generated: '2026-08-14' method: searched source: https://help.spekit.com/hc/en-us/articles/31041891807643-Spekit-API-Overview sources: - https://help.spekit.com/hc/en-us/articles/31041891807643-Spekit-API-Overview - https://help.spekit.com/hc/en-us/articles/53991678945435-Spekit-MCP-Data-Privacy-Security-FAQs rate_limits: - surface: REST API base_url: https://api.spekit.co scope: per-account (the API token's organization) limit: 30 window: 10 seconds burst: >- Short bursts are tolerated; sustained requests above the threshold are rejected. Spekit's guidance is to implement exponential backoff. status_on_exhaustion: 429 status_name: Too Many Requests response_headers: [] response_headers_note: >- NOT DOCUMENTED and NOT OBSERVABLE. Spekit publishes no X-RateLimit-* / RateLimit-* / Retry-After header contract, and the OpenAPI declares only a 200 response per operation — there is no 429 response object in the spec at all. Every operation requires a token, so the limit could not be observed on a live unauthenticated response. An agent therefore has the documented number but no runtime signal: it cannot read remaining quota or a reset time, and must infer backoff from the bare 429 status. documented: true source: https://help.spekit.com/hc/en-us/articles/31041891807643-Spekit-API-Overview - surface: MCP connector endpoint: https://mcp.spekit.co/mcp scope: per-session limit: null window: null status_on_exhaustion: null documented: false documented_note: >- Spekit confirms the connector is rate limited "per session" with quota enforcement validated across service instances, but publishes no number, window or exhaustion behaviour. source: https://help.spekit.com/hc/en-us/articles/53991678945435-Spekit-MCP-Data-Privacy-Security-FAQs pagination_caps: - surface: REST API parameter: page_size default: 50 maximum: 100 note: Applies to all five collection operations; caps result volume independently of the request-rate limit. limit_count: 2 checked: '2026-08-14'