remove --all
        +Available Commands:
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
         Flags:
        -      --agents string       Agents to remove from, comma-separated (default: every agent the server is installed for)
        -      --all[=true|false]    Remove every server this CLI installed
        -  -h, --help[=true|false]   help for remove
        -      --names string        Servers to remove: server names or service names, comma-separated
        +  -h, --help[=true|false]   help for aitools
         
         Global Flags:
               --debug[=true|false]   enable debug logging
           -o, --output type          output type: text or json (default text)
           -p, --profile string       ~/.databrickscfg profile
           -t, --target string        bundle target to use (if applicable)
        +
        +Use "databricks aitools [command] --help" for more information about a command.
TestAccept/cmd/aitools/mcp/hidden: branch ✅ | main (cb94103) ❌ | latest (v1.27.0) ❌

main (cb94103):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/hidden/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcphidden1514221292/002/output.txt
        @@ -8,9 +8,7 @@
         
         >>> [CLI] aitools mcp --help
         Databricks managed MCP servers give coding agents tools for Databricks services, such as GitHub or web search.
        -Use 'add' to install them into coding agents, bound to a Databricks profile, 'list' to see what is installed,
        -and 'remove' to uninstall them. Some services use your account with an external provider, such as GitHub:
        -'login' signs you in to one and 'logout' removes the sign-in.
        +Use 'add' to install them into coding agents, bound to a Databricks profile.
         
         Usage:
           databricks aitools mcp [command]
        @@ -18,8 +16,6 @@
         Available Commands:
           add         Install Databricks managed MCP servers into coding agents
           list        List Databricks managed MCP servers installed by this CLI
        -  login       Sign in to a Databricks managed MCP service with your account
        -  logout      Remove your sign-in to a Databricks managed MCP service
           remove      Remove Databricks managed MCP servers from coding agents
         
         Flags:
        @@ -58,7 +54,6 @@
               --force[=true|false]   Overwrite an existing server with the same name that this CLI did not write
           -h, --help[=true|false]    help for add
               --location string      Schema to resolve leaf names in, as <catalog>.<schema> (default: system.ai, or the one schema that the full names in --names share when there are no leaf names)
        -      --login[=true|false]   Sign in to each sign-in-required server that was installed, after the installs
               --name string          Server name to use instead of the generated one; only with exactly one service
               --names string         Services to install: full names (system.ai.github) or leaf names (github), comma-separated
         
        
    acceptance_test.go:1166: 
        LOG.aitools-help: Install Databricks skills and plugins into your coding agent so it can work
        LOG.aitools-help: effectively with Databricks resources (bundles, jobs, SQL, and more).
        LOG.aitools-help: 
        LOG.aitools-help: Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        LOG.aitools-help: 
        LOG.aitools-help: Skills and plugins are sourced from
        LOG.aitools-help: https://github.com/databricks/databricks-agent-skills
        LOG.aitools-help: 
        LOG.aitools-help: Usage:
        LOG.aitools-help:   databricks aitools [command]
        LOG.aitools-help: 
        LOG.aitools-help: Available Commands:
        LOG.aitools-help:   auto-install Turn the automatic installation of Databricks AI tools on or off
        LOG.aitools-help:   install      Install Databricks skills and plugins for coding agents
        LOG.aitools-help:   list         List installed skills and plugins
        LOG.aitools-help:   uninstall    Uninstall Databricks skills and plugins
        LOG.aitools-help:   update       Update installed skills and plugins
        LOG.aitools-help:   version      Show installed skills and plugins version
        LOG.aitools-help: 
        LOG.aitools-help: Flags:
        LOG.aitools-help:   -h, --help[=true|false]   help for aitools
        LOG.aitools-help: 
        LOG.aitools-help: Global Flags:
        LOG.aitools-help:       --debug[=true|false]   enable debug logging
        LOG.aitools-help:   -o, --output type          output type: text or json (default text)
        LOG.aitools-help:   -p, --profile string       ~/.databrickscfg profile
        LOG.aitools-help:   -t, --target string        bundle target to use (if applicable)
        LOG.aitools-help: 
        LOG.aitools-help: Use "databricks aitools [command] --help" for more information about a command.

latest (v1.27.0):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/hidden/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcphidden3803255900/002/output.txt
        @@ -7,23 +7,27 @@
         === The hidden group still answers help for its subcommands
         
         >>> [CLI] aitools mcp --help
        -Databricks managed MCP servers give coding agents tools for Databricks services, such as GitHub or web search.
        -Use 'add' to install them into coding agents, bound to a Databricks profile, 'list' to see what is installed,
        -and 'remove' to uninstall them. Some services use your account with an external provider, such as GitHub:
        -'login' signs you in to one and 'logout' removes the sign-in.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        +
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
        +
         Usage:
        -  databricks aitools mcp [command]
        +  databricks aitools [command]
         
         Available Commands:
        -  add         Install Databricks managed MCP servers into coding agents
        -  list        List Databricks managed MCP servers installed by this CLI
        -  login       Sign in to a Databricks managed MCP service with your account
        -  logout      Remove your sign-in to a Databricks managed MCP service
        -  remove      Remove Databricks managed MCP servers from coding agents
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
         Flags:
        -  -h, --help[=true|false]   help for mcp
        +  -h, --help[=true|false]   help for aitools
         
         Global Flags:
               --debug[=true|false]   enable debug logging
        @@ -31,39 +29,35 @@
           -p, --profile string       ~/.databrickscfg profile
           -t, --target string        bundle target to use (if applicable)
         
        -Use "databricks aitools mcp [command] --help" for more information about a command.
        +Use "databricks aitools [command] --help" for more information about a command.
         
