geoff-ji/jiharn/documentation-and-decisions
Claude Code and agent skill by geoff-ji/jiharn
- community
Checking that the SKILL.md file is still in the repository.
Produce concise docs and Architectural Decision Records (ADRs) when a workflow or design becomes durable. Favours short artifacts over exhaustive manuals. Adapted from addyosmani's "documentation-and-adrs" and the skill-authoring doctrine.
Install
npx vskill@latest install geoff-ji/jiharn/documentation-and-decisionsSKILL.md
documentation-and-decisions
Trigger
Activate when:
- A recurring workflow is about to be repeated a third time — promote
it to a skill (via skill-authoring) OR to a doc.
- A design decision is made that future sessions would want to know
("we chose SQLite over Redis for this because...").
- The user explicitly asks for a README / ADR / decision record.
- A feature lands and the deployment behaviour would surprise a new
operator.
Principle
Short, lived docs > long, stale docs. Write the minimum that futureyou would thank present you for. Delete anything that's wrong.
Doc Shapes (pick one)
- README section — goal, usage, gotchas. Max ~60 lines.
- ADR (Architectural Decision Record) — Context, Decision,
Alternatives, Consequences. Max ~40 lines. File name docs/adr/YYYY-NN-<slug>.md.
- Runbook entry — "when X happens, do Y; if Y fails, Z".
Max ~30 lines. File name docs/runbooks/<area>.md.
- Skill docstring update — inline inside an existing
SKILL.md
when the doc IS the workflow.
Inputs
target: what we're documenting (feature, decision, workflow).audience: future-self, teammates, operator.shape: which of the four shapes above.
Steps
- Pick the shape from the list above — don't invent a new format.
- For ADRs, state what was considered and rejected. A one-line
"we chose X" is not an ADR.
- Name concrete commands / file paths / links. No "see the diagram"
without a diagram.
- Include a "revisit when..." trigger so the doc has a life-cycle
signal ("revisit when the SQLite row count > 100k").
- Write via
write_fileto the correct path;skill_managefor
embedded skill docs.
ADR Template
Verification
- The doc is under 100 lines unless it's a reference manual.
- Every code / command snippet actually runs on the project as of
today.
- The "revisit when" trigger is concrete, not "later".
- Links resolve (check via
read_urlif external).
Failure Signals
- "This document is out of date" comments added and left forever —
delete the stale section instead.
- ADR with no "Alternatives considered" — it's a status report, not a
decision record.
- The same decision keeps being re-discussed: the ADR was never read.
Link to it from the code / skill it governs.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingworkspace/skills/meta/documentation-and-decisions/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/geoff-ji/jiharn/documentation-and-decisions)