trace-to-skill weekly downloads statistics

trace-to-skill

Codex Issue Radar and maintainer-readiness tooling for open-source projects using AI coding agents.

trace-to-skill turns live GitHub issue demand, failed Codex/agent runs, and repository readiness checks into evidence-backed reports, AGENTS.md rules, SKILL.md workflows, and CI gates.

Start Here

Run a live Codex issue radar:

npx trace-to-skill issue-map --repo openai/codex --state all --limit 100 --output codex-issue-radar.md

Install a weekly radar in any repository:

npx trace-to-skill init --issue-map-repo openai/codex --issue-map-state all --issue-map-limit 100
npx trace-to-skill init --issue-map-repo openai/codex --issue-map-comment-issue 8

Check whether a repo is Codex-ready:

npx trace-to-skill scorecard .

Package a failed run into a better OpenAI/Codex issue:

npx trace-to-skill codex-report ./runs --output openai-codex-issue.md

See the live generated example: docs/CODEX_ISSUE_RADAR.md, the recency-weighted heat map: docs/CODEX_ISSUE_HEAT.md, and the surface support matrix: docs/CODEX_SURFACE_MATRIX.md. For the full command catalog, use docs/USE_CASES.md.

AI coding agents are getting good enough to change real repositories, but they still repeat the same workflow mistakes: claiming success without tests, ignoring repo instructions, over-editing, inventing files, leaking secrets into traces, or enabling risky MCP tools.

trace-to-skill closes that loop:

failed agent run -> failure class -> reusable rule/skill -> eval gate -> keep or revise

It is built for maintainers using Codex, Claude Code, Cursor, Copilot coding agent, Gemini CLI, OpenCode, or MCP-enabled workflows.

Fast Use Cases

Use it when you need to:

• Gate Codex-ready PRs: run trace-to-skill scorecard . in CI and post a reviewer-friendly readiness comment.
• Try it before collecting traces: run trace-to-skill demo to generate a real Codex issue report from packaged public fixtures in one command.
• Prepare OpenAI OSS evidence: run trace-to-skill oss-brief . to generate application-ready proof, 500-character summary fields, readiness score, benchmark status, license, and next steps.
• Mine GitHub issue demand: run trace-to-skill issue-map --repo openai/codex or pass an exported issue JSON file to rank maintainer pain by deterministic failure class, comments, reactions, and evidence gaps.
• See what is heating up right now: run trace-to-skill issue-heat --repo openai/codex --state open --limit 100 --window-hours 24 to rank recent GitHub issue movement by recency, comments, labels, reactions, severity, and the first support command to run.
• Install a weekly issue radar: run trace-to-skill init --issue-map-repo owner/name to add a scheduled GitHub Action that turns the repo's hottest issues into a Codex failure-class report in the job summary.
• Harden agent instructions: run trace-to-skill lint-agents . to catch missing AGENTS.md, conflicting tool instructions, missing includes, nested instruction drift, encoding issues, and risky MCP config.
• Protect agent context: run trace-to-skill guard-github-event "$GITHUB_EVENT_PATH" before feeding issue, PR, comment, discussion, check-run, or commit text into an agent.
• Prevent unsafe patch overwrites: run trace-to-skill guard-patch ./change.patch --root . before applying generated patches so *** Add File cannot silently replace an existing file or symlink target.
• Audit local Codex session history: run trace-to-skill session-audit ~/.codex --format json to summarize rollout JSONL sizes, huge lines, parse errors, state files, short session_index.jsonl evidence, bloated transcript-like sidebar titles, subagent lifecycle signals, and recoverable unindexed thread ids without publishing private transcripts.
• Preflight sensitive paths before agent runs: run trace-to-skill sensitive-audit . --format json to find .env, private keys, package auth files, cloud credentials, local databases, signing files, and secret manifests by filename/path without reading file contents. The report now checks project-level .codexignore, .agentignore, .aiexclude, and .gitignore coverage for the recommended patterns; add --format ignore --ignore-target codexignore to generate a reviewable .codexignore candidate without mutating the repo.
• Preflight language-server readiness: run trace-to-skill lsp-audit . --format json to detect repo languages, missing LSP commands, install hints, and evidence files before asking Codex for symbol-aware edits.
• Audit Codex config drift: run trace-to-skill config-audit ~/.codex --format json to summarize legacy profile config, model pins, Speed/Fast service-tier persistence drift, sandbox/approval posture, Windows elevated sandbox mode, missing permission profiles, plugin cache drift, and MCP approval sprawl.
• Audit bundled plugin drift: run `trace-to-skill plugin-audit…