         >>> [CLI] aitools mcp add --help
        -Install Databricks managed MCP servers into coding agents, bound to a Databricks profile.
        -Each server runs through 'databricks aitools mcp serve' with that profile's credentials.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        -Services come from --names: full names such as system.ai.github, or leaf names such as github. Leaf names
        -resolve in --location, which defaults to system.ai, or to the one schema that the full names in --names share when there are no leaf names.
        -Agents come from --agents. Without it, every detected MCP-capable agent is used.
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
         
        -Interactive selection is not available yet: in a terminal, pass both --names and --agents.
        -With --output json, both are required.
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
         
        -Supported agents: claude-code, cursor, codex, opencode, copilot, antigravity, gemini
        -
         Usage:
        -  databricks aitools mcp add [flags]
        +  databricks aitools [command]
         
        -Examples:
        -  # Install GitHub and web search for Claude Code and Cursor, bound to the dogfood profile
        -  databricks aitools mcp add --names github,web_search --agents claude-code,cursor --profile dogfood
        +Available Commands:
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
         Flags:
        -      --agents string        Agents to install for, comma-separated (default: every detected MCP-capable agent)
        -      --force[=true|false]   Overwrite an existing server with the same name that this CLI did not write
        -  -h, --help[=true|false]    help for add
        -      --location string      Schema to resolve leaf names in, as <catalog>.<schema> (default: system.ai, or the one schema that the full names in --names share when there are no leaf names)
        -      --login[=true|false]   Sign in to each sign-in-required server that was installed, after the installs
        -      --name string          Server name to use instead of the generated one; only with exactly one service
        -      --names string         Services to install: full names (system.ai.github) or leaf names (github), comma-separated
        +  -h, --help[=true|false]   help for aitools
         
         Global Flags:
               --debug[=true|false]   enable debug logging
           -o, --output type          output type: text or json (default text)
           -p, --profile string       ~/.databrickscfg profile
           -t, --target string        bundle target to use (if applicable)
        +
        +Use "databricks aitools [command] --help" for more information about a command.
        
    acceptance_test.go:1166: 
        LOG.aitools-help: Install Databricks skills and plugins into your coding agent so it can work
        LOG.aitools-help: effectively with Databricks resources (bundles, jobs, SQL, and more).
        LOG.aitools-help: 
        LOG.aitools-help: Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        LOG.aitools-help: 
        LOG.aitools-help: Skills and plugins are sourced from
        LOG.aitools-help: https://github.com/databricks/databricks-agent-skills
        LOG.aitools-help: 
        LOG.aitools-help: Usage:
        LOG.aitools-help:   databricks aitools [command]
        LOG.aitools-help: 
        LOG.aitools-help: Available Commands:
        LOG.aitools-help:   auto-install Turn the automatic installation of Databricks AI tools on or off
        LOG.aitools-help:   install      Install Databricks skills and plugins for coding agents
        LOG.aitools-help:   list         List installed skills and plugins
        LOG.aitools-help:   uninstall    Uninstall Databricks skills and plugins
        LOG.aitools-help:   update       Update installed skills and plugins
        LOG.aitools-help:   version      Show installed skills and plugins version
        LOG.aitools-help: 
        LOG.aitools-help: Flags:
        LOG.aitools-help:   -h, --help[=true|false]   help for aitools
        LOG.aitools-help: 
        LOG.aitools-help: Global Flags:
        LOG.aitools-help:       --debug[=true|false]   enable debug logging
        LOG.aitools-help:   -o, --output type          output type: text or json (default text)
        LOG.aitools-help:   -p, --profile string       ~/.databrickscfg profile
        LOG.aitools-help:   -t, --target string        bundle target to use (if applicable)
        LOG.aitools-help: 
        LOG.aitools-help: Use "databricks aitools [command] --help" for more information about a command.
TestAccept/cmd/aitools/mcp/list-and-remove: branch ✅ | main (cb94103) ❌ | latest (v1.27.0) ❌

main (cb94103):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/list-and-remove/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcplist-and-remove1893157584/002/output.txt
        @@ -10,7 +10,7 @@
         
         github-logfood  system.ai.github
           claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        +  system.ai.github needs sign-in. Run: databricks aitools mcp login system.ai.github --profile logfood
           Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
         
         >>> [CLI] aitools mcp add --profile dogfood --names system.ai.genie_one_mcp --agents claude-code
        
