generated: '2026-09-19'
method: searched
source: https://raw.githubusercontent.com/SongT-50/korean-news-mcp/master/mcpize.yaml
docs:
- https://github.com/SongT-50/korean-news-mcp#readme
- https://github.com/SongT-50/korean-public-data-mcp#readme
limit_count: 0
exhaustion_status: null
headers:
  on_exhaustion: []
  on_success: []
  note: >-
    No rate limit is documented for the A2A agents or for the directly hosted MCP endpoints, and none was
    observable — the A2A hosts answered two probes each without any RateLimit-*, X-RateLimit-* or Retry-After
    header, and the MCP hosts were suspended (503). Being JSON-RPC surfaces, neither would signal exhaustion with
    HTTP 429 semantics unless the platform in front did.
limits: []
marketplace_quotas:
  note: >-
    The provider's mcpize.yaml for the news server declares per-plan quotas for the MCPize-hosted listing only —
    Free 20 calls/day, Pro 500 calls/day, Enterprise unlimited (see plans/songt50-us-plans-pricing.yml). These
    are MCPize marketplace quotas and do not apply to the onrender.com endpoints or the A2A agents. The public-data
    mcpize.yaml declares pricing model free with no quota.
upstream_limits:
  note: >-
    The READMEs say the Korean government upstreams (data.go.kr, ECOS) are free with daily call limits that are
    "sufficient for personal use" but state no number; the servers mitigate them with in-process caches (TTLs
    recorded in conventions/songt50-us-conventions.yml). When self-hosting, the operator's own upstream key
    carries those limits.
probes:
- {url: https://publicdata-agent.songt50.us/, method: POST, status: 200, note: 'no rate-limit headers in the response (date, content-type, server, cf-ray only)'}
- {url: https://news-agent.songt50.us/, method: POST, status: 200, note: no rate-limit headers}