generated: '2026-10-03'
method: probed
status: published
source:
- https://qf-api.quietforge-studio.workers.dev/mcp
- https://qf-api.quietforge-studio.workers.dev/apis.json
- https://qf-api.quietforge-studio.workers.dev/.well-known/ard.json
summary: 'One first-party remote MCP server (streamable HTTP) on the API host. initialize and tools/list
  answered anonymously on 2026-10-03: 7 tools, which the server instructions describe as free and read-only
  (x402 directory search/detail/stats/demand, a single live x402 probe of a public URL, one sudoku puzzle,
  and a listing of the paid x402 routes). The paid HTTP routes are not exposed as MCP tools. Full tool
  list with inputSchemas saved verbatim in mcp/quietforge-tools.json.'
deployment:
  mode: remote
  endpoint: https://qf-api.quietforge-studio.workers.dev/mcp
  install: null
  package: null
  auth: none
  verified: probed
server:
  name: quietforge-x402-tools
  title: Quietforge x402 Tools
  version: 1.2.1
  transport: streamable-http
  url: https://qf-api.quietforge-studio.workers.dev/mcp
  protocol_version: '2025-06-18'
  websiteUrl: https://quietforge-studio.pages.dev/docs-api/
tools:
- name: x402_search
  description: Use this to find a pay-per-call x402 API for a given task, network or price cap before
    calling it (mostly USDC on Base; check each row's network and asset, some are testnet or other assets).
    Searches a health-probed directory of live x402 pay-per-call HTTP endpoints on Base, Solana and other
    networks (x402_index_stats gives the current count). Returns id, resource URL, method, network, USD
    price, description, 30-day call/payer counts and our own uptime/probe data. Data is re-crawled a few
    times a day, not real-time. Quietforge operates this directory and also operates some listed endpoints
    (source=quietforge) under the same per-host cap as every operator; ordering is by 30-day usage, then
    liveness, then freshness, never by operator.
  inputs:
  - q
  - network
  - alive
  - max_price_usd
  - source
  - limit
  - offset
- name: x402_service
  description: 'Use this to check whether one endpoint from x402_search is reliable enough to pay: its
    record plus its last 20 health probes (status code, latency, alive) and uptime across them.'
  inputs:
  - id
- name: x402_index_stats
  description: 'Use this to see how much this index covers (it is a subset of the x402 market, not all
    of it) and to sanity-check x402_search coverage before relying on it. Summary of the index: totals,
    how many endpoints answered on the last probe, counts per network and per source, and when the crawl
    ran.'
  inputs: []
- name: x402_demand
  description: 'Use this to answer "is anyone actually paying for x402 APIs today" with settled on-chain
    numbers instead of listings. Market-level demand signal for the x402 ecosystem, verified against on-chain
    USDC settlement on Base rather than self-reported call counts: how many seller addresses were paid
    in the last 24h, total USD settled, and the distribution of sellers by payment count. Free and aggregate-only;
    the per-seller breakdown (distinct payers per address; payer counts include sampler/grader wallets
    that pay many sellers once, so read them as an upper bound on customers) is the paid GET /v1/x402/demand
    route. Refuses to answer rather than serve a stale snapshot.'
  inputs: []
- name: x402_probe
  description: 'Use this to check a URL you are about to pay (or one you operate) for a well-formed x402
    challenge before sending funds; a challenge is no guarantee the service delivers. Makes ONE live unpaid
    request to a public http(s) URL and reports whether it is an x402 endpoint: status code, latency,
    alive flag, whether a PAYMENT-REQUIRED challenge came back and the parsed USD price. GET only, except
    that a URL already in the index is probed with its listed method. Public hosts only (redirects re-checked),
    delisted hosts never contacted, results cached 60 s, 3 probes per host per 5 min and 30 per minute
    overall.'
  inputs:
  - url
  - method
- name: sudoku_puzzle
  description: Use this when a user wants a sudoku to play or print, or an app needs a verified puzzle
    at a named difficulty. Generates ONE sudoku puzzle with a unique solution, 180-degree symmetry, graded
    by the solving techniques it needs (easy | medium | hard | expert). Returns 81-char puzzle and solution
    strings (row by row, 0 = empty), given count and technique list. Deterministic when a seed is supplied.
    20 per minute shared by all clients.
  inputs:
  - level
  - seed
- name: quietforge_pricing
  description: 'Price-check Quietforge''s paid x402 HTTP routes before paying (free read-only list, not
    tools here): convert docx, xlsx, pptx, csv to pdf; PDF text; URL/HTML to PNG; sudoku books; manuscript
    typesetting; crosswords; family trees from GEDCOM; dataset quality audits; x402 index API. USD prices,
    pay-to address, how an x402 client pays.'
  inputs: []
tool_quality:
  tool_count: 7
  tokens_total: 1909
  tokens_per_tool_median: 259
  heavy_tools: []
  heavy_tool_count: 0
  params_total: 12
  params_undescribed: 0
  params_described_pct: 100.0
  tools_with_undescribed_params: 0
  tools_annotated: 7
  annotated_pct: 100.0
  state_changing_unhinted: []
  tools_missing_schema: 0
  short_descriptions: 0
  duplicate_descriptions: 0
  duplicate_names: 0
  nonportable_names: []
  output_schema_count: 7
  token_estimate: chars/3.6 of each compact tool definition
  source: mcp/quietforge-tools.json
  measured: '2026-10-20'
  scored: false
  note: Facts from the saved tools/list capture (mcp_tool_quality.py); recorded, not scored. Checks learned
    from Glimind's agent-readiness audit.
tools_count: 7