    acceptance_test.go:1166: 
        LOG.create-genie_one_mcp-dogfood: {
        LOG.create-genie_one_mcp-dogfood:   "comment": "Ask questions about your data",
        LOG.create-genie_one_mcp-dogfood:   "create_time": "2026-11-19T15:18:40.543482645Z",
        LOG.create-genie_one_mcp-dogfood:   "created_by": "tester@databricks.com",
        LOG.create-genie_one_mcp-dogfood:   "effective_owner": "tester@databricks.com",
        LOG.create-genie_one_mcp-dogfood:   "metastore_id": "7a0e7a3c-077d-fff0-7a0e-7a3c077e00b3",
        LOG.create-genie_one_mcp-dogfood:   "name": "mcp-services/system.ai.genie_one_mcp",
        LOG.create-genie_one_mcp-dogfood:   "updated_by": "tester@databricks.com"
        LOG.create-genie_one_mcp-dogfood: }
    acceptance_test.go:1166: 
        LOG.create-genie_one_mcp-logfood: {
        LOG.create-genie_one_mcp-logfood:   "comment": "Ask questions about your data",
        LOG.create-genie_one_mcp-logfood:   "create_time": "2026-11-19T15:18:40.500363600Z",
        LOG.create-genie_one_mcp-logfood:   "created_by": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-genie_one_mcp-logfood:   "metastore_id": "7a0e7a3c-04ec-0c59-7a0e-7a3c04ec0d81",
        LOG.create-genie_one_mcp-logfood:   "name": "mcp-services/system.ai.genie_one_mcp",
        LOG.create-genie_one_mcp-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-genie_one_mcp-logfood: }
    acceptance_test.go:1166: 
        LOG.create-github-logfood: {
        LOG.create-github-logfood:   "comment": "Repositories, issues, pull requests",
        LOG.create-github-logfood:   "config": {
        LOG.create-github-logfood:     "source_connection": {
        LOG.create-github-logfood:       "name": "connections/github",
        LOG.create-github-logfood:       "options": {
        LOG.create-github-logfood:         "oauth_provider": "OAUTH_PROVIDER_GITHUB_MCP"
        LOG.create-github-logfood:       }
        LOG.create-github-logfood:     }
        LOG.create-github-logfood:   },
        LOG.create-github-logfood:   "create_time": "2026-11-19T15:18:40.521457953Z",
        LOG.create-github-logfood:   "created_by": "tester@databricks.com",
        LOG.create-github-logfood:   "effective_owner": "tester@databricks.com",
        LOG.create-github-logfood:   "metastore_id": "7a0e7a3c-062d-edf9-7a0e-7a3c062def14",
        LOG.create-github-logfood:   "name": "mcp-services/system.ai.github",
        LOG.create-github-logfood:   "updated_by": "tester@databricks.com"
        LOG.create-github-logfood: }

latest (v1.27.0):

acceptance_test.go:1388: Diff:
        --- cmd/aitools/mcp/list-and-remove/output.txt
        +++ /tmp/TestAcceptcmdaitoolsmcplist-and-remove1965080564/002/output.txt
        @@ -2,195 +2,29 @@
         === Two profiles, add twice: logfood gets two services, dogfood gets one
         
         >>> [CLI] aitools mcp add --profile logfood --names system.ai.genie_one_mcp,system.ai.github --agents claude-code
        -Profile logfood ([DATABRICKS_URL])
        -Agents (named with --agents): claude-code
        +Error: unknown flag: --names
         
        -genie_one_mcp-logfood  system.ai.genie_one_mcp
        -  claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        +Usage:
        +  databricks aitools [command]
         
        -github-logfood  system.ai.github
        -  claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        -  system.ai.github needs sign-in. Run: DATABRICKS_CONFIG_FILE=[TEST_TMP_DIR]/.databrickscfg databricks aitools mcp login system.ai.github --profile logfood
        -  Public repositories only until a repository owner installs https://github.com/apps/databricks-github-connector (on AWS GovCloud, https://github.com/apps/databricks-gh-connector-gov-cloud).
        +Available Commands:
        +  auto-install Turn the automatic installation of Databricks AI tools on or off
        +  install      Install Databricks skills and plugins for coding agents
        +  list         List installed skills and plugins
        +  uninstall    Uninstall Databricks skills and plugins
        +  update       Update installed skills and plugins
        +  version      Show installed skills and plugins version
         
        ->>> [CLI] aitools mcp add --profile dogfood --names system.ai.genie_one_mcp --agents claude-code
        -Profile dogfood ([DATABRICKS_URL])
        -Agents (named with --agents): claude-code
        +Flags:
        +  -h, --help[=true|false]   help for aitools
         
        -genie_one_mcp-dogfood  system.ai.genie_one_mcp
        -  claude-code  installed  [TEST_TMP_DIR]/home/.claude.json
        +Global Flags:
        +      --debug[=true|false]   enable debug logging
        +  -o, --output type          output type: text or json (default text)
        +  -p, --profile string       ~/.databrickscfg profile
        +  -t, --target string        bundle target to use (if applicable)
         
        -=== list groups the servers by profile
        +Use "databricks aitools [command] --help" for more information about a command.
         
        ->>> [CLI] aitools mcp list
        -Profile dogfood  ([DATABRICKS_URL], config file [TEST_TMP_DIR]/.databrickscfg)
        -  genie_one_mcp-dogfood    system.ai.genie_one_mcp        Claude Code
         
