agentsclimarketplace

Compounding

Skill quanpersie2001/pulse/skills/compounding

A documentation-first skill plugin for Claude Code and Codex, built to take work from vague intent to validated, reviewed, and reusable delivery.

Install
npx -y skills add quanpersie2001/pulse --skill compounding

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use after completed Pulse work when the user asks to capture post-cycle machine-readable learnings for future planning and execution.

SKILL.md

2.2 KB, 388 tokens by cl100k_base, as published. Nobody here has run it

Compounding

If preflight readiness is missing, stale, or blocked (check .pulse/tooling-status.json), stop and invoke pulse:using-pulse before continuing.

Compounding turns completed feature work into reusable memory for future planning and execution.

Compounding is the canonical post-cycle, machine-readable learning pass for completed Pulse work. Use pulse:dream for runtime-artifact consolidation outside that post-cycle moment.

When to Use

  • after pulse:reviewing completes and feature is merged
  • after systematic bug-fix work reveals non-obvious root cause
  • after abandoned work that produced reusable lessons

Skip only when no durable/reusable learning emerged.

Runtime Contract

All operational rules live in references/runtime-appendix.md. Treat that file as canonical for:

  • 3-stream analysis (pattern/decision/failure)
  • synthesis quality bar (applicable-when must be specific)
  • propagation taxonomy and routing destinations
  • promotion rules for critical-patterns.md
  • durable memory capture behavior in .pulse/memory/*
  • state updates and handoff outputs

Minimum Flow

  1. Gather context from history artifacts, verification evidence, and bead graph.
  2. Run three analysis streams and collect outputs.
  3. Synthesize one learnings file at .pulse/memory/learnings/YYYYMMDD-<slug>.md.
  4. Classify each learning by propagation type and route to destination.
  5. Promote only truly global-critical learnings.
  6. Update .pulse/STATE.md and .pulse/state.json.

References

  • references/runtime-appendix.md — canonical runtime contract
  • references/learnings-template.md — learnings file structure
  • references/analysis-prompts.md — prompts for pattern/decision/failure analysis
  • references/corrections-and-ratchets.md — correction and ratchet file structures

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.