# Compact AI Coding Sessions

Long coding chats lose decisions, blockers, and file state. This skill creates concise handoff summaries for Claude, Codex, and Claude Code.

## Install

```bash
npx skillstore add zhanlincui/chat-compactor
```

## Metadata

- Status: approved
- Slug: zhanlincui-chat-compactor
- Skillstore revision: r1
- Version status: missing
- Tree hash: 3e87ee009e98a230e3b6182010fbe990a2d8a6297a0da221ff01c365e20aaf47
- Author: ZhanlinCui
- GitHub username: ZhanlinCui
- License: MIT
- Repository: https://github.com/ZhanlinCui/Ultimate-Agent-Skills-Collection/tree/main/chat-compactor
- Ref: 64ca8af0f54a325752f08bd54e52151061ea659a
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands
- Quality score: 76
- Public page: https://skillstore.pages.dev/skills/zhanlincui-chat-compactor
- Manifest: https://skillstore.pages.dev/api/skills/zhanlincui-chat-compactor/manifest

## Capabilities

- Generates a structured markdown handoff with context, accomplishments, current state, blockers, and next steps.
- Captures decision rationale and rejected alternatives for future agents.
- Documents failed approaches so later sessions avoid repeating them.
- Records modified files and relevant environment notes when available in the conversation.
- Recommends a predictable session note filename for saved summaries.
- Defines trigger phrases for wrapping up long or completed sessions.

## Use Cases

- End a Coding Session: Create a concise handoff before stopping work on an implementation or debugging task.
- Transfer Context Between Agents: Give a new Claude, Codex, or Claude Code session the decisions, files, blockers, and next actions.
- Record Debugging History: Capture failed approaches and current state so the team avoids repeating costly investigation paths.

## Prompt Templates

### Basic Wrap-Up

```
Use chat-compactor to summarize this session for the next AI agent. Include current state, completed work, blockers, and next steps.
```

### Focused Debugging Handoff

```
Use chat-compactor for this debugging session. Emphasize symptoms, hypotheses tested, failed approaches, useful files, and the next experiment.
```

### Implementation Handoff

```
Use chat-compactor to create a handoff for this feature. Include design decisions, modified files, validation status, risks, and prioritized follow-up work.
```

### Multi-Session Continuity Plan

```
Use chat-compactor to prepare a durable project handoff. Separate confirmed facts, assumptions, unresolved questions, environment notes, and actions for the next session.
```

## Limitations

- Summary quality depends on the available conversation history.
- It does not inspect the repository unless the agent uses tools separately.
- It can preserve sensitive details if users include them in the chat.
- It does not replace version control, issue tracking, or formal documentation.

## Best Practices

- Ask for compaction before the context window becomes difficult to review.
- Review the summary for secrets before saving it to a project directory.
- Include validation status and remaining uncertainty, not only completed work.

## Anti Patterns

- Do not use vague progress notes without file names, blockers, or next actions.
- Do not include secrets, credentials, private tokens, or unreleased customer data.
- Do not treat the handoff as a substitute for commits or issue updates.

## Security Audit

- Audited at: 2026-07-08T12:31:41.32\+00:00
- Summary: All four static command execution findings are false positives caused by markdown fences and inline filename examples in SKILL.md. No prompt injection or malicious command execution intent was found. A low risk privacy concern remains because saved summaries may preserve sensitive project context.

## Stats

- Views: 157
- Downloads: 62
- Favorites: 0
- Popularity score: 0