Skip to content

agent-skills.md

Latest commit

 

History

History
93 lines (61 loc) · 2.56 KB

agent-skills.md

File metadata and controls

93 lines (61 loc) · 2.56 KB
Raw

Agent skills — ponytail + caveman

How ponytail and caveman integrate with this platform.

Why both

Skill Shrinks Where it applies
Ponytail What agents build Edit prompts, code diffs, guardrailed file writes
Caveman What agents say Plan step strings, observations, chat-style output

They compose: ponytail picks <input type="date"> over a date-picker library; caveman writes "Use native date input" instead of a paragraph.

Cursor (local development)

File Purpose
.cursor/rules/ponytail.mdc Always-on YAGNI ladder
.cursor/rules/caveman.mdc Optional terse prose mode

Install / refresh:

chmod +x scripts/install-agent-skills.sh
./scripts/install-agent-skills.sh

Optional full caveman skill via skills CLI (installs to gitignored .cursor/skills/):

npx skills add JuliusBrussee/caveman -a cursor --yes

Temporal workers (autonomous agents)

Prompt injection lives in temporal/config/agent_skills.py and is appended to:

  • Plan nodes — DRAFT_PLAN_SYSTEM_PROMPT, REFINE_PLAN_SYSTEM_PROMPT (caveman)
  • Edit node — EDIT_SYSTEM_PROMPT (ponytail)
  • Observations — compact memory writes (caveman hint)

Environment toggles on the worker:

export AGENT_PONYTAIL=full   # default
export AGENT_CAVEMAN=lite    # default
export AGENT_PONYTAIL=off    # disable ponytail ladder
export AGENT_CAVEMAN=off     # disable terse plans

Multi-agent orchestration (planned)

When MissionWorkflow lands, assign skills by role:

MissionWorkflow
  ├── planner task   → caveman (board + task titles)
  ├── builder tasks  → ponytail (minimal edits per task)
  └── reviewer task  → ponytail-review pattern (delete-list)

Shared Postgres mission board holds progress; shared RAG holds observations tagged by mission/task. Target graph design: multi-agent.md.

Third-party layout

See third_party/README.md.

  • third_party/ponytail/ — optional git submodule (full upstream)
  • third_party/caveman-skill/SKILL.md — MIT skill text only

Not included: Caveman Proxy / Engine (BSL-1.1). Install separately if you want input-token compression:

npm install -g @caveman-ai/cli && caveman setup --install

License

  • Ponytail: MIT
  • Caveman skill: MIT
  • Caveman engine/proxy: BSL-1.1 — not vendored in this repo