Doable

Get Code Context Connection

Check the Doable organization and recover an existing workspace binding. Call this before preparing or syncing the routing profile. It is the single authentication preflight for Code Context: the same MCP Authorization bearer token is used for every later workspace, Round, suite, TRD, case, and run operation. Args: local_workspace_id: Existing private local workspace UUID, when known round_code: Exact DQ-... request code when resuming a published Round

Sync Code Context Workspace

Upload one user-approved, locally validated sanitized routing profile. The profile must be produced by the Doable Code Context local helper. Do not construct it from raw repository data in the model or include paths, repository names, branches, commits, source, logs, URLs, or credentials. Args: workspace_ref: Opaque client workspace UUID from the local helper profile: Privacy-safe profile JSON emitted by the local helper

Start Code Context Round

Start one bounded coding-agent-origin context Round for a feature. Resolve the feature locally before calling this tool. Search/reuse or create the suite first, and pass only the exact developer request plus explicit supplemental questions. Doable may add focused questions without widening the named feature. Args: suite_id: Existing or newly created Doable feature suite feature_request: Exact bounded feature-testing request user_questions: Optional explicit developer questions

Get Code Context Round

Pull the current frozen Round for one DQ-... context connection. For TRD follow-up, the original code may resolve to a later published Round on the same TRD and workspace. This tool never lists or guesses unrelated rounds. Save its structured response privately, then let the bundled helper validate the connection and create the local answer candidate.

Submit Code Context Round

Submit one helper-validated privacy-safe Round answer payload. The MCP adapter derives a deterministic idempotency key. Never pass local evidence ledgers, paths, source code, repository identities, or secrets.

Finalize Code Context Round

Consume one ready coding-agent Round into TRD create or follow-up. Use mode=auto unless the developer explicitly selected create/follow_up. Retry is safe: the backend owns finalization idempotency and state.

Search Test Suites

Find the user's existing Doable feature suites before asking for an ID. Search uses suite name, deployed entry URL, and current TRD title/content. Reuse a unique result automatically, then call get_trd and list_test_cases before deciding whether the current assets cover the request. Ask the user to select only when the result is ambiguous or the scan is incomplete. Args: org_slug: Optional organization slug. Omit for an organization API key; Doable searches the organization bound t

Create Test Suite

Create a new Doable test suite. Duplicate names are allowed — each call produces a fresh suite_id. org_slug is optional; if omitted, Doable uses the API key's default org. Multi-org callers should pass it explicitly. Args: name: Human-readable suite name entry_url: Deployed app URL the suite will target by default org_slug: Doable organization slug (optional)

Update Test Suite

Update a test suite's entry_url and/or name. At least one of entry_url or name must be provided. Args: suite_id: Suite to update entry_url: New entry URL (optional) name: New suite name (optional)

List Test Cases

Read case inventory for coverage review, confirmation, and rerun selection. Returns each case's goal, preconditions, steps, expected results, execution flags, a stable inventory digest, and the suite's current TRD revision. The current backend does not persist exact case-to-TRD-snapshot lineage, so callers must compare coverage instead of assuming every case is current. Args: suite_id: Suite whose existing cases should be reviewed include_recent_attempts: Include compact metadata for u

Get Trd

Read the current TRD and its latest revision for a suite. Use this before inspecting or proposing a change to an existing feature. Set wait=true to long-poll until current TRD work finishes. This tool is read-only and intentionally omits internal session prompt history. Args: suite_id: Suite whose existing feature TRD should be read wait: Wait for current TRD work to finish before returning

Generate Test Cases

Generate cases from the current approved TRD revision without recreating it. Call get_trd first and bind this action to the displayed trd_id and latest snapshot. Existing cases are preserved by default. Set replace_existing_cases=true only after the user explicitly approves replacing the current case set. This tool never deletes or recreates the TRD. Args: suite_id: Suite containing the approved TRD expected_trd_id: TRD ID shown during approval expected_snapshot_id: Latest snapshot

Init Test Suite

Legacy clean-slate workflow: create a TRD, generate cases, and import them. Round-based coding-agent workflows should call generate_test_cases against the approved existing TRD revision instead. This tool's default clean path removes the current TRD and cases. Args: suite_id: Existing Doable suite (from create_test_suite) prd: Product Requirements Document with acceptance_criteria and user_flows clean_before_test: Delete existing cases and TRDs first (default True)

Run Test Cases

Run test cases in a suite. Omit test_case_ids to run every executable, non-exploratory case. Pass a subset to filter (e.g., to re-run just the failed cases after a fix). Exploratory and explicitly non-executable cases are always skipped — they don't count toward scheduling, completion, or results. If base_url is provided, the suite's entry_url is PATCHed to base_url before the run is scheduled — useful when the deployed app's URL has changed. Args: suite_id: Suite containing the cases

Get Test Run Status

Check progress of a test run. Returns status (pending/running/completed/failed) and completion percent. Poll until status is 'completed', then call get_test_run_results. Provide a job_id to track a specific run, OR a suite_id to check every executable, non-exploratory case (the default set run_test_cases would schedule). The returned job_id can be reused for stable polling. Pass exactly one. Args: job_id: Job ID from run_test_cases suite_id: Suite to check when no job_id is available

Get Test Run Results

Fetch results from a completed test run. Returns findings, test case results, and artifacts. Only call after get_test_run_status returns 'completed'. Use result_type="verify" for a compact pass/fail view without findings detail. Provide a job_id for a specific run, OR a suite_id to fetch results for every executable, non-exploratory case in the suite. Pass exactly one. Args: job_id: Job ID from a completed run suite_id: Suite to fetch when no job_id is available — covers all