        -Profile logfood  ([DATABRICKS_URL], config file [TEST_TMP_DIR]/.databrickscfg)
        -  genie_one_mcp-logfood    system.ai.genie_one_mcp        Claude Code
        -  github-logfood           system.ai.github               Claude Code
        -
        -=== list as JSON
        -
        ->>> [CLI] aitools mcp list --output json
        -{
        -  "profiles": [
        -    {
        -      "profile": "dogfood",
        -      "host": "[DATABRICKS_URL]",
        -      "config_file": "[TEST_TMP_DIR]/.databrickscfg",
        -      "status": "ok",
        -      "servers": [
        -        {
        -          "name": "genie_one_mcp-dogfood",
        -          "service": "system.ai.genie_one_mcp",
        -          "host": "[DATABRICKS_URL]",
        -          "sign_in_required": false,
        -          "status": "ok",
        -          "agents": [
        -            {
        -              "name": "claude-code",
        -              "path": "[TEST_TMP_DIR]/home/.claude.json",
        -              "status": "installed"
        -            }
        -          ]
        -        }
        -      ]
        -    },
        -    {
        -      "profile": "logfood",
        -      "host": "[DATABRICKS_URL]",
        -      "config_file": "[TEST_TMP_DIR]/.databrickscfg",
        -      "status": "ok",
        -      "servers": [
        -        {
        -          "name": "genie_one_mcp-logfood",
        -          "service": "system.ai.genie_one_mcp",
        -          "host": "[DATABRICKS_URL]",
        -          "sign_in_required": false,
        -          "status": "ok",
        -          "agents": [
        -            {
        -              "name": "claude-code",
        -              "path": "[TEST_TMP_DIR]/home/.claude.json",
        -              "status": "installed"
        -            }
        -          ]
        -        },
        -        {
        -          "name": "github-logfood",
        -          "service": "system.ai.github",
        -          "host": "[DATABRICKS_URL]",
        -          "sign_in_required": true,
        -          "status": "ok",
        -          "agents": [
        -            {
        -              "name": "claude-code",
        -              "path": "[TEST_TMP_DIR]/home/.claude.json",
        -              "status": "installed"
        -            }
        -          ]
        -        }
        -      ]
        -    }
        -  ],
        -  "unmanaged": [],
        -  "warnings": []
        -}
        -
        -=== remove --profile removes that profile only
        -
        ->>> [CLI] aitools mcp remove --profile dogfood
        -Removed genie_one_mcp-dogfood  system.ai.genie_one_mcp  profile dogfood  (Claude Code)
        -
        ->>> [CLI] aitools mcp list
        -Profile logfood  ([DATABRICKS_URL], config file [TEST_TMP_DIR]/.databrickscfg)
        -  genie_one_mcp-logfood    system.ai.genie_one_mcp        Claude Code
        -  github-logfood           system.ai.github               Claude Code
        -
        -=== A Databricks entry changed by hand is listed as modified; an unmanaged ug entry is listed and never changed
        -
        ->>> [CLI] aitools mcp list
        -Profile logfood  ([DATABRICKS_URL], config file [TEST_TMP_DIR]/.databrickscfg)
        -  genie_one_mcp-logfood    system.ai.genie_one_mcp        Claude Code
        -  github-logfood           system.ai.github               Claude Code (modified)
        -  changed outside this CLI in Claude Code: delete the entry from [TEST_TMP_DIR]/home/.claude.json by hand and run `databricks aitools mcp remove --names github-logfood` again
        -
        -Unmanaged Databricks MCP entries (not changed by this command):
        -  Claude Code    web_search           ug mcp-proxy
        -
        -=== --profile lists that profile only, without the unmanaged entries
        -
        ->>> [CLI] aitools mcp list --profile logfood
        -Profile logfood  ([DATABRICKS_URL], config file [TEST_TMP_DIR]/.databrickscfg)
        -  genie_one_mcp-logfood    system.ai.genie_one_mcp        Claude Code
        -  github-logfood           system.ai.github               Claude Code (modified)
        -  changed outside this CLI in Claude Code: delete the entry from [TEST_TMP_DIR]/home/.claude.json by hand and run `databricks aitools mcp remove --names github-logfood` again
        -
        -=== remove leaves the modified entry in place and says why
        -
        ->>> musterr [CLI] aitools mcp remove --names github-logfood,genie_one_mcp-logfood
        -Skipped github-logfood: Claude Code has an entry named github-logfood that was changed outside this CLI; left in place. To remove it, delete the entry from [TEST_TMP_DIR]/home/.claude.json by hand and run `databricks aitools mcp remove --names github-logfood` again
        -Error: nothing was removed: named MCP servers are left in place: github-logfood
        -
        ->>> jq .mcpServers | keys home/.claude.json
        -[
        -  "genie_one_mcp-logfood",
        -  "github-logfood",
        -  "web_search"
        -]
        -
        ->>> jq .mcp | [.servers[] | {name, profile, agents: (.agents | keys)}] home/.databricks/aitools/skills/.state.json
        -[
        -  {
        -    "name": "genie_one_mcp-logfood",
        -    "profile": "logfood",
        -    "agents": [
        -      "claude-code"
        -    ]
        -  },
        -  {
        -    "name": "github-logfood",
        -    "profile": "logfood",
        -    "agents": [
        -      "claude-code"
        -    ]
        -  }
        -]
        -
        -=== remove --all removes what this CLI wrote and nothing else
        -
        ->>> [CLI] aitools mcp remove --all --output json
        -{
        -  "removed": [
        -    {
        -      "name": "genie_one_mcp-logfood",
        -      "service": "system.ai.genie_one_mcp",
        -      "profile": "logfood",
        -      "agents": [
        -        "claude-code"
        -      ]
        -    }
        -  ],
        -  "skipped": [
        -    {
        -      "name": "github-logfood",
        -      "agent": "claude-code",
        -      "reason": "Claude Code has an entry named github-logfood that was changed outside this CLI; left in place. To remove it, delete the entry from [TEST_TMP_DIR]/home/.claude.json by hand and run `databricks aitools mcp remove --names github-logfood` again"
... (57 more lines truncated, raise DATABRICKS_CLI_TEST_MAX_LOG to see more)

2 more failure details omitted, see the full report in the run.

