Repository navigation
feat(task-steps): native OSWorld initialize and evaluate on computer envs (FD-5492) - #606
williamloo-scale wants to merge 9 commits into
Conversation
|
Review the following changes in direct dependencies. Learn more about Socket for GitHub.
|
|
Warning Review the following alerts detected in dependencies. According to your organization's Security Policy, it is recommended to resolve "Warn" alerts. Learn more about Socket for GitHub.
|
…envs Adds the osworld_deploy_controller, osworld_initialize and osworld_evaluate steps and the osworld-server computer app (agent_env.osworld), so an OSWorld task (classic init_config/evaluator or an OSWorld-V2 task_NNN.py) runs on a framework computer env. - The OSWorld controller runs in a container sandbox of its own beside the computer, in remote mode, so the grader, a task's checks and the judge's key stay out of the agent's reach (a computer's apps run as its session user). - The computer runs only the OSWorld guest server, as the osworld-server app. Its card advertises urn:agentenv:osworld-guest/v1 at the prefix the app forwards to the guest server; the deploy step resolves the guest server's address from the computer's card (stored, else live until the app answers) and refuses a computer without the app before creating a sandbox. - osworld_initialize: OSWorld-V2 via /reset + /initialize_task; classic via /setup with the whole init_config. - osworld_evaluate: OSWorld-V2 via /evaluate; classic sends every func to the controller's /evaluate (async-polled, blocking fallback) and combines by conj; verifications shape matches cua_evaluate's. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…tion test (FD-5492) A computer env with the osworld-server app, the controller beside it, a classic task set up and graded through it: pass scores 1.0, fail 0.0. Running it found: - The controller got the guest server's address as this process reaches it; a local container reaches this machine's loopback by another name. It now gets it, and the model endpoint, through url_from_sandbox, as an agent's deploy does. - supervisord looks a bare command up on PATH, so the guest server's service never started: it runs as `bash start-osworld-server.sh`. - python-xlib refuses to connect without an Xauthority file; the start script makes an empty one. - pyautogui's screenshots need scrot on X11; the app installs it. - A deployed env's record is stored once, when it is deployed, so the controller's address written on it lasted only the process. It is recorded in the run's context metadata too, and read from there. tst/integration/computer/osworld_test.py stages the app from OSWORLD_SERVER_SOURCE and runs the controller image OSWORLD_CONTROLLER_IMAGE (one with scale-cua #360); it skips without them. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…ility (FD-5492) It needs scale-cua's guest server source and a controller image, which this repository doesn't carry, so it skips without them. CI's skip policy fails a skip that isn't a declared capability gap, so the skip names osworld_sources, the slow tier allows it as it does mcp_server_sources, and CONTRIBUTING asks a change to the OSWorld code to run the test with both set and say so. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
… addresses, and review fixes (FD-5492) - osworld_deploy_controller starts the controller with CONTROLLER_AUTH_TOKEN, an HMAC of the env's control token, and refuses one that answers a call without it: its port can be reachable from the computer, and so by the agent. Every controller call carries the token (controller.Endpoint). - The controller's sandbox is created link by link, each given the guest server and model endpoint as its own sandboxes reach them; a chain's url_from_sandbox can't say for the link it ends up on. - osworld_evaluate refuses a result/expected/options list whose length isn't the funcs'. - A failed setup is reported by its error alone, bounded, never the controller's whole answer. - Context failures raise StepContextError; the controller's disk is a named constant. - The osworld-server app pins Pillow 13.0.0, past every Pillow advisory. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…ry shipped agent's budget stop (FD-5492) - osworld_deploy_controller calls the link's prepare_image before creating its container, so an image stored as its build context alone is built rather than pulled, and lets go a sandbox the link says can't run the image for the next link's, as the provider chain does. - osworld_evaluate reads Claude Code's and Codex's "Reached maximum number of turns (N)" and each CLI's "<name> timed out after Ns" as a budget stop, grading the desktop and marking partial credit, beside the harness's own sentences. A stall, a tool's timeout or any other error still isn't one. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
b1e420c to
c3a2f60
Compare
…ter (FD-5492) osworld_deploy_controller, osworld_initialize and osworld_evaluate run on computer envs, so they live in task_steps/computer/osworld beside snapshot_computer_env. No behavior change. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…world (FD-5492) The controller client, the guest-server lookup, staging and its CLI, and the osworld-server computer app move from agent_env.osworld into the steps' own package, agent_env.task_step.task_steps.computer.osworld, so OSWorld lives in one place beside the other computer steps. The unit tests move to tst/unit/task_step/osworld. No behavior change. Stage the app with: python -m agent_env.task_step.task_steps.computer.osworld stage ... Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
| logger = logging.getLogger(__name__) | ||
|
|
||
| #: The controller image's artifact id when a step names none. | ||
| DEFAULT_CONTROLLER_IMAGE_ARTIFACT_ID = "cua-controller" |
There was a problem hiding this comment.
