Skills & Workflows

The Fedora Agent Skills Library is a collection of reusable AI skills designed for Fedora community workflows. Skills provide AI coding agents with specialized context, procedural instructions, and reference materials to assist with common tasks across the Fedora Project.

Overview

Agent Skills follow the open Agent Skills specification. Skills in this library are self-contained packages that community members can load into their own local AI agent environments, such as Claude Code, OpenClaw, or pi-coding-agent.

This project operates on an agentic-first development model:

  • AI coding agents assist contributors with routine, multi-step, or mechanical tasks.

  • Human contributors set intent, guide execution, review all generated outputs, and retain final accountability.

  • All skills in the library are licensed under CC BY-SA 4.0 or another Fedora Allowed License.

Available Skills

The library currently provides the following skills for Fedora contributor workflows:

fedora-compose-triage

A staged investigation workflow for examining FINISHED_INCOMPLETE and DOOMED Fedora composes using compose-tracker issues.

fedora-ftbfs-search

Automated searching in Bugzilla for existing Failed To Build From Source (FTBFS) and Failed To Install (FTI) bugs by package name.

meeting-secretary

Transforms Meetbot HTML log files into structured Markdown summaries for Fedora Discussion and ticket comments for Fedora Forge. Target migration name: fedora-meeting-followup.

packit-pull-from-upstream-onboarding

Sets up automated upstream release tracking in Fedora dist-git using Packit. Target migration name: packit-pull-from-upstream.

Contributing Skills

The Fedora AI/ML SIG welcomes community contributions to the skills library. Contributors can add new skills, improve existing instructions, or update reference materials.

To author or review a skill, consult the technical guides:

  • Creating Agent Skills: Covers skill directory anatomy, YAML frontmatter, audience and category tagging, and the canonical <domain>-<subject>[-<qualifier>]-<verb> naming schema.

  • Reviewing Agent Skills: Details the automated Skillsaw CI checks, human evaluation criteria (grounding, hallucination mitigation, safety), empirical agent testing, and merge workflow.

To propose a new skill:

  1. Search the skills library issue tracker to verify that the workflow is not already being developed.

  2. Open an issue describing the proposed skill, its intended user audience, and the workflow steps it will automate.

  3. Once discussed with maintainers, submit a pull request against the ai-ml/skills-library repository.

For broader pathways to get involved with the SIG, see Get Involved.

Maintainers

The Fedora Agent Skills Library is curated by community maintainers who steward the repository. Maintainers focus on reviewing community pull requests, verifying safety guardrails, maintaining instruction quality, and ensuring compliance with Fedora licensing and AI policies, rather than solely generating new content.

The current maintainers of the Fedora Agent Skills Library, in alphabetical order:

Communication