@eng-dev-ecosystem-bot

eng-dev-ecosystem-bot commented Nov 12, 2026 •

Copy link
Copy Markdown
Collaborator

✅ Integration test report

Commit: e6ae53a

Run: 44552070587

Env 💚​RECOVERED ✅​pass 🙈​skip Time
💚​ aws linux-2core-8gb 1 272 25 16:18
💚​ aws-windows-latest-4core-16gb 1 274 23 9:55
💚​ azure linux-2core-8gb 1 271 25 9:53
💚​ azure-windows-latest-4core-16gb 1 273 23 11:45
💚​ gcp linux-2core-8gb 1 272 25 13:43
💚​ gcp-windows-latest-4core-16gb 1 274 23 7:40
Test Name aws linux-2core-8gb aws-windows-latest-4core-16gb azure linux-2core-8gb azure-windows-latest-4core-16gb gcp linux-2core-8gb gcp-windows-latest-4core-16gb
💚​ TestAccept 💚​R 💚​R 💚​R 💚​R 💚​R 💚​R
Top 28 slowest tests (at least 2 minutes):
duration env testname
4:41 gcp linux-2core-8gb TestFilerWorkspaceFilesExtensionsReadDir
4:14 azure-windows-latest-4core-16gb TestFilerWorkspaceFilesExtensionsDelete
4:05 aws-windows-latest-4core-16gb TestAccept
3:44 azure-windows-latest-4core-16gb TestAccept
3:38 aws linux-2core-8gb TestFilerWorkspaceFilesExtensionsReadDir
3:25 gcp-windows-latest-4core-16gb TestAccept
3:06 aws-windows-latest-4core-16gb TestImportDirDoesNotOverwrite
3:00 azure-windows-latest-4core-16gb TestExportDir
2:59 aws linux-2core-8gb TestLock
2:59 aws linux-2core-8gb TestSyncIncrementalFileSync
2:56 aws linux-2core-8gb TestFilerWorkspaceFilesExtensionsDelete
2:48 aws linux-2core-8gb TestFilerWorkspaceFilesExtensionsRead
2:44 azure-windows-latest-4core-16gb TestImportDirDoesNotOverwrite
2:41 gcp linux-2core-8gb TestFilerRecursiveDelete/workspace_files_extensions
2:39 aws-windows-latest-4core-16gb TestFilerWorkspaceFilesExtensionsReadDir
2:37 gcp linux-2core-8gb TestExportDirDoesNotOverwrite
2:33 azure-windows-latest-4core-16gb TestFilerRecursiveDelete/workspace_files_extensions
2:29 aws linux-2core-8gb TestFilerReadWrite/workspace_files
2:29 azure-windows-latest-4core-16gb TestSyncFullFileSync
2:26 gcp linux-2core-8gb TestFilerReadWrite/workspace_files_extensions
2:24 azure-windows-latest-4core-16gb TestFilerWorkspaceFilesExtensionsRead
2:23 azure-windows-latest-4core-16gb TestFilerWorkspaceFilesExtensionsReadDir
2:19 aws linux-2core-8gb TestFilerRecursiveDelete/workspace_files_extensions
2:16 gcp-windows-latest-4core-16gb TestFilerRecursiveDelete/workspace_files_extensions
2:08 aws linux-2core-8gb TestImportDirWithOverwriteFlag
2:04 gcp linux-2core-8gb TestFilerWorkspaceFilesExtensionsRead
2:03 azure linux-2core-8gb TestFilerWorkspaceFilesExtensionsRead
2:00 aws linux-2core-8gb TestImportDirDoesNotOverwrite

@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from eec9d1a to 5b1d279 Compare November 12, 2026 14:37
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 5b1d279 to 9187003 Compare November 12, 2026 17:10
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 9187003 to 48fac6f Compare November 13, 2026 00:15
@simonfaltum
simonfaltum force-pushed the aitools-mcp/07-login branch 2 times, most recently from 0aa7987 to a08bc42 Compare November 13, 2026 07:16
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 7064bbc to 10450c9 Compare November 13, 2026 15:05
@pietern pietern added the aitools AI tools (databricks aitools) related changes label Nov 16, 2026
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 10450c9 to 9ec98a0 Compare November 16, 2026 12:47
@simonfaltum
simonfaltum force-pushed the aitools-mcp/07-login branch 2 times, most recently from 8ef469f to d779ae1 Compare November 16, 2026 13:19
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 9ec98a0 to 1d09e53 Compare November 16, 2026 13:19
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 1d09e53 to 01b4249 Compare November 16, 2026 13:32
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from 01b4249 to 11e6e00 Compare November 16, 2026 14:49
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch 2 times, most recently from 2a9b99f to c7d75cb Compare November 16, 2026 15:53
@simonfaltum
simonfaltum force-pushed the aitools-mcp/07-login branch 2 times, most recently from 109bc73 to 661f21e Compare November 17, 2026 10:52
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from c7d75cb to 0b9a961 Compare November 17, 2026 10:52
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch 2 times, most recently from 1eef59f to f2e0cd4 Compare November 17, 2026 18:57
@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch from f2e0cd4 to e41d98d Compare November 17, 2026 19:58
@simonfaltum
simonfaltum force-pushed the aitools-mcp/07-login branch 2 times, most recently from 9abcc11 to 79d56db Compare November 17, 2026 22:17
@simonfaltum
simonfaltum marked this pull request as ready for review November 18, 2026 16:20
@simonfaltum
simonfaltum requested review from a team and lennartkats-db as code owners November 18, 2026 16:20

@review-oss-bot review-oss-bot Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

Looks good to me — no issues found.

@anton-107 anton-107 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Approving: I found nothing that blocks the merge. I reviewed only what this PR adds on top of aitools-mcp/06b-serve (c5aa08b). At 79d56db, the touched packages pass with -race (libs/aitools/mcp, libs/browser, libs/exec, cmd/aitools, libs/testserver), and so does TestAccept/cmd/aitools/mcp.