The question to settle before the rest: should this be in core?
The controller this step deploys is built from scale-cua, which is private. This default names an image in Scale's store, and the env the controller gets (PROVIDER=remote, REMOTE_VM_IP, REMOTE_SERVER_HEADERS, CONTROLLER_AUTH_TOKEN) is scale-cua's contract. Upstream OSWorld has no HTTP controller, no remote provider and no /evaluate_async. So nobody outside Scale can run these three steps, and the end-to-end test can never run in public CI: it adds an osworld_sources skip to the slow tier instead. AGENTS.md asks core to stay vendor-neutral and not to add a seam that nothing in the repository consumes or defaults. Moving the package under task_steps/computer puts it further into core.
Two ways out:
- A plugin: the steps through
[task_steps] implsand the app as its own env, from scale-cua or agentenvhub-sdk. Every piece already has an extension point. - Public pieces: the guest server (
main.py,pyxcursor.py) is upstream OSWorld's, Apache-2.0, so it could be a pinned download checked by sha256, as the drivers' installs are. The controller would need a public image built in this repository, so CI runs the test.
There was a problem hiding this comment.
Agreed that as written nobody outside Scale can run these steps: the controller image comes from private scale-cua, and the env it gets (PROVIDER=remote, REMOTE_VM_IP, REMOTE_SERVER_HEADERS, CONTROLLER_AUTH_TOKEN) is scale-cua's contract. We'll take both of your ways out, in parallel:
-
Now, Scale-only: a plugin. The three steps, the controller client, the guest lookup and the
osworld-serverapp go into agentenvhub-sdk asagent_env.task_stepsentry points, next to thecua_*steps the hub already runs from there. That unblocks the hub (scaleapi/scaleapi#174344) without anything vendor-specific in core. -
In parallel, core: public pieces. This PR stays open as the core version and doesn't merge until the steps run on public parts:
- Guest server: upstream OSWorld's
main.pyandpyxcursor.py(Apache-2.0) as a pinned download checked by sha256, like the drivers' installs, instead of staging scale-cua's copy. - Controller: an image built in this repository from public code, so the end-to-end test runs in public CI and the
osworld_sourcesskip goes away. - Default:
controller_image_artifact_iddefaults to that public image, not a Scale store id. - Contract: the env the step sets becomes the public controller's documented contract.
Once core has these, the plugin's copies go away and the hub uses the core steps.
- Guest server: upstream OSWorld's
Comments 2 to 5 are fixed here (20fdec6), so both versions start from the same code. On staging, from comment 5: in core the app becomes a bundle folder beside the other shipped apps, with an agent-env subcommand, as part of step 2.
| CONTROLLER_TOKEN_ENV: guest.controller_token(deployed)} | ||
| config = get_config() | ||
| try: | ||
| env["LITELLM_API_KEY"] = config.get_litellm_api_key() |
There was a problem hiding this comment.
Per-run judge overrides don't reach the controller. This passes the configured [model] key only, so a run's user_overrides.judge_litellm_api_key (or litellm_api_key) is ignored. osworld_evaluate reads only its own judge_model field, not user_overrides.judge_model. rubrics_verifier honours all three (rubrics_verifier.py:772 and 814), and AGENTS.md makes user_overrides the contract for per-run keys.
So agent-env task run --judge-litellm-api-key K --judge-model M grades the judge funcs with another key and the controller's default model. With no [model] key configured, the judge gets no key, and those funcs score 0.
