05-improve — Claude Code skill: what it does, how to install
05-improve examines conversation logs, evaluates visible token usage, and suggests targeted refinements without touching your code.
MIT ★ 571 updated 2026-10-29
/plugin marketplace add ai-driven-dev/framework
/plugin install aidd-refine@aidd-frameworkWhat 05-improve does
The skill processes a conversation ID or export file across three sequential stages in the actions/ directory. It inspects the chat record, generates advice, and identifies files for revision without asking for confirmation between steps.
All existing project files stay read-only. The skill writes its findings to a single temporary report and stops if the conversation data cannot be found.
It does not perform code delivery reviews or apply automatic edits. Its sole purpose is diagnosing conversational friction from recorded evidence.
What it helps with
- pinpoint why an agent conversation burned excessive tokens
- identify specific files that need revision after a prompt session
- cut wasted interactive time during complex tasks
Example prompts for the agent
- Review the conversation export.json and highlight wasted token spend
- Inspect session-42 to suggest minimal improvements for the next run
- Examine conversation session-id and recommend which files to edit
How to install 05-improve
1Plugin marketplace
/plugin marketplace add ai-driven-dev/framework/plugin install aidd-refine@aidd-frameworkThis installs the whole aidd-refine plugin with all its skills
2Skills CLI (Vercel)
npx skills add ai-driven-dev/framework --skill 05-improve -a claude-code3Manually: put the files in place
~/.claude/skills/05-improve/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/05-improve/
1Skills CLI (Vercel)
npx skills add ai-driven-dev/framework --skill 05-improve -a codex2Manually: put the files in place
.agents/skills/05-improve/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add ai-driven-dev/framework --skill 05-improve -a cursor2Manually: put the files in place
.cursor/skills/05-improve/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install ai-driven-dev/framework plugins/aidd-refine/skills/05-improve2Manually: put the files in place
.github/skills/05-improve/SKILL.md1Skills CLI (Vercel)
npx skills add ai-driven-dev/framework --skill 05-improve -a gemini-cli2Platform CLI
gemini skills install https://github.com/ai-driven-dev/framework --path plugins/aidd-refine/skills/05-improveManually: put the files in place
.devin/skills/05-improve/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add ai-driven-dev/framework --skill 05-improve -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about 05-improve
How to install 05-improve in Claude Code?
Run npx skills add ai-driven-dev/framework --skill 05-improve -a claude-code in your terminal. You can also install it via marketplace commands: run /plugin marketplace add ai-driven-dev/framework followed by /plugin install aidd-refine@aidd-framework.
What does 05-improve do?
It evaluates a session log or export file, audits visible costs, and produces minimal recommendations to make future chats more efficient.
Does 05-improve modify project files directly?
No. The skill keeps repository files read-only and writes only an isolated temporary report.
Source text
SKILL.md · MIT
--- name: 05-improve description: Reads complete conversation evidence, measures visible cost, and recommends minimal improvements. Use when the user wants to improve a conversation, reduce wasted time or tokens, or choose files to revise. Not for delivery review or automatic edits. argument-hint: conversation | export --- # Improve ```mermaid flowchart LR start([conversation ID or export]) --> read-conversation read-conversation -->|complete| recommend --> target-edits read-conversation -->|unavailable| unavailable([stop]) target-edits --> question([ask next intent]) --> stop([stop]) ``` ## Actions Run all three actions in order without confirmation. Read only the next file in `actions/`. | Order | Action | | --- | --- | | 1 | `01-read-conversation.md` | | 2 | `02-recommend.md` | | 3 | `03-target-edits.md` | ## Transversal rules - Access only the selected conversation and its relevant sources. - Use recorded evidence only; mark unavailable measurements as such. - Keep project files read-only; write only a unique temporary report.
© AI-Driven Dev · MIT · ai-driven-dev/framework