The change is reversible. aitools mcp is still hidden, and the state file format doesn't change. The one change everyone sees is that the default browser.Open now returns when the launcher exits, and that is easy to revert.

Non-blocking notes:

  1. PathSafeName refuses more names than the API does. Unity Catalog names may contain %, # and ?; only ., space, / and control characters are forbidden. serviceResourceName already escapes the name with url.PathEscape, so login, logout and now serve refuse names the request would handle correctly. The system.ai services don't use these characters, so this only matters for a custom schema. Refusing just / and control characters would be enough.
  2. A DEGRADED credential keeps waiting without being reset. SignIn deletes a FAILED credential before opening the page, but a DEGRADED one is kept and the wait starts. list --check reports DEGRADED as "sign-in needed". Is there a check against the real API that the provider page turns a DEGRADED credential back to ACTIVE without a delete first? If it can't, the user waits the full 5 minutes.
  3. add --login reports the not-required state inconsistently. In signInInstalled, a server that doesn't need sign-in gets sign_in: {state: NOT_REQUIRED} even when it reached no agent. A server that needs sign-in and reached no agent gets no sign_in at all. Checking installedAnywhere before the not-required branch would make the two cases match.
  4. A failed sign-in appears twice in the add --login JSON: in servers[].sign_in.error and in the top-level error. TestTopLevelFailureKeepsErrorsOutsideAgents shows this is intended. Mentioning it in the JSON contract would help consumers know they can read either.

@simonfaltum

Copy link
Copy Markdown
Member Author

Thanks!

  1. 64aeab0: PathSafeName refuses only / and control characters now. ?, # and % are escaped when the request is built.
  2. 49ed447: a DEGRADED credential is deleted before the page opens, as a FAILED one is. I haven't seen a DEGRADED credential on a live workspace, so I couldn't check whether the page recovers one by itself; deleting it means the sign-in never waits on one.
  3. 8275026: signInInstalled checks whether a server reached an agent first, so a not-required server that reached none gets no sign_in, the same as a required one.
  4. The description now says a failed sign-in is reported in both places.

@simonfaltum
simonfaltum force-pushed the aitools-mcp/06b-serve branch 2 times, most recently from 171374b to ad9ad88 Compare November 19, 2026 14:29
@simonfaltum
simonfaltum requested a review from a team as a code owner November 19, 2026 14:29
Base automatically changed from aitools-mcp/06b-serve to main November 19, 2026 15:08
One commit for the restack onto main. The individual commits, up to 49ed447, are in the pull request's history.

Co-authored-by: Isaac <no-reply@databricks.com>
@simonfaltum
simonfaltum added this pull request to the merge queue Nov 19, 2026
Merged via the queue into main with commit 4285bf9 Nov 19, 2026
29 checks passed
@simonfaltum
simonfaltum deleted the aitools-mcp/07-login branch November 19, 2026 15:47
@eng-dev-ecosystem-bot