Suggest resolving the key as rubrics_verifier does, config.resolve_reference(overrides.get("judge_litellm_api_key") or overrides.get("litellm_api_key")), then falling back to [model]. And let user_overrides.judge_model set the judge funcs' model ahead of the step's own judge_model, as rubrics_verifier does.
There was a problem hiding this comment.
Fixed in 20fdec6, following rubrics_verifier:
- Key:
osworld_deploy_controllerpassesconfig.resolve_reference(user_overrides.judge_litellm_api_key or user_overrides.litellm_api_key), and falls back to[model]'s key when the override is unset or resolves to nothing. - Base URL:
judge_litellm_base_url/litellm_base_urlwork the same way. - Model:
osworld_evaluatesendsuser_overrides.judge_modelas the judge funcs'model. It wins over a func's ownmodeloption and over the step'sjudge_model, and it's recorded withcontext.record_model("judge", ...).
Tests cover the judge key, the fallback to the run's key and then to [model], and the model override.
| if len(scores) == 1: | ||
| score = scores[0] | ||
| else: | ||
| score = aggregate_score([{"score": s, "result": s == PASS_SCORE} for s in scores], |
There was a problem hiding this comment.
Several funcs are combined as pass/fail, which isn't how OSWorld combines them. A func counts as passed only at exactly 1.0, and ALL_PASS/ANY_PASS return 1 or 0, while a single func keeps its partial score.
OSWorld's DesktopEnv.evaluate (desktop_env.py L580-587):
andreturns 0 when any metric is 0, else their mean;orreturns 1 when any metric is 1, else their max.
So and over [1.0, 0.5] scores 0.0 here and 0.75 in OSWorld, and or over [0.5, 0.3] scores 0.0 here and 0.5 in OSWorld. If this deliberately matches cua_evaluate, say so in the docstring. Otherwise suggest following OSWorld's rule, since these steps exist to give OSWorld's scores.
There was a problem hiding this comment.
Fixed in 20fdec6. Several funcs now combine as OSWorld's DesktopEnv.evaluate does: and is 0 when any func scores 0, otherwise their mean; or is their max. That's also what the controller does itself (scale_cua_environment.py:871-880), so a task scores the same either way. One func keeps its own score, as before.
The conj test has your two cases: and over [1.0, 0.5] scores 0.75, and or over [0.5, 0.3] scores 0.5.
|
|
||
| def __init__(self, id: str, version: Optional[int], env_id: str, evaluator: Optional[dict] = None, | ||
| verifier_id: Optional[str] = None, osworld_v2_task_path: Optional[str] = None, | ||
| osworld_v2_task_url: Optional[str] = None, judge_model: Optional[str] = None, |
There was a problem hiding this comment.
Smaller ones, none blocking:
osworld_v2_task_urlis called the older name, here and in initialize.py:26, but these step types are new, so no stored task has it. Unless hub tasks are being converted into these types, suggest dropping it (AGENTS.md: no defensive code for a case no caller produces)._BUDGET_EXHAUSTED(line 37) matches each shipped agent's stop sentence as free text, so a reworded message silently turns partial credit into a skip. A structured stop reason onPromptResponsewould hold up better. Short of that, a test could check each pattern against the agent's own source.start-osworld-server.shrestarts the server in its own loop (line 9), under a service manager that already restarts services. One restart policy is enough.- The timeouts at controller.py:84 (60) and controller.py:131 and 146 (15) want names, as Greptile noted.
- Staging is
python -m agent_env.task_step.task_steps.computer.osworld stage ...since the move. Anagent-envsubcommand, or a bundle folder likebrowsers/envs/chromium, would be easier to find, and would put the app where the other shipped apps live.
There was a problem hiding this comment.
