Stats
Actions
Tags
From agent-caveman
Invoke before spawning subagents, or have subagents invoke before returning their final report. Forces caveman-style compact output for agent↔agent handoffs to cut tokens on inter-agent channel. Use when orchestrating multi-agent workflows, delegating research tasks, or when a subagent is about to summarize findings back to the caller.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agent-caveman:grunt-terseThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Goal: cut tokens on the inter-agent channel. Subagent prose is wasted on orchestrator — orchestrator wants facts, paths, line numbers, next step. Not narrative.
Goal: cut tokens on the inter-agent channel. Subagent prose is wasted on orchestrator — orchestrator wants facts, paths, line numbers, next step. Not narrative.
Structure final message as fragments. No preamble. No "I investigated...". No closing summary.
RESULT: <one line — done | partial | blocked>
FINDINGS:
- <fact with path:line when code>
- <fact>
FILES: <comma list of paths touched or relevant>
NEXT: <one line — what caller should do, or "none">
Omit any section with nothing to say. No empty headers.
[thing] [action] [reason] patternpath/to/file.py:42 so caller can click❌ Verbose:
I searched through the codebase and found that authentication is handled in src/auth/middleware.ts. The file defines a middleware function that validates JWT tokens using the jsonwebtoken library. It checks for token expiry and signature validity. There's also a helper in src/auth/utils.ts that handles token refresh. Let me know if you'd like me to investigate further!
✅ Terse:
RESULT: done FINDINGS:
- JWT validation:
src/auth/middleware.ts:14usesjsonwebtoken- expiry + sig checked at
:22-31- refresh helper:
src/auth/utils.ts:8NEXT: none
Brief subagent terse too:
npx claudepluginhub carlet0n/agent_caveman --plugin agent-cavemanGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.