kevmoo/kevmoo_skills/sharpen-later
Claude Code and agent skill by kevmoo/kevmoo_skills
- community
Checking that the SKILL.md file is still in the repository.
Bookmarks agent workflow friction (a failing CLI, an ignored rule, a missing skill or hook) in one command, then resumes the active task in the same turn. Use when the user invokes /sharpen-later, says "log this for sharpen-saw", or wants to note a papercut without derailing current work. Don't use for multi-session transcript audits (use sharpen-saw) or when the user explicitly asks to edit a specific named rule or skill file right now.
Install
npx vskill@latest install kevmoo/kevmoo_skills/sharpen-laterSKILL.md
Sharpen-Later (/sharpen-later)
Capture a friction bookmark in one tool call and return to the user's task in the same turn. A dedicated /sharpen-saw session investigates it later.
The Same-Turn Contract
- Log it with one command.
scripts/later.pysits in this skill's
directory: replace <skill-dir> with that directory.
``bash python3 <skill-dir>/scripts/later.py "<what the user said went wrong>" \ --agent-note "<the failing command, tool or rule, in one line>" \ --cat <category> ``
--cat | Use for | |
|---|---|---|
cli-gap | A CLI that hung, prompted, or needed a flag you guessed. | |
rule-violation | An instruction that existed and was not followed. | |
skill-gap | A skill that was missing, or present and never loaded. | |
hook-gap | Something a hook or permission rule should have caught. | |
handoff-gap | A constraint, path, or tool bound left out of a subagent brief. | |
tool-bug | A harness tool that misbehaved. | |
ui-glitch | Output that rendered wrong for the user. | |
perf | Work that was needlessly slow or token-heavy. | |
review-escape | A real defect a human reviewer caught after local review | |
other | Anything else. |
- Echo the receipt. Put the printed
📌 Logged for /sharpen-saw (...)line
first in your reply.
- Resume. Continue the interrupted task in the same turn.
Guardrails
- Infer
--catand--agent-notefrom the conversation and proceed; the user
already said everything they wanted to say.
- Keep the bookmark detour to that single command. Multi-session root-cause
audits, transcript reading, and unsolicited edits to rules, skills or CLIs belong to the later /sharpen-saw session, which finds the surrounding steps from the bookmark's session and timestamp.
Details
- Bookmarks append to
${XDG_STATE_HOME:-~/.local/state}/sharpen-saw/later.jsonl (--later-file overrides).
- The session id comes from
$CLAUDE_CODE_SESSION_ID; under another agent, pass
--session <id>. A subagent logs against its parent session.
- Needs Python 3.9+ and nothing else.
Security
- Security scanPassed, 95 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 repositoryCheckingskills/sharpen-later/SKILL.md
Popularity
- 7-day trend score
- 5
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/kevmoo/kevmoo_skills/sharpen-later)