All in 20fdec6 unless noted:
osworld_v2_task_url: kept, because hub tasks are being converted into these types. scaleapi/scaleapi#174344 makes the hub's CUA importer emitosworld_initialize/osworld_evaluatewithosworld_v2_task_urlwhen a CSV row gives a task URL, as it does for thecua_*steps. The docstrings now say that instead of "older name"._BUDGET_EXHAUSTED: addedtest_each_agents_budget_stop_is_read_as_its_source_writes_it. For each in-repo bridge (codex, claude-code, gemini, hermes, letta-code, opencode), it checks that the bridge's source still contains its stop f-string and that the pattern matches that sentence, so a rewording fails the test. A structured stop reason onPromptResponsewould hold up better; it touches every bridge, so I'd rather do it as its own change.start-osworld-server.sh: the loop is gone. Itexecs the server, and the composer's supervisord program (autorestart=true, composer.py:960) restarts it. The end-to-end test passes with this.- Timeouts: named
START_TIMEOUT_S(60) andREADY_CALL_TIMEOUT_S(15). - Staging: this depends on where the code ends up after your first comment, so I'll settle it there.
…object answers (FD-5492) - osworld_deploy_controller gives the judge the run's key and endpoint (user_overrides.judge_litellm_api_key / litellm_api_key, judge_litellm_base_url / litellm_base_url) ahead of [model]'s, as rubrics_verifier does. - osworld_evaluate gives judge funcs user_overrides.judge_model ahead of their own option and the step's judge_model, records it as the run's judge model, and treats every agent_judge_* variant as a judge (agent_judge_zip, _blend, _godot, _log, _gz, _fcstd were sent as deterministic funcs, without the model or the rubric). - Several funcs combine as OSWorld's DesktopEnv.evaluate and the controller do: `and` is 0 when any func scores 0, else their mean; `or` is their max. - A controller answer that isn't a JSON object raises ControllerCallError, which the steps score 0, instead of an AttributeError that failed the step. - The controller timeouts are named; start-osworld-server.sh leaves restarts to supervisord; the osworld_v2_task_url field is documented as what the hub's importer writes. - A test pins each shipped agent's budget stop sentence to its bridge's source. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
| if any(is_judge(func) for func, *_ in self._funcs()): | ||
| context.record_model("judge", self.id, run_judge_model or self.judge_model) |
There was a problem hiding this comment.
When a judge function sets options["model"] and the run has no judge_model override, this records the step's judge_model instead. _payload keeps the function's model, so the run summary names a model that did not grade the task. This makes score differences harder to explain.
Record the model selected for the request, including function-level options.
Prompt To Fix With AI
This is a comment left during a code review.
Path: src/agent_env/task_step/task_steps/computer/osworld/evaluate.py
Line: 149-150
Comment:
**Summary names the wrong judge**
When a judge function sets `options["model"]` and the run has no `judge_model` override, this records the step's `judge_model` instead. `_payload` keeps the function's model, so the run summary names a model that did not grade the task. This makes score differences harder to explain.
Record the model selected for the request, including function-level options.
---
For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.…e used (FD-5492) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Please read this Claude Artifact https://claude.ai/artifact/JCkogrfdgTxvBhXLM2PoZS?sk=kv0m_DF3_tK47coA7YHyfQ
Note
Status after review: the controller these steps deploy is built from private scale-cua, so this PR doesn't merge to core until the controller and guest server are public (see Edgar's first comment). Meanwhile the same steps ship as an agentenvhub-sdk plugin for Scale, in scaleapi/scaleapi#175341, which unblocks the hub (scaleapi/scaleapi#174344). Both copies have the review fixes (20fdec6, ab7db64).
Summary
This PR adds three task steps to core:
osworld_deploy_controller,osworld_initializeandosworld_evaluate. It also adds theosworld-servercomputer app (all insrc/agent_env/task_step/task_steps/computer/osworld/, beside the other computer steps). Together they let an OSWorld task run on a frameworkcomputerenv. Both task kinds are supported:init_configplusevaluator.task_NNN.pythat definesTASK_CLASS.No plugin is needed.
Important
Merge order: scaleapi/scale-cua#360 must merge, and a controller image built from it must be published, before this PR. This PR refuses any controller that doesn't enforce
CONTROLLER_AUTH_TOKEN, and today's image doesn't.Design
The controller runs as a separate sidecar.
Only the framework may call the controller.
