Audit developer experience across CLI, API, service, and GitHub Action surfaces with scored findings.
npx skills add https://github.com/SgtPooki/skills --skill dx-audit-sgtpooki
Please help me install this Agent Skill. Skill: dx-audit Source: https://github.com/SgtPooki/skills/tree/main/skills/dx-audit Command: npx skills add https://github.com/SgtPooki/skills --skill dx-audit-sgtpooki
na renormalization) and rolls up category, surface, and overall scores via a dependency-free Node scorer. - Two audit tiers: A static tier for docs/source inspection and a fixture tier that runs clean-room consumer installs for real TTFS, success-rate, and example-reliability metrics. - Use Case: Run /dx-audit in a repo you maintain to get a report.md, scorecard.json, and prioritized remediation plan written to a temp directory, with quick wins ranked by impact and effort. ## Quick Start Ask the agent to run a DX audit on the current repository, for example by saying "audit the developer experience of this repo across all detected surfaces and give me the scored report."High-intent search queries and answers about installing and using this skill.
Run the dx-audit skill against your repository; it auto-detects which surfaces exist (CLI, library/API, service, GitHub Action) and walks prioritized checklists for each. It produces a scored report, machine-readable scorecard, and remediation plan in a temp output directory.
The audit covers four surfaces: CLI, programmatic library/API, server/service, and GitHub Action. Each surface has its own P0/P1/P2 checklist grounded in sources like the CLI Guidelines, RFC 9457, OpenAPI, and GitHub's action authoring docs.
No. The default static tier inspects docs, source, and cheap probes like --help and exit codes without credentials. The fixture tier, which measures real TTFS and success rates, uses testnet or sandbox credentials when available and marks run-dependent metrics n/a otherwise.
Each check is scored 0, 1, 3, or 5, with 'na' for genuinely absent or out-of-scope items. A bundled Node scorer script rolls checks into category, surface, and overall scores using fixed weights, renormalizing around na entries so scoped-out work is never penalized.
No. All artifacts — report.md, scorecard.json, remediation-plan.md, and checks.json — are written to a temp directory keyed by repo name. Files are only written inside the repo if you explicitly request it and the path is already gitignored.
nihalnihalaniBuilds Jac projects into PyPI wheels and npm tarballs from jac.toml metadata.
rshadeScore technical debt across architecture, testing, documentation, and infrastructure.
innexar-platReview pull requests and diffs using mode-specific checklists and decision trees.
erikliebenAutomate persona-based usability testing across viewports and themes.