Policy Catalog

Real-world policies, deterministically mapped

Each entry dissects a project's AI contribution stance and translates it into a machine-readable covenant.yml equivalent.

Filter by Stance

Comparison Matrix

Some projects in this matrix were sourced from open-source-ai-contribution-policies.

Project Stance Agent PRs Agent Reviews Issues Human Threads Provenance Attestation
Polars accepting allowallowallowwarn nonenone
Matplotlib accepting warnwarnallowdeny model, providernone
SciPy accepting allowwarnallowwarn model, providernone
Pandas accepting warnwarnallowdeny model, providernone
Django accepting warnwarnwarndeny strict (all fields)none
CPython accepting warnwarnwarndeny model, providernone
LLVM accepting allowallowallowwarn nonenone
Homebrew accepting warnallowwarnwarn model, providernone
Firefox accepting warnwarnallowwarn nonenone
NetBSD restricting denydenydenydeny strictrequired
Kornia restricting warnallowallowdeny strict (all fields)required
GNOME Extensions restricting denydenyallowdeny nonenone
Gentoo Linux rejecting denydenydenydeny n/an/a
Linux man-pages rejecting denydenydenydeny n/an/a
GoToSocial rejecting denydenydenydeny n/an/a

Example Policies

Complete covenant.yml files based on real-world stances.

Accepting (Polars-style)

spec_version: 1.0.0
defaults:
  unmatched: allow
rules:
  - id: agent-pr-open
    actor: agent
    action: pull_request.open
    outcome: allow
  - id: protect-human-threads
    actor: agent
    action: conversation.intervene_human_thread
    outcome: warn

Restricting (NetBSD-style)

spec_version: 1.0.0
defaults:
  unmatched: deny
actors:
  managers:
    - id: core-team
      match:
        usernames: [netbsd-core]
requirements:
  on_failure: deny
  default_provenance_profile: strict
  provenance_profiles:
    strict:
      required_fields: [model, provider, prompt_record, test_proof]
      on_failure: deny
rules:
  - id: manager-override
    actor: manager
    action: '*'
    outcome: allow
  - id: deny-agent-pr
    actor: agent
    action: pull_request.*
    outcome: deny
  - id: human-pr
    actor: human
    action: pull_request.open
    outcome: allow

Rejecting (Gentoo-style)

spec_version: 1.0.0
defaults:
  unmatched: deny
enforcement:
  deny:
    - type: comment
      message: This repository does not accept AI-generated contributions.
    - type: close_pull_request
rules:
  - id: human-all
    actor: human
    action: '*'
    outcome: allow
  - id: deny-agent-all
    actor: agent
    action: '*'
    outcome: deny

Submit a Policy

Know a project with a defined AI contribution stance? Submit it for inclusion using the Policy Submission issue template. Submitted policies are validated against the Covenant.yml v1.0.0 schema before acceptance.