Copy link
Copy Markdown
Collaborator

❌ Integration test report

Commit: 4285bf9

Run: 44558946331

Env 💥​PANIC ❌​FAIL 🟨​KNOWN 🤯​MISS 💚​RECOVERED ✅​pass 🙈​skip Time
💥​ aws linux-2core-8gb 1 778 15 836 1411 320:00
💥​ aws-windows-latest-4core-16gb 1 2 517 20 1006 1439 320:00
💥​ azure linux-2core-8gb 1 6 545 14 1044 1419 320:00
💥​ azure-windows-latest-4core-16gb 1 2 6 472 12 1033 1447 320:00
💥​ gcp linux-2core-8gb 1 709 15 892 1417 320:00
💥​ gcp-windows-latest-4core-16gb 1 637 15 880 1445 320:00
1256 interesting tests: 1231 MISS, 9 RECOVERED, 6 PANIC, 6 KNOWN, 4 FAIL
Test Name aws linux-2core-8gb aws-windows-latest-4core-16gb azure linux-2core-8gb azure-windows-latest-4core-16gb gcp linux-2core-8gb gcp-windows-latest-4core-16gb
🤯​ TestAccept 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/apps/job_permissions 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/apps/job_permissions/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/apps/job_permissions/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_doesnot_exist 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_doesnot_exist/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_doesnot_exist/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_not_deployed 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_not_deployed/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/artifacts/artifact_path_with_volume/volume_not_deployed/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cli_defaults 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/cli_defaults/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cli_defaults/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_cross_target 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_cross_target/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_cross_target/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_permissions 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_permissions/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_permissions/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_remote_addition 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_remote_addition/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/cluster_policy_remote_addition/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/config_edits 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/config_edits/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/config_edits/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/dashboard_etag 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/dashboard_etag/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/dashboard_etag/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/external_cluster_policy_cross_target 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/external_cluster_policy_cross_target/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/external_cluster_policy_cross_target/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/flushed_cache 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/flushed_cache/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/flushed_cache/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/formatting_preserved 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/formatting_preserved/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/formatting_preserved/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_fields 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/job_fields/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_fields/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_multiple_tasks 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/job_multiple_tasks/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_multiple_tasks/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_params_variables 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/job_params_variables/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_params_variables/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_pipeline_task 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/job_pipeline_task/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/job_pipeline_task/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/json_status_missing_state 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/json_status_missing_state/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/json_status_missing_state/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/json_status_selectors 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/json_status_selectors/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/json_status_selectors/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/multiple_files 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/multiple_files/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/multiple_files/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/multiple_resources 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/multiple_resources/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/multiple_resources/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/output_json 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/output_json/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/output_json/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/output_no_changes 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/output_no_changes/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/output_no_changes/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/pipeline_fields 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/pipeline_fields/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/pipeline_fields/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/policy_injected_cluster_fields 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/policy_injected_cluster_fields/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/policy_injected_cluster_fields/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/resolve_variables 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/resolve_variables/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/resolve_variables/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/select_basic 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/select_basic/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/select_basic/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/select_multiple 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/select_multiple/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/select_multiple/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/skip_deployment 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/skip_deployment/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/skip_deployment/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/skip_permissions 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/skip_permissions/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/skip_permissions/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/target_override 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/target_override/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/target_override/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/task_rename_revert 🤯​M 🙈​s 🤯​M 🙈​s 🤯​M 🙈​s
🤯​ TestAccept/bundle/config-remote-sync/task_rename_revert/DMS= 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/config-remote-sync/task_rename_revert/DMS=true 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/debug/fetch-repository-info 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/debug/fetch-repository-info/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/debug/fetch-repository-info/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/empty-bundle 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/empty-bundle/DATABRICKS_BUNDLE_ENABLE_EXPERIMENTAL_YAML_SYNC=/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/empty-bundle/DATABRICKS_BUNDLE_ENABLE_EXPERIMENTAL_YAML_SYNC=/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/empty-bundle/DATABRICKS_BUNDLE_ENABLE_EXPERIMENTAL_YAML_SYNC=true/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/empty-bundle/DATABRICKS_BUNDLE_ENABLE_EXPERIMENTAL_YAML_SYNC=true/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/files/no-snapshot-sync 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/files/no-snapshot-sync/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/files/no-snapshot-sync/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/basic 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/basic/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/breakglass 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/breakglass/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/can-manage 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/can-manage/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/destroy-redeploy 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/destroy-redeploy/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/dev-mode 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/dev-mode/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/legacy-state 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/legacy-state/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/missing-snapshot 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/missing-snapshot/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/no-artifacts 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/no-artifacts/DMS=/READPLAN= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/no-artifacts/DMS=/READPLAN=1 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/permissions-change 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/permissions-change/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/select 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/immutable/select/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/mlops-stacks 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/mlops-stacks/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/mlops-stacks/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/spark-jar-task 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/spark-jar-task/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deploy/spark-jar-task/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/alert 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/alert/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/alert/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/catalog 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/catalog/DMS= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/catalog/DMS=true 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/cluster 🤯​M 🤯​M ✅​p 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/cluster/DMS= 🤯​M 🤯​M ✅​p 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/cluster/DMS=true 🤯​M 🤯​M ✅​p 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/DMS=/READPLAN= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/DMS=/READPLAN=1 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/DMS=true/READPLAN= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/DMS=true/READPLAN=1 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/recreation 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/recreation/DMS=/READPLAN= 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
🤯​ TestAccept/bundle/deployment/bind/dashboard/recreation/DMS=/READPLAN=1 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M 🤯​M
Top 50 slowest tests (at least 2 minutes):
duration env testname
18:44 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
16:34 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
15:54 azure-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=
14:35 gcp linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
14:07 aws linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
13:37 aws-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
12:52 azure-windows-latest-4core-16gb TestAccept/bundle/resources/schemas/rename_then_change/DMS=true
12:28 azure linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
12:19 aws-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
11:56 aws linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
11:52 gcp linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
11:26 azure-windows-latest-4core-16gb TestAccept/bundle/resources/schemas/rename_then_change/DMS=
10:59 azure linux-2core-8gb TestAccept/bundle/resources/materialization_sets/immutable/DMS=
10:46 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started-toggle/DMS=true
10:39 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started-omitted/DMS=true
10:37 azure linux-2core-8gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=true
10:28 aws-windows-latest-4core-16gb TestAccept/bundle/resources/schemas/rename_then_change/DMS=
10:28 gcp linux-2core-8gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=
10:21 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started-toggle/DMS=
10:00 azure linux-2core-8gb TestAccept/bundle/resources/jobs/shared-root-path/DMS=
9:55 azure-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
9:38 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started-omitted/DMS=
9:34 gcp linux-2core-8gb TestAccept/pipelines/run/refresh-flags
9:32 azure linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started-toggle/DMS=
9:31 azure linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=
9:31 gcp linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started-omitted/DMS=
9:15 gcp linux-2core-8gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=true
9:10 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=true
9:10 azure-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started/DMS=true
9:05 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started-toggle/DMS=
8:58 azure linux-2core-8gb TestAccept/bundle/templates/default-python/combinations/classic/DLT=no/DMS=/NBOOK=no/PY=yes/READPLAN=
8:56 aws-windows-latest-4core-16gb TestAccept/bundle/resources/registered_models/basic/DMS=
8:54 aws-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=true
8:47 azure linux-2core-8gb TestAccept/bundle/resources/clusters/lifecycle-started-toggle/DMS=
8:46 aws-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/lifecycle-started/DMS=
8:46 aws-windows-latest-4core-16gb TestAccept/bundle/resources/schemas/rename_then_change/DMS=true
8:40 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/lifecycle-started-toggle/DMS=true
8:40 aws linux-2core-8gb TestAccept/bundle/resources/apps/lifecycle-started-omitted/DMS=true
8:35 azure linux-2core-8gb TestAccept/bundle/templates/default-python/integration_classic/DMS=/UV_PYTHON=3.9
8:30 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/clusters/deploy/local_ssd_count/DMS=true
8:23 azure-windows-latest-4core-16gb TestAccept/bundle/templates/default-python/combinations/serverless/DLT=yes/DMS=/NBOOK=no/PY=yes/READPLAN=
8:09 azure-windows-latest-4core-16gb TestAccept/bundle/templates/default-python/combinations/serverless/DLT=no/DMS=/NBOOK=no/PY=no/READPLAN=1
7:54 azure linux-2core-8gb TestAccept/bundle/resources/materialization_sets/batch/DMS=true
7:53 gcp linux-2core-8gb TestAccept/bundle/resources/experiments/basic/DMS=true
7:53 aws linux-2core-8gb TestAccept/bundle/resources/materialization_sets/immutable/DMS=true
7:52 aws-windows-latest-4core-16gb TestAccept/bundle/resources/materialization_sets/batch/DMS=
7:51 azure linux-2core-8gb TestAccept/bundle/templates/default-python/combinations/serverless/DLT=no/DMS=/NBOOK=no/PY=yes/READPLAN=
7:48 gcp linux-2core-8gb TestAccept/bundle/resources/clusters/deploy/local_ssd_count/DMS=
7:46 gcp-windows-latest-4core-16gb TestAccept/bundle/resources/apps/inline_config/DMS=
7:44 gcp linux-2core-8gb TestAccept/bundle/resources/clusters/lifecycle-started-toggle/DMS=
(1107 table rows omitted to keep the report under 60000 bytes)

