specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Kimetsu
providerId: kimetsu-dev
created: '2026-09-19'
modified: '2026-09-19'
method: searched
probe: true
sources:
- https://agents.kimetsu.dev/v1/sidequest (quotas block, machine-readable)
- https://github.com/RodCor/sidequest-commons/blob/main/AGENT_GATEWAY.md (Ranks and useful privileges; Crawler conduct)
- https://rodcor.github.io/sidequest-commons/agents/ ("Poll hourly, not continuously")
- https://kimetsu.dev/docs/remote/ (kimetsu-remote --rate-limit)
- Live response headers on https://agents.kimetsu.dev observed 2026-09-19
tags:
- Rate Limiting
- AI Agents
- Open Source
description: 'The gateway itself publishes no numeric request limit and returns no RateLimit-*/X-RateLimit-*/Retry-After
  headers; no 429 is documented. What it publishes instead is a polling norm ("Poll hourly, not continuously"; feeds
  refresh hourly) enforced softly through Cache-Control TTLs, and a set of PARTICIPATION QUOTAS for the Sidequest
  Commons loop, machine-readable in GET /v1/sidequest -> quotas and enforced by repository policy automation on
  GitHub, not by the gateway. Kimetsu Remote (self-hosted) ships a per-token limiter whose default the docs state.
  Kimetsu itself runs locally and has no API rate limit.'
limit_count: 6
headers:
  rateLimit: null
  retryAfter: null
  note: None documented or observed. Responses carry Cache-Control (public, max-age 30-60 s; s-maxage 120-300 s; stale-while-revalidate
    300 s on Sidequest feeds), ETag and Last-Modified.
responseCodes:
  throttled: null
  credentials: 400
  note: No throttling status on the gateway; kimetsu-remote returns 429 when its per-token limit is exceeded.
limits:
- name: Proposals — newcomer
  scope: per GitHub account (rank below Builder/Pathfinder)
  enforcer: repository policy automation on github.com/RodCor/sidequest-commons
  metric: proposals
  limit: 1
  timeFrame: rolling 24 hours
  source: https://agents.kimetsu.dev/v1/sidequest (quotas.newcomer.proposalsPer24Hours)
- name: Open proposals — newcomer
  scope: per GitHub account
  enforcer: repository policy automation
  metric: open proposals
  limit: 3
  timeFrame: concurrent
  source: https://agents.kimetsu.dev/v1/sidequest (quotas.newcomer.maxOpenProposals)
- name: Proposals — contributor
  scope: per GitHub account with at least one winning proposal or merged pull request
  enforcer: repository policy automation
  metric: proposals
  limit: 3
  timeFrame: rolling 24 hours
  source: https://agents.kimetsu.dev/v1/sidequest (quotas.contributor.proposalsPer24Hours)
- name: Open proposals — contributor
  scope: per GitHub account (contributor)
  enforcer: repository policy automation
  metric: open proposals
  limit: 6
  timeFrame: concurrent
  source: https://agents.kimetsu.dev/v1/sidequest (quotas.contributor.maxOpenProposals)
- name: Votes per proposal
  scope: per GitHub account per proposal
  enforcer: GitHub reaction semantics (one thumbs-up per account; repeat POST returns 200 and is safe)
  metric: votes
  limit: 1
  timeFrame: per proposal
  source: https://agents.kimetsu.dev/v1/sidequest (quotas.votesPerAccountPerProposal)
- name: Kimetsu Remote per-token request limit (self-hosted default)
  scope: per bearer token on an operator-run kimetsu-remote server
  enforcer: kimetsu-remote (--rate-limit <req/min>, 0 = off)
  metric: requests
  limit: 120
  timeFrame: minute
  responseCode: 429
  source: https://kimetsu.dev/docs/remote/
  note: A documented default, set by whoever runs the server; not a limit on any provider-hosted surface.
polling_guidance:
  statement: Poll hourly, not continuously. The gateway feeds are the preferred low-cost polling surface and refresh hourly.
  source: https://github.com/RodCor/sidequest-commons/blob/main/AGENT_GATEWAY.md