> ## Documentation Index > Fetch the complete documentation index at: https://docs.signalmd.dev/llms.txt > Use this file to discover all available pages before exploring further. # List improvement batches > When one run of several tests finds the same kind of gap in more than one of them, those tests improve together, and a gap they share is fixed once for all. Newest first, up to `limit` (one page). ## OpenAPI ````yaml /api-reference/openapi.json get /v1/agents/{agentId}/improvement-batches openapi: 3.1.0 info: title: SignalMD API version: 1.0.0 description: >- SignalMD reads a voice agent's production calls and says what to fix, then tests whether the fix worked. This API exposes the same data the web app shows: agents, calls and their traces, signals, issues, versions and their comparisons, tests, and test runs. Authenticate with `Authorization: Bearer sig_...` (an API key from the API keys page). Read keys may call every `read` operation; write keys may also call `write` operations, within their monthly budget. Each operation's scope is in `x-signal-scope`. Errors share one envelope (`Error`), lists share one cursor scheme (`limit`, `cursor`, `hasMore`, `nextCursor`), and every response carries `Request-Id` and `RateLimit-*` headers. servers: - url: https://api.signalmd.dev description: Production - url: http://localhost:43124 description: Local development security: - bearerAuth: [] tags: - name: Account description: Who the token belongs to and what it may do. - name: Agents description: Voice agents connected to the workspace. - name: Calls description: Production calls, their traces and per-call signals. - name: Signals description: What SignalMD noticed across calls. - name: Issues description: 'Problems worth fixing: where, why, and what to change.' - name: Versions description: Agent versions and how each one did. - name: Comparisons description: Stored comparisons between two versions. - name: Tests description: Tests, test sets, personas and coverage. - name: Runs description: Test runs and the attempts inside them. paths: /v1/agents/{agentId}/improvement-batches: get: tags: - Tests summary: List improvement batches description: >- When one run of several tests finds the same kind of gap in more than one of them, those tests improve together, and a gap they share is fixed once for all. Newest first, up to `limit` (one page). operationId: listImprovementBatches parameters: - schema: type: string format: uuid pattern: >- ^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}$ example: 3f1c2d4e-5a6b-4c7d-8e9f-0a1b2c3d4e5f description: The agent's id. required: true description: The agent's id. name: agentId in: path - schema: type: integer minimum: 1 maximum: 100 default: 25 description: Items per page, 1 to 100. example: 25 required: false description: Items per page, 1 to 100. name: limit in: query responses: '200': description: The newest improvement batches. content: application/json: schema: $ref: '#/components/schemas/ImprovementBatchList' '400': description: 'Invalid request: a parameter or the body failed validation.' content: application/json: schema: $ref: '#/components/schemas/Error' '401': description: Missing, invalid or revoked credentials. content: application/json: schema: $ref: '#/components/schemas/Error' '403': description: Not permitted for this principal. content: application/json: schema: $ref: '#/components/schemas/Error' '404': description: >- The agent, or a resource under it, does not exist or is not visible to this key. content: application/json: schema: $ref: '#/components/schemas/Error' '429': description: Rate limited. See Retry-After. content: application/json: schema: $ref: '#/components/schemas/Error' '500': description: Server error. The message is generic; quote the request id. content: application/json: schema: $ref: '#/components/schemas/Error' x-codeSamples: - lang: bash label: curl source: >- curl "https://api.signalmd.dev/v1/agents/6f1d2c3b-4a5e-4f60-8b7a-9c0d1e2f3a4b/improvement-batches" \ -H "Authorization: Bearer $SIGNAL_API_KEY" components: schemas: ImprovementBatchList: type: object properties: object: type: string enum: - list data: type: array items: $ref: '#/components/schemas/ImprovementBatch' hasMore: type: boolean nextCursor: type: - string - 'null' description: Pass as `cursor` for the next page; null on the last page. required: - object - data - hasMore - nextCursor Error: type: object properties: error: type: object properties: type: $ref: '#/components/schemas/ErrorType' code: type: string example: agent_not_found description: The specific reason; stable within v1. message: type: string example: Agent not found. param: type: - string - 'null' description: The request field at fault, for a validation error. requestId: type: - string - 'null' example: req_8c1f0e2b4a6d9e3f7a5c1b0d quota: type: object properties: metric: type: string example: credits description: >- credits when the workspace's monthly credits are used up, or agents when the plan's agent limit is reached. activity: type: string enum: - observation - testing - chat example: testing description: >- With metric credits: what the request would have spent them on. observation is analysing calls, testing is test runs, chat is the copilot. requested: type: integer example: 18 description: >- With metric credits: the credits the request needed to start. limit: type: integer description: 'The plan''s limit: the month''s credits, or its agents.' used: type: integer description: How much of the limit is used this month. plan: type: string example: free_org description: The workspace's plan slug. upgradePath: type: string example: /billing description: Where in the web app the plan is changed. required: - metric - limit - used - plan - upgradePath description: >- Present when code is quota_exceeded: the plan limit the request would pass. proposalId: type: string format: uuid description: >- Present when code is criteria_locked: the pending criteria proposal the change was filed as. A person approves or rejects it. required: - type - code - message - param - requestId required: - error example: error: type: not_found_error code: agent_not_found message: Agent not found. param: null requestId: req_8c1f0e2b4a6d9e3f7a5c1b0d ImprovementBatch: type: object properties: object: type: string enum: - improvement_batch id: type: string format: uuid pattern: >- ^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}$ example: 3f1c2d4e-5a6b-4c7d-8e9f-0a1b2c3d4e5f status: type: string enum: - running - paused - done - refused description: >- running: some test is still improving. paused: the rest settled and some test waits on the budget, a Retell key or credits; resume it from its page in the app once that is fixed. done: every test settled. refused: it did not start (see reason). reason: type: - string - 'null' description: Why it is paused or refused, in one line. spendCeilingUsd: type: - number - 'null' description: The most of the test budget this batch may spend. createdAt: type: string format: date-time example: '2026-10-02T14:03:07.120Z' updatedAt: type: string format: date-time example: '2026-10-02T14:03:07.120Z' totals: type: object properties: tests: type: integer ready: type: integer description: Passes as written, or reproduces the issue it was made for. improving: type: integer botProblem: type: integer description: The test is right and the agent fails it. needsYou: type: integer description: A person has to decide; each test's state says why. paused: type: integer stopped: type: integer notStarted: type: integer retellUsd: type: number description: Retell spend in this batch, counted against the test budget. required: - tests - ready - improving - botProblem - needsYou - paused - stopped - notStarted - retellUsd description: How the batch stands. tests: type: array items: type: object properties: testId: type: string format: uuid pattern: >- ^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}$ example: 3f1c2d4e-5a6b-4c7d-8e9f-0a1b2c3d4e5f title: type: - string - 'null' revision: type: integer description: >- The test's revision now. GET .../tests/{testId}/revisions has what each one changed. autoImprove: allOf: - $ref: '#/components/schemas/TestAutoImprove' - type: - object - 'null' description: >- Where the test stands. Null until auto-improve starts on it. retellUsd: type: number skipped: type: - string - 'null' description: Set when the test was not improved with the batch, with why. required: - testId - title - revision - autoImprove - retellUsd - skipped description: One entry per test, in the order they were approved. batchFixes: type: array items: type: object properties: tests: type: integer text: type: string required: - tests - text description: Gaps found on several tests at once and fixed once for all of them. required: - object - id - status - reason - spendCeilingUsd - createdAt - updatedAt - totals - tests - batchFixes description: Tests one run found at fault, improving together. ErrorType: type: string enum: - invalid_request_error - authentication_error - budget_error - permission_error - not_found_error - conflict_error - rate_limit_error - api_error TestAutoImprove: type: object properties: status: type: - string - 'null' enum: - improving - ready - reproduces - bot_problem - needs_you - paused - stopped - null description: >- Where auto-improve stands on this test. improving: runs or a fix are in progress. ready: the test passes as written. reproduces: an issue test that captures the issue. bot_problem: the test is right and the agent fails it, confirmed on Retell. needs_you: a person has to decide, including a possible problem in the agent seen in a text run and not confirmed on Retell (run the test on Retell to confirm it). paused: waiting on the budget, a Retell key or credits (the workspace has none left this month). stopped: a person edited or archived the test. Null when it never ran on this test. label: type: - string - 'null' description: The same in one line, as the app shows it. active: type: boolean description: Whether it is still working on the test. updatedAt: type: - string - 'null' format: date-time example: '2026-10-02T14:03:07.120Z' required: - status - label - active - updatedAt securitySchemes: bearerAuth: type: http scheme: bearer description: An API key (`sig_...`), or a session token from the web app. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.