autofix-skills¶
Claude Code plugin for the Jira autofix pipeline. Provides orchestrator skills for automated bug fixing, CVE remediation, ticket triage, and spike research. Designed to run inside a Claude Code container as part of a CI pipeline.
The plugin implements the inner layer of the autofix pipeline — orchestrator skills dispatch to sub-agents via prompt files and delegate deterministic work to Python scripts. The outer layer (ticket fetching, repo cloning, container launch, verdict reading) lives in separate repos (jira-autofix, ai-agentic-lib).
Plugin Details
- Version: 0.1.0
- Author: opendatahub-io
- License: Apache-2.0
- Category: Development Tools
- Repository: opendatahub-io/autofix-skills
- Tags: autofix jira cve bug-fixing triage pipeline ci-cd
Pipeline¶
Skills¶
| Skill | Description | Invocable |
|---|---|---|
/autofix-resolve |
Orchestrate end-to-end bug fixing via implement and review agent loop (max 3 iterations) | |
/autofix-cve-resolve |
CVE remediation across multiple repos with state-machine dispatch | |
/autofix-triage |
Assess bug tickets for AI autofix readiness (ready/needs_info/not_fixable) | |
/autofix-research |
Investigate spike tickets with no associated repository |
Installation¶
/plugin install autofix-skills@opendatahub-skills
Architecture¶
Four skills form two tiers: orchestrators (autofix-resolve, autofix-cve-resolve) and standalone assessors (autofix-triage, autofix-research).
autofix-resolve uses an implement → review → evaluate loop (max 3 iterations) with state.py for persistence across context compression. Extension skills are discovered via .autofix-context/config.json and called at post-implement and post-review hook points.
autofix-cve-resolve uses a Python state machine (cve_pipeline.py) for deterministic routing between phases: parse → resolve-repos → scan → fix → verify → VEX → review → create-PR → finalize. Each phase dispatches to a specialized agent prompt.
All skills write a structured verdict to autofix-output/.autofix-verdict.json and treat .autofix-context/ files as untrusted input.