/setupor/evaluateitself.CONTROLLER_AUTH_TOKEN, and every call to the controller carries it as a bearer token. The controller answers 401 to any call without it (scale-cua#360).osworld-controller:<env_id>. It's derived again whenever it's needed and never stored, because context redaction strips stored secrets.The computer runs only the OSWorld guest server, as the
osworld-serverapp./osworld/<route>to127.0.0.1:5000.The guest server is found through the computer's card.
urn:agentenv:osworld-guest/v1at its forward prefix. AGETon that endpoint reports whether the guest server is up.osworld_deploy_controllerfinds the guest server by the extension's URI in the computer's card, and no/svc/<name>route is built anywhere.Step reference
osworld_deploy_controllerenv_id,controller_image_artifact_id(adocker_imageartifact; defaultcua-controller),ttl_seconds,cpu,memory_mb,env_vars.PROVIDER=remote,OS_TYPE=Ubuntu,REMOTE_VM_IP=<guest URL from the card>, andREMOTE_SERVER_HEADERS=<JSON of DeployedEnv.control_headers>;CONTROLLER_AUTH_TOKEN=<the derived token>;LITELLM_API_KEYandLITELLM_BASE_URLfrom[model], when they're configured.env_varsgo first, so the step's own values win.url_from_sandboxper link), andprepare_imageruns on that link beforecreate_container.GET /configureanswers (the controller has no health route), then checks that the controller requires the token.DeployedEnv.metadata):osworld_controller_url,osworld_controller_sandbox_idandosworld_guest_url.context.deployed_sandboxesasosworld-controller-<env_id>.osworld_initializeenv_id,init_config,osworld_v2_task_path, and the legacyosworld_v2_task_url(the path wins if both are set).POST /reset, thenPOST /initialize_task {id, osworld_v2_task_url}. It fails onsuccess: falseor a truthyerror.init_configtoPOST /setup, which serves every setup type. It fails onsuccess: falseor on any seeded site withok: false. Seeded sites are recorded incontext.metadata["osworld_seeded_sites"].osworld_evaluateenv_id,evaluator,verifier_id(default: a uuid hex),osworld_v2_task_path/osworld_v2_task_url, andjudge_model./evaluate {osworld_v2_task_url}./evaluate, one at a time, with itsresultandexpectedgetters and its options. Results are combined byconj. No grading runs in the worker.infeasibleis the one exception: it is graded on the agent's answer (the[INFEASIBLE]tag), ascua_evaluatedoes.rubricandparameters, plusmodel=judge_modelwhen the func's options set none. Deterministic funcs have anymodeloption removed.result,expected,options,rubric,parameters) whose length doesn't match the number of funcs is an error, not silently cut short./evaluate_async, then polling/evaluate_result. A controller without those routes (404) gets the blocking/evaluate. Every call carries the controller's bearer token.stage_files(fromstep_overrides) is downloaded through/setupbefore anything reads the computer. The step refuses to grade if staging fails.postconfig: runs once through/setupbefore grading. A failure is logged (error type plus at most 300 characters) and grading continues.error_typeis set is recorded as a skip with score 0.partial_credit(on the verification and incontext.metadata["partial_credit"][verifier_id]). The check matches the stop message of every agent that ships (Max turns (N) reached,Reached maximum number of turns (N), and the agent, CLI, Letta Code, hermes and opencode timeouts).http(s)URL is passed through. An object-store ref is signed for 6 hours, but only if it's in the object store and under<prefix>artifacts/; anything else is refused.Result shape (same as
cua_evaluate)context.metadata["verifications"][verifier_id] = {"results": [...], "score": float}results=[{"id": "osworld_v2", "score", "result": not error, "message"}].results=[{"func", "results": [{"id", "score", "result", "message"}]}], plusdebug(overall_score,conjunction,metrics,error,message) andpartial_creditwhen it applies.Staging the app
The guest server's source isn't vendored; it's staged under
payload/. The app's declaration passes the checks a put runs. Then add the app to the computer env'sapps. The app's requirements pinPillow==13.0.0; every 12.x release has known vulnerabilities.Dependencies
Controller (scaleapi/scale-cua#360, open): must
http(s)://REMOTE_VM_IPbase URL that includes a path;REMOTE_SERVER_HEADERSon every call to the guest server;CONTROLLER_AUTH_TOKENas a bearer token on every request.It must merge and its image must be published before this PR (see Merge order above).
