Skip to content

README.md

Latest commit

 

History

History
57 lines (40 loc) · 5.9 KB

README.md

File metadata and controls

57 lines (40 loc) · 5.9 KB
Raw

Agent Skills

This repository is a home for many named agent skills.

Each skill lives in its own flat directory under skills/ and should include a SKILL.md file plus any optional agents/, scripts/, references/, or assets/ needed to make the skill reliable and reusable.

Skills are kept flat: a skill never contains another skill's SKILL.md nested inside it. Agents such as Claude Code load skills from a flat skills directory and do not support nested skills, so a related set of actions (for example the labnb-* family) ships as several top-level skill directories rather than one skill with sub-skill folders. This keeps the whole skills/ tree installable into Claude Code, Codex, and other agents.

Skills

The list below is kept in alphabetical order by skill name.

Skill Description
duct Wrap any command with con/duct to capture wall-clock time, CPU, and memory usage as structured logs, so agents and reviewers can inspect what a run actually consumed.
kya Govern and review agents with veldt-kya (Know Your Agents) — risk-score, consensus-judge, and drift-check an agent, emit compliance evidence, and write a governance verdict that the labnb loop can break on.
labnb Create and maintain a concurrency-safe global lab notebook outside project roots, with startup summaries of related prior work, first-class idea capture and promotion, isolated experiment workspaces, focused companion skills, and append-only indexing across projects, investigations, and tasks.
labnb-idea Record a promising but not-yet-implemented experiment idea in the shared lab notebook index.
labnb-promote Promote a lab notebook idea into a concrete experiment with explicit budgets, source links, and provenance.
labnb-resume Summarize prior ideas and experiments for a project slug, then choose whether to resume, promote, branch, or start new work.
labnb-run Create and run a concrete lab notebook experiment with isolated workspace, explicit budgets, and iterative logging.
orcd Use MIT ORCD (Engaging) as a remote execution environment: set up key-based SSH through the OnDemand portal, discover the Slurm partitions, GPU models, and storage tiers the current user is actually entitled to, place job IO on the fast bcs flash scratch, and submit and track work.
structsense-skills StructSense Skills transforms unstructured text and PDFs into validated, ontology-grounded structured JSON using a model-agnostic extraction pipeline.

Installing Into An Agent

Use scripts/install_skills.py to copy the flat skill directories into an agent's skills directory:

python scripts/install_skills.py --list                  # show available skills
python scripts/install_skills.py --agent claude           # -> ~/.claude/skills
python scripts/install_skills.py --agent codex            # -> ~/.agents/skills
python scripts/install_skills.py --agent codex --scope project   # -> ./.agents/skills
python scripts/install_skills.py --agent codex --skills labnb duct
python scripts/install_skills.py --dest /path/to/skills   # any other agent

Skill directories follow the cross-agent Agent Skills format: a flat directory with a SKILL.md (with name and description frontmatter). The optional agents/openai.yaml adds Codex-specific UI metadata and invocation policy and is ignored by agents that do not use it. Default install locations:

  • Claude Code: ~/.claude/skills/ (user) or <project>/.claude/skills/ (project).
  • Codex: ~/.agents/skills/ (user) or <repo>/.agents/skills/ (project), the Agent Skills open-standard location.

The installer refuses to install a skill that contains a nested SKILL.md, which guards against re-introducing the structure that some agents cannot load.

Continuous Integration

.github/workflows/ci.yml runs on every push and pull request, with no human interaction or API keys required:

  • Unit tests: uv run python -m unittest discover -s tests.
  • Validate skill format: scripts/validate_skills.py checks every skill against the Agent Skills format (flat directory, required name/description frontmatter, matching name, no nested skills, parseable agents/openai.yaml, and that referenced scripts exist) and dry-runs the installer.
  • duct skill smoke test: installs con-duct, captures a real command with duct, and summarizes the run with the skill's helper.

Utilities

  • scripts/install_skills.py: install the flat skills in this repository into an AI coding agent's skills directory (--agent claude/codex, --scope user/project, or an explicit --dest), validating that no skill contains nested skills.
  • scripts/validate_skills.py: deterministically validate that every skill is loadable by SKILL.md-based agents; used by CI.
  • scripts/launch_agent_container.py: launch codex or claude inside a tightly-scoped Docker or Apptainer container using a reusable TOML config, including explicit auth mounts when credentials live outside the main agent state directory. See docs/agent-container-launcher.md.