---
title: "Read one readiness run"
method: GET
path: "/projects/{projectId}/readiness-runs/{runId}"
tags: ["Directory readiness"]
---

# Read one readiness run

`GET /projects/{projectId}/readiness-runs/{runId}`

Lane statuses, coverage and the observation axis. The full report — every finding with its citation and provenance — is a separate fetch.

## Path parameters

- `projectId` string, required
- `runId` string, required

## Response `200`

The run.

- ReadinessRun — A directory-readiness run. Readiness carries no numeric score and is excluded from `pooledConformanceScore`: it answers whether a publisher would list this server, which is a different question from whether the server speaks MCP correctly.
  - `id` string, required
  - `readinessKind` 'claude' | 'openai', required
  - `serverId` string, nullable, required — The saved project server this run graded. Null only on rows written before the field existed.
  - `serverUrl` string, required — The saved server's URL exactly as stored. Canonicalizing it would change the grade.
  - `submissionMode` 'mcp-only' | 'mcp-imported-skills', nullable, required — The declared submission shape, for an OpenAI run. Never inferred.
  - `status` 'pending' | 'running' | 'completed' | 'failed' | 'cancelled', required — `cancelled` is distinct from `failed`: a person who stopped a run has learned nothing about their server.
  - `overallStatus` 'ready' | 'not-ready' | 'incomplete', nullable, required — The headline verdict. `incomplete` means nothing was violated and something the run needed was never evaluated.
  - `lanes` ReadinessLaneCoverage[], required
    - `lane` 'runtime-compatibility' | 'directory-policy' | 'optional-features' | 'submission-artifacts' | 'experience-insights' | 'plugin-package' | 'release-contract', required — Anthropic defines five lanes and OpenAI seven, overlapping on four. Which subset applies is decided by the run's `readinessKind`.
    - `status` 'ready' | 'not-ready' | 'incomplete', required
    - `evaluated` integer, required — Findings that reached a satisfied or violated verdict.
    - `notEvaluated` integer, required — Applicable but never exercised. Each one is an unanswered question.
    - `notApplicable` integer, required — Could not apply to this target; not a gap.
    - `missingInputs` string[], required — Named inputs the caller could supply to close the gap, for example `toolListing`. Empty when the gap is not the caller's to close.
  - `stages` ReadinessStageResult[], required
    - `stage` 'technical-preflight' | 'submission-ready', required
    - `status` 'ready' | 'not-ready' | 'incomplete', required
    - `lanes` string[], required — The lanes this stage rolled up, so the verdict is reproducible by hand.
  - `authMode` 'headless' | 'interactive' | 'provided-token', nullable, required
  - `capabilities` string[], required — What the executing surface could actually do. Two surfaces grading one target agree only on their shared capability subset.
  - `attemptCount` integer, required
  - `terminalReason` string, nullable, required
  - `errorMessage` string, nullable, required
  - `policySnapshotDate` string, nullable, required — The publisher-documentation snapshot this run graded against, so a stale grade says so.
  - `engineVersion` string, nullable, required
  - `sdkVersion` string, nullable, required
  - `includeLlmObservations` boolean, required — Whether this run was started with the paid observation opt-in.
  - `llmObservations` ReadinessObservationState, required — The model-observation axis, independent of the run's own status. A run whose lanes graded cleanly is `completed` even when the observation call was refused for credit: a payment problem belongs to the account, never to the server under grading.
    - `status` 'not-requested' | 'pending' | 'completed' | 'billing-blocked' | 'provider-failed' | 'invalid-output', required
    - `reason` 'not_requested' | 'billing_limit_reached' | 'provider_error' | 'provider_timeout' | 'schema_invalid' | 'no_evidence' | 'cancelled' — Machine-readable, so a client branches rather than string-matching. `billing_limit_reached` is the value to key a top-up prompt on.
    - `detail` string — A sentence for a human. Never parse it; branch on `reason`.
  - `hasReport` boolean, required
  - `reportUrl` string, nullable, required
  - `createdAt` integer, required
  - `updatedAt` integer, required

## Other responses

- `401` — Missing, invalid, revoked, or orphaned key (`UNAUTHORIZED`) — or the **target MCP server** needs an OAuth grant (`OAUTH_REQUIRED`), which is a property of the server, not your key.
- `403` — Key is valid but not allowed to do this.
- `404` — Unknown project, server, or resource.
- `429` — Per-key rate limit exceeded (60 requests/minute sustained, bursts up to 10). Honor `Retry-After` and back off with jitter.
- `500` — Something failed on MCPJam's side.

---

[API](https://skmtc.net/mcpjam/apis/mcpjam-api.md) · [All operations](https://skmtc.net/mcpjam/apis/mcpjam-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/mcpjam/mcpjam-api/revisions/8978a4e735fa/schema)