ccp_skill

Overview

This skill continuously records and compresses session events so conversation context survives crashes, restarts, and inter-session gaps. It appends only key decisions, discoveries, and essential context to a rolling session archive and automatically prunes old files based on a user-selected retention period.

How this skill works

At session start the skill creates a dated session file and, during heartbeats, appends a compressed summary of progress: topics, key decisions, and items to remember. It avoids raw transcript dumps by acting as a live compressor that keeps only high-value context. A weekly cleanup routine removes files older than the configured retention duration.

When to use it

• You need crash-resistant context continuity between agent sessions.
• You want a lightweight archive of recent session history without storing full transcripts.
• You need quick recovery of decisions and discovery points after a restart or error.
• You want to maintain short-term context while keeping long-term memory separate.

Best practices

• Choose a retention duration during initial setup (7, 10, 30, or 180 days) and confirm it before enabling the skill.
• Log concise, high-level summaries: topics covered, key decisions, and facts to remember — not raw chat logs.
• Run or schedule the cleanup routine weekly to enforce the retention window.
• Use session files as a short-term safety net and keep curated long-term facts in your primary memory system.
• Name and timestamp session files consistently so recovery and manual inspection are easy.

Example use cases

• A developer debugging a multi-step deployment who needs to resume after a crash with decisions preserved.
• A support agent who wants key troubleshooting steps and outcomes persisted between shifts.
• A research workflow that accumulates discoveries over several short sessions and needs a rolling history.
• An automation agent that must avoid losing configuration decisions when sessions restart unexpectedly.