generated: '2026-08-11'
method: searched
source: https://infer.flow7.org/.well-known/agent-skills/index.json (HTTP 200, application/json)
description: >-
  Agent Skills PUBLISHED BY INFER and saved here verbatim — not generated by API Evangelist. Infer
  serves an Agent Skills 0.2.0 discovery index with a SHA-256 digest per artifact, which makes this
  one of the few providers in the catalog whose agent surface is both first-party and integrity-
  verifiable. All three skills share one unusual property: they instruct the agent to read live
  product state (catalog + status) before recommending anything, and forbid it from asking the user
  to paste an API key.
discovery:
  url: https://infer.flow7.org/.well-known/agent-skills/index.json
  schema: https://schemas.agentskills.io/discovery/0.2.0/schema.json
  status: 200
  saved: well-known/infer-by-flow7-agent-skills-index.json
skill_count: 3
skills:
  - name: infer-codex-provider
    type: skill-md
    file: skills/infer-by-flow7-infer-codex-provider.md
    source: https://infer.flow7.org/.well-known/agent-skills/infer-codex-provider/SKILL.md
    digest: sha256:ad221543d11a898bfe3f9a947f877565027bdd1406bd51d8b78963753a2b2031
    method: searched
    description: >-
      Configure or diagnose Infer by Flow7 as a Codex custom provider over the Responses wire API.
    api: Infer Responses API
    operations:
      - getPublicCatalog
      - getPublicStatus
      - createResponse
    grounded_in: >-
      /api/public/catalog and /api/public/status by URL; ~/.codex/config.toml provider block with
      base_url https://infer.flow7.org/v1, env_key INFER_API_KEY, wire_api "responses".
  - name: infer-opencode-provider
    type: skill-md
    file: skills/infer-by-flow7-infer-opencode-provider.md
    source: https://infer.flow7.org/.well-known/agent-skills/infer-opencode-provider/SKILL.md
    digest: sha256:acf5773eaa0abbc4231ab05dc2c2fb2154c55fdab9765310d01ce53c2dee0883
    method: searched
    description: >-
      Configure or review Infer by Flow7 as an OpenCode custom provider over the Responses API,
      against a versioned compatibility boundary (OpenCode 1.18.16).
    api: Infer Responses API
    operations:
      - getPublicCatalog
      - getPublicStatus
      - createResponse
  - name: codex-provider-doctor
    type: archive
    file: skills/infer-by-flow7-codex-provider-doctor.md
    file_note: >-
      SKILL.md extracted verbatim from the published archive. The archive also contains
      agents/openai.yaml and scripts/infer-codex-provider-doctor.py, which are not copied into this
      repo — the archive URL and its digest are recorded in packages/ instead.
    source: https://infer.flow7.org/assets/skills/codex-provider-doctor-0.1.0.tar.gz
    version: 0.1.0
    digest: sha256:2876c89162b2bfe278682bb7b1d3ed80ad63ea62202d5e90cc37dac51d5b11ba
    method: searched
    description: >-
      Provider-neutral. Diagnoses ANY OpenAI Responses-compatible endpoint with a local
      standard-library doctor — credential-free first pass, then optional paid probes gated behind a
      typed terminal confirmation.
    api: n/a (provider-neutral diagnostic)
    notable: >-
      Infer shipped a tool that works against its competitors as readily as against itself, and built
      it so no key ever leaves the user's machine.
generated_skills: []
generated_skills_note: >-
  API Evangelist authored NO skills for this provider. Infer already publishes its own, and every
  marquee flow the four-operation surface supports (check status, read prices, configure a client,
  send a bounded request) is covered by the three above. Adding a generated skill here would add
  noise, not coverage.