collect_artifactson computer envs: feat(collect_artifacts): collect from framework computer envs #601 (separate PR, open).Known gaps
cua_evaluate's for the same task. These funcs are:compare_docx_files,compare_docx_tables,compare_pptx_files,compare_table,check_json_settings,check_json_keybindings,is_extension_installed,check_include_exclude,file_contains,run_sqlite3,check_thunderbird_prefsandcheck_thunderbird_filter.cua_evaluatereads (vm_file,cloud_file,rule,vm_command_line, the vscode getters) exists in its getters. So nothing becomes unservable.python3-pyatspi, which reaches the app's venv throughsys.path. This needs checking on a live computer.ubuntu-24.04..pytask, LLM-judge func or cloud sandbox provider has run end to end.Testing
tst/unit/task_step/osworld/(guest resolution, controller transport, the three steps, the app). All pass./evaluate_asyncpolling through transient errors, the 404 fallback, failed jobs, score normalization, signing refused outside the artifacts,require_authrefusing a controller that doesn't answer 401, and the token kept out ofrepr.osworld_deploy_controller: env vars (including the token), metadata, the sandbox record, no secrets in the logs, refusal before any sandbox is created, no double deploy, a controller that doesn't require the token is refused, chain links each get their own addresses,prepare_imagebeforecreate_container, a sandbox with a problem is let go, and every link failing.osworld_initialize: V2 and classic, including site failures and error text cut to 300 characters.osworld_evaluate: V2, a V2 controller error scoring 0, the error_type skip for V2 and classic,conjand/or across funcs, a single func's partial score, a failing func among others, infeasible, partial credit for each shipped agent's stop message (9 cases) and a message that isn't one, mismatched per-func list lengths, postconfig running once, andstage_files.ruff checkpasses.main; rerunning on c3a2f60. The rebase's one conflict wasrules.py, which keeps bothmain's browser steps and the threeosworld_*steps.test_docker_image_github_clone(a credential test), andtest_disk_recipe(Windows ISO boot) failed on macOS; neither touches this PR, and both pass in CI.Docs
The site should cover:
osworld-serverapp and theurn:agentenv:osworld-guest/v1extension;controller_image_artifact_id) and itsCONTROLLER_AUTH_TOKENrequirement.End-to-end test
tst/integration/computer/osworld_test.pyruns on local Docker:osworld-desk: ubuntu-24.04, the reference driver, theosworld-serverapp);init_configexecuteset up through the controller;check_include_excludeovervm_command_line, graded by the controller. A rule that matches scores 1.0 and one that doesn't scores 0.0.It passes against the token-checking controller image with Pillow 13 (132 s and 110 s on the latest two runs). It needs
OSWORLD_SERVER_SOURCEandOSWORLD_CONTROLLER_IMAGEand skips without them (theosworld_sourcescapability).The first live run found four bugs, fixed in a2b0fa7:
~/.Xauthority;scrot.The controller's address is also now saved in the run's context metadata, since a deployed env's record isn't re-saved after deploy.
Commits
osworld_sourcescapability.StepContextErrorfor context errors, bounded error text, and Pillow 13.prepare_imageon each link, and every shipped agent's stop message.task_step/task_steps/computer/osworld/, besidesnapshot_computer_env. No behavior change.and/or; everyagent_judge_*sent as a judge; a non-object answer scores 0; named timeouts; supervisord restarts the guest server.osworld_evaluate.osworld-serverapp) moves fromagent_env.osworldinto that same package, and the unit tests totst/unit/task_step/osworld/. No behavior change; the end-to-end test passes (136 s).🤖 Generated with Claude Code


Confidence Score: 5/5
The latest changes appear safe to merge; the previously reported wrong judge name remains a non-blocking issue.Fix with agent prompt
Summary
Adds OSWorld setup and grading on computer environments through a separate controller.
Diagram
%%{init: {'theme': 'neutral'}}%% flowchart LR Worker[Task steps] -->|Bearer token| Controller[OSWorld controller] Controller -->|Computer control token| App[osworld-server app] App --> Guest[Local guest server] Controller -->|Scores| WorkerReviews (14) · Last reviewed commit: "fix(osworld): narrow the prompt response..." · Reviewed by Greptile