idelafuente-oc pushed a commit to idelafuente-oc/cli that referenced this pull request Nov 20, 2026
…databricks#7553)

This pull request is part of a stack that adds the hidden `databricks
aitools mcp` command group, on top of databricks#7551. A service that needs
provider sign-in rejects `tools/list` until the user signs in, so the
agent has no tool to start it. The proxy now offers one `authenticate`
tool that runs the sign-in.

## Changes
- A challenge makes `tools/list` return only `authenticate`. A challenge
is a JSON-RPC `-32006`, a `-32042` answer with a sign-in URL, or an
error result that names a sign-in URL. A `-32042` URL is accepted on the
profile's host, or on any host whose `o=` or `w=` query names the bound
workspace.
- On a server that needs sign-in, a 401 is a challenge only when its
body asks for sign-in. Any other 401 gets the `databricks auth login`
hint. A server that needs no sign-in has no such handling, so its 401
passes through unchanged.
- A challenged `tools/call` names `authenticate` and the login command.
If `tools/list` still passes, `authenticate` is added to its first page
until a call succeeds or the sign-in completes.
- If the service still challenges after an "Already signed in" answer, a
challenged `tools/call` is answered with the `mcp logout` command to run
first, since neither `authenticate` nor `login` can fix it then.
- `authenticate` runs `login`'s sign-in, sends progress naming the page,
and sends `tools/list_changed` when signed in. A timeout names the URL,
the last failed check and the login command.
- The known-service note is printed by `mcp add`, by `mcp login` and by
the authenticate tool's sign-in. `mcp list` does not read it.
- The proxy never deletes an active credential. `SignIn` deletes a
FAILED or DEGRADED one before it shows the page.
- The page opens with no stdin, since stdin carries the protocol, and
stays open when the request is cancelled.
- A server with an install record takes its workspace ID from the
record, so its sign-in makes no request for it.
- A server with no install record reads the service at start. A rejected
token there exits 2 with the login hint.

## Why
Discovery fails before sign-in, so the agent can only start a sign-in
through a tool that exists before it. A rejected Databricks token is not
fixed by a provider sign-in, so it gets the login hint instead.

## Decisions to review
- No automatic recovery when a credential reads as active but the
service keeps rejecting it. Deleting and re-creating an active
credential would also delete working ones, for example while the gateway
catches up with a sign-in just made in a terminal. Cost: a stale
provider grant needs one `mcp logout`.
- A 401 that does not ask for sign-in is an auth failure, whatever its
body. Cost: a gateway that answers a missing provider sign-in with a
bare 401 gets the login hint instead of a sign-in.
- A page is not reopened after a cancelled attempt that showed it, so a
retry does not open a second tab.

## Tests
- `libs/aitools/mcp/proxy/challenge_test.go` covers the challenge
bodies, 401s, call-only challenges, the logout answer, page reopening,
queued and cancelled calls, and progress.
`TestSignInRequiredNamesTheLogoutWhileStillChallenged` covers the logout
answer after "Already signed in".
- `TestServeSignInUsesTheRecordedWorkspaceID` covers the workspace ID
taken from the record.
- Acceptance `serve-authenticate-tool` runs the stdio flow with
`BROWSER=none`.
- `go test -race ./cmd/aitools/ ./libs/aitools/mcp/...`, the
`TestAccept/cmd/aitools/mcp` acceptance tests and `GOOS=windows go vet`
on the same packages pass.
- Not tested: a live `-32042` challenge, and a real browser.

## How to review
Refactors that keep behavior: the sign-in path that `mcp login` and the
proxy share (`libs/aitools/mcp/login.go`, `cmd/aitools/mcp_serve.go`),
and the shared JSON-RPC response envelope in `libs/aitools/mcp/proxy/`.
Read closely: `libs/aitools/mcp/proxy/challenge.go` (new), its `Result`
method and the 401 rule, and `serveChallenge` in
`cmd/aitools/mcp_serve.go`.

This pull request and its description were written by Isaac.

Co-authored-by: Isaac <no-reply@databricks.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

aitools AI tools (databricks aitools) related changes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants