principle-encode-lessons-in-structure
Encode lessons in structure. Apply when writing the same instruction a second time, or noticing a correction that keeps recurring.
- License 许可
- MIT
- 最近更新
- 2026-11-15
- Stars
- ★ 2
- Forks
- 0
01 前言 Front Matter
- name
- principle-encode-lessons-in-structure
- license
- MIT
- metadata
- { "source": "pstack principle-encode-lessons-in-structure and the ladder, proof and rule table of pstack's correct skill, MIT, Copyright (c) 2026 Lauren Tan, github.com/cursor/plugins, adapted; checked against pstack 0.15.36 at commit 0bb6f37" }
- description
- Encode lessons in structure. Apply when writing the same instruction a second time, or noticing a correction that keeps recurring.
03 SKILL.md 正文
<!-- Generated by scripts/build.mjs from the sources; edit the source, then run npm run build. -->
Encode lessons in structure
Encode a recurring fix in a mechanism (a lint, a metadata flag, a runtime check or a script), not in more text. A written instruction needs the reader to notice, remember and comply; a mechanism does not.
When you write an instruction a second time, encode it and delete the instruction. Make a judgement rule prominent, with an example of the failure.
Pick the highest level that works: the architecture (one owner per state, one supported way, old ways deleted); a type that cannot hold the bad state; a check whose error names the fix; a test; a written rule last. Prove that each new check fails on a real past mistake. Agents copy nearby code, so a weak guard spreads.
Keep one rule table that pairs each standing rule with its mechanism. Drop a rule once its mistake cannot happen.
Close the loop. Capture and route every error, correction and surprise: a one-off correction gets a note, a recurring fix gets the highest level that works, and a systemic issue changes the vision or a flow. Apply it now or add a concrete task. Acknowledging without recording, recording without routing, and fixing one instance without the pattern change nothing.
04 同仓库更多 Skills
writing-standard
Write at 80% of Simplified Technical English. Apply to any text an agent writes for a person: messages, reports, specs, findings, reviews, plans.
sage
The state tool of sage mode: records tasks, routes, agent runs, findings, verdicts by head SHA and the user's gates in ~/.claude/sage/<project>/. Use in sage mode for every change of a task's state, and before any merge. Claude Code only.
sage-help
Help with sage: how to set it up, switch sage mode and autopilot, read the board, answer the chief's questions, pick a skill or a principle, fix a run that went wrong, and find where sage keeps its files. The user types /sage:sage-help (or /sage-help) with a question. Load it only when the user types it; do not load it on your own. Claude Code only.
report
The report that every sage agent ends with: STATUS, RESULT, EVIDENCE, FINDINGS, QUESTIONS, NOT VERIFIED, BRANCH. Use at the end of any task that the sage chief of staff gave you. Claude Code only.
remote-control
Set up Remote Control, so that the user can follow, steer and start Claude Code sessions from the Claude app on a phone. It turns Remote Control on for every new session, and it can run a server in a folder (such as ~/workspace) that starts at each login. Use when the user asks to reach sessions or start new ones from a phone, or to set up, check or undo Remote Control. Claude Code only.
readme-guide
Write a project README. Apply when you write, change or review a README of any project, in any language.