Skip to content

About

Reusable agent skills by Pierre Ortega

Resources

Stars

1 star

Watchers

0 watching

Forks

Latest commit

 

History

10 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Pierre Skills

Reusable agent skills by Pierre Ortega.

Skills

Agent Laboratory

Builds a self-verifying feedback loop before optimizing performance, quality, cost, correctness, or visual fidelity. It combines a laboratory for controlled experiments with a stopwatch for objective before-and-after measurement.

The skill includes:

  • A baseline-to-report experimental workflow.
  • Domain-specific measurement patterns and evidence rules.
  • A dependency-free Python ledger for recording and comparing trials.

See skills/agent-laboratory/SKILL.md.

Truthful TDD

Runs integration- and E2E-first red–green–refactor through broad vertical surfaces, with unit tests reserved for pure functions. Every bug gets regression coverage and every feature gets acceptance coverage, while the suite stays small by testing capabilities rather than incidental details.

See skills/truthful-tdd/SKILL.md.

Test Suite Auditor

Inventories every test suite in a repository, judges each suite and case for real defect prevention, independent oracles, redundancy, and predicted flake risk, then can consolidate or delete weak tests after replacement coverage is proven. A bundled manifest tool prevents sample-based audits from claiming completion while suites remain unreviewed. Existing CI jobs and Git history can add reachability, runtime, retry, churn, and minute-usage evidence; a JUnit normalizer combines machine-readable local or CI reports across repeated runs.

See skills/test-suite-auditor/SKILL.md.

Design

Designs, builds, reviews, and refines production web interfaces through ten directly routed craft playbooks. It covers visual direction, semantic color and theme systems, typography and copy, layout and surfaces, controls, responsive accessibility, data evidence, editorial pages, performance, and rendered QA. Direct sentence-case headings replace eyebrow/kicker layers, while behavioral evals and a bundled validator preserve the design rules over time.

See skills/design/SKILL.md.

Animation

Decides, specifies, builds, debugs, and reviews purposeful web motion. It combines frequency-aware motion decisions with easing and timing craft, interruptibility, spatial continuity, reduced-motion variants, frame-budget discipline, and real-interface observation.

See skills/animation/SKILL.md.

Shaders

Builds, integrates, debugs, tests, and optimizes WebGPU shaders with vgpu, typed WGSL modules, browser canvases, and headless Node rendering. Eight routed playbooks cover effects, geometry, compute, multipass textures, debugging, performance, setup, and product integration. Bundled tools inventory an existing vgpu project and enforce deterministic raw-pixel sanity gates.

See skills/shaders/SKILL.md.

Infrastructure Autopilot

Runs a persistent 15-minute infrastructure control loop with explicit merge and production authority. It discovers and builds monitors, responds to outages through bounded runbooks, improves performance, cost, SQL, capacity, and alert quality through measured experiments, and keeps a GitHub issue plus machine ledger as its audit trail. Twelve directly routed playbooks keep the large domain progressively disclosed, while bundled tools handle probe execution, loop state, slow-query ranking, cost-per-unit trends, alarm quality, audit rendering, and structural validation.

See skills/infrastructure-autopilot/SKILL.md.

Install

Copy the skill directory into the skills folder used by your agent:

git clone https://github.com/PierreAndreis/pierre-skills.git
cp -R pierre-skills/skills/agent-laboratory ~/.agents/skills/
cp -R pierre-skills/skills/truthful-tdd ~/.agents/skills/
cp -R pierre-skills/skills/test-suite-auditor ~/.agents/skills/
cp -R pierre-skills/skills/design ~/.agents/skills/
cp -R pierre-skills/skills/animation ~/.agents/skills/
cp -R pierre-skills/skills/shaders ~/.agents/skills/
cp -R pierre-skills/skills/infrastructure-autopilot ~/.agents/skills/

For Codex installations that use $CODEX_HOME/skills, copy it there instead:

cp -R pierre-skills/skills/agent-laboratory "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/truthful-tdd "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/test-suite-auditor "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/design "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/animation "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/shaders "$CODEX_HOME/skills/"
cp -R pierre-skills/skills/infrastructure-autopilot "$CODEX_HOME/skills/"

Restart the agent or begin a new session after installation so it refreshes its skill catalog.

License

MIT

About

Reusable agent skills by Pierre Ortega

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages