Apache Software Foundation
9 organization-specific tools, 16 skills, and the contributor growth and release management skill families.
Skills describe the work. Tools connect that work to your agent, services, and project conventions.
The workflows: what to check, prepare, and bring back for approval.
The interfaces skills need: 19 capability contracts, such as reading an issue or drafting mail, and 8 low-level capabilities.
The implementations for your tracker, model, or other service; 45 already implemented.
Skills ask for a capability rather than a named vendor. Tools implement each capability for a backend, so your project can change its tracker, mail, model or agent while keeping the workflow. Neutrality holds on six independent axes; a choice on one never constrains the others.
14 of 19 capability contracts (74%) have at least 2 backends or are neutral by construction. 85 of 85 skills are vendor-neutral, and 36 of 36 low-level tools work with any agent.
Any model, hosted or local.
Any agent that reads skills.
Any forge or issue tracker.
Mail, chat and forums.
Git, Subversion, Fossil and more.
Your project’s own rules.
| Contract | Backends | Coverage |
|---|---|---|
tracker | Atlassian, Forgejo / Gitea, Fossil, GitHub, GitLab, SourceHut | 6 backends |
source-control | Forgejo / Gitea, Fossil, Git, GitHub, GitLab, SourceHut, Subversion | 7 backends |
change-request | Atlassian, Forgejo / Gitea, GitHub, GitLab, email | 5 backends |
mail-archive | ASF, Discourse, Google, SourceHut | 4 backends |
chat | Discord, Slack | 2 backends |
mail-source | ASF, Google, Maildir | 3 backends |
mail-create | Google, Maildir | 2 backends |
cve-authority | CVE.org, Vulnogram | 2 backends |
report-relay | Any backend | Neutral by construction |
scan-format | Any backend | Neutral by construction |
project-metadata | ASF | Single organization (ASF) |
release-staging | Nexus Repository Manager (ASF-hosted) | 1 backend; needs another |
security-cross-ref | OSV.dev | 1 backend; needs another |
typed-decision | TypeSafe + self-hosted OpenAI-compatible | 1 backend; needs another |
people | Atlassian, GitHub | 2 backends |
security-tracker | ASF, Atlassian, agnostic | 3 backends |
body-field | Atlassian, GitHub, agnostic | 3 backends |
hook-manager | Any backend | 0 backend; needs another |
sbom-pipeline | Any backend | 0 backend; needs another |
Most tool choices are the same for every project under one governing organization: every ASF project allocates CVEs the same way, reads the same mail archive and gates on PMC membership. An organization profile holds those shared defaults once: its governance vocabulary, the backend each capability uses, and its identity. A project names its organization and inherits the rest; its own configuration still wins, then the organization, then the framework default. The same skill runs unchanged for an ASF project and an independent one; the organization carries the difference.
9 organization-specific tools, 16 skills, and the contributor growth and release management skill families.
The baseline for projects without a formal organization: no extra vocabulary, framework defaults throughout.
Any foundation, company or collective can add its own: contribute it to the framework so its projects and others reuse it, keep it in the organization’s own repository, or keep one locally in a project’s overrides.
The capability and tool definitions below come from the framework. Implementation status and supported vendors are available in the tools directory.
tracker | Issue / board / label backend. |
source-control | Branch / commit / diff / push (VCS). |
change-request | Proposed-change review + merge gate (pull request / merge request / Gerrit change). |
mail-archive | Mailing-list / forum archive reads. |
chat | Project chat reads — public channels only (Slack, Discord). Read-only; never posts. |
mail-source | Inbound-mail ingestion (mbox / IMAP / …). |
mail-create | Outbound mail composition. Always produces an editable draft; sending is a separate human-approved step on that draft (draft mode = default and the only mode implemented today; send mode declared but unimplemented — no autonomous send). |
cve-authority | CVE allocation / record management / publication. |
report-relay | Inbound security-report relay detection. |
scan-format | Security-scanner report parsing. |
project-metadata | Governance rosters / people / releases. |
release-staging | Release-staging repository reads (ASF Nexus at `repository.apache.org`). Read-only — never closes, drops, or promotes. |
security-cross-ref | Vulnerability database / cross-reference alias lookup (OSV.dev / NVD). |
typed-decision | Provider-agnostic typed decision backend (Choice / Score / Noul). |
people | Forge / tracker accounts: profile lookup, repository collaborators, team membership (one confirmed write). Distinct from `contract:project-metadata`, which reads an organization's governance rosters. |
security-tracker | Where the security skills keep embargoed reports: body fields, status rollup, attachments, a confidentiality guard, and a publish step for working-copy backends (GitHub, local files, private Jira). |
body-field | How the `### <Name>` fields of a tracker issue body are found, read and rewritten: field boundaries, closed code blocks, the refusals, and a language-neutral conformance suite every implementation passes. |
hook-manager | Git-hook managers (prek, pre-commit, lefthook, husky, …) under the secure agent setup: where a manager's install puts its hook, and how its home is split between sandboxed and host runs. |
sbom-pipeline | SBOM generation in a project's build: one CycloneDX 1.7 SBOM per distributable artifact, with the same meaning whatever ecosystem produced it, and content from other ecosystems embedded as assemblies. |
analytics | Read-only metrics / dashboards / renderers. |
sandbox | Agent isolation, egress control, settings audit. |
action-guard | Deterministic pre-tool-use command guards. |
privacy | PII redaction / approved-LLM gating. |
framework-dev | Build / validate / eval the framework itself. |
release | Release-artefact helpers an adopter's release process runs: reproducible-archive build, lint and comparison; maven-artifact-verify for staged JVM artefacts; release-config loading and pre-flight checks; release-verify checks of a staged release candidate; release-management rendering of the paste-ready release blocks; secret-prompt for the release manager's upload credentials. |
setup | Adopter-side setup state an agent resolves at runtime rather than at development time: lock parsing, floor comparison, reconciliation fingerprints. |
review | A second model's read of a change before it is published: other models' CLIs run read-only, findings merged. |