Skills chat-compactor
📦

chat-compactor

Safe

Generate Session Summaries for AI Continuity

Lose critical context when sessions end. This skill produces structured handoff documents that preserve decisions, dead ends, and next steps for seamless AI agent continuity.

Supports: Claude Codex Code(CC)
🥉 74 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "chat-compactor". User says: 'Let's wrap up, compact this session'

Expected outcome:

  • Session saved to session-auth-refactor-2025-01-06.md
  • Summary includes: 3 key decisions with rationale, 2 dead ends documented, 5 modified files listed, 4 prioritized next steps

Using "chat-compactor". User says: 'Save context before we switch to the API task'

Expected outcome:

  • Checkpoint created: session-frontend-setup-2025-01-06.md
  • Captured: React component structure, Tailwind configuration, build command, and deployment notes

Security Audit

Safe
v1 • 2/24/2026

All static analysis findings are false positives. The 'external_commands' pattern matched markdown code fence backticks (```) not shell execution. The 'weak cryptographic algorithm' finding at line 3 incorrectly flagged YAML frontmatter text. This skill contains only instructional markdown content with no executable code, network access, or security risks.

1
Files scanned
100
Lines analyzed
0
findings
1
Total audits
No security issues found
Audited by: claude

Quality Score

38
Architecture
100
Maintainability
87
Content
50
Community
100
Security
91
Spec Compliance

What You Can Build

End-of-Session Handoff

Developer finishing a debugging session generates a summary so the next AI session can immediately continue without re-explaining the problem context and attempted solutions.

Long Project Continuity

Team lead maintains project momentum across multiple AI sessions by creating structured handoff documents that preserve architectural decisions and implementation status.

Context Window Management

User proactively compacts lengthy conversations before the context window fills, preserving essential information while freeing up tokens for continued work.

Try These Prompts

Basic Session Summary
Compact this session. Generate a structured summary with: what was accomplished, key decisions made, current state of the code, and next steps. Save as session-summary.md
Detailed Handoff Document
Create a comprehensive handoff document for the next AI agent. Include decision tables with alternatives rejected, all modified files with change descriptions, dead ends that should not be retried, and environment gotchas discovered.
Mid-Session Context Checkpoint
Generate a lightweight checkpoint summary before we continue. Capture current state, what's working, what's blocked, and immediate next steps. Keep it under 300 tokens.
Project Transition Summary
We're switching from feature A to feature B. Create a handoff document for feature A that captures: completed work, pending items, file locations, naming conventions established, and any quirks the next session should know.

Best Practices

  • Document the 'why' behind decisions, not just the 'what' - future agents need rationale
  • Always list failed approaches in dead ends section to prevent wasted effort
  • Keep summaries scannable with clear headings, tables, and bullet points

Avoid

  • Writing narrative prose instead of structured, scannable content
  • Creating vague summaries like 'made good progress' without specific outcomes
  • Omitting failure documentation - dead ends are the most valuable content

Frequently Asked Questions

When should I use this skill?
Use at session end, when the user says 'compact' or 'summarize', before context switches, when context window exceeds 50%, or before taking breaks on ongoing work.
Where are session files saved?
Sessions are saved to the project root or a designated sessions directory with naming format: session-[topic]-[date].md
What makes a good session summary?
Good summaries are scannable in under 30 seconds, actionable with clear next steps, defensive by documenting dead ends, and minimal with no unnecessary content.
Can I customize the output format?
Yes, the structured format is a template. You can request modifications like adding sections, changing table formats, or adjusting detail levels based on your needs.
Should I create a summary for every session?
Not necessarily. Use for meaningful work sessions where continuity matters. Quick Q&A or standalone tasks typically don't need session documentation.
How detailed should the dead ends section be?
Include enough detail to prevent retrying: what approach was tried, why it failed, and any error messages or symptoms. This prevents the most common AI re-exploration loops.

Developer Details

File structure

📄 SKILL.md