slack-context-memory

What This Skill Does

The slack-context-memory skill is an advanced context-optimization engine designed specifically for OpenClaw AI agents operating within Slack-heavy work environments. It addresses the 'context window exhaustion' problem common in long-running AI sessions by intelligently distilling massive Slack message archives into structured, high-value data points. By identifying thematic conversation boundaries, the skill generates summaries comprising TL;DRs, specific decision logs, and topic classifications, while simultaneously creating vector embeddings for future semantic retrieval. This allows your agent to maintain a persistent, "long-term memory" of channel history without the prohibitive cost or token limit penalties of re-processing raw message logs.

Installation

To integrate this skill into your environment, ensure you have Node.js 18+ and a PostgreSQL database with the pgvector extension enabled. Run the following command in your terminal:

clawhub install openclaw/skills/skills/davidrudduck/slack-context-memory

Once installed, navigate to the skill directory at /home/david/clawd/scripts/slack-context-memory and execute node setup-schema.js to initialize the necessary database structure. Ensure your OpenClaw agent has read access to your Slack data dump or API exports for the skill to perform its compaction processes.

Use Cases

This skill is perfect for teams looking to centralize organizational knowledge found in chat. Primary use cases include:

• Project Onboarding: Summarize months of channel discussion to bring a new team member up to speed instantly.
• Decision Auditing: Quickly query the AI to find exactly when and why a specific architectural or policy decision was made.
• Long-term Support: Maintain context on troubleshooting tickets discussed across Slack threads without inflating prompt tokens.
• Meeting Prep: Summarize relevant project channels before a meeting to ensure the AI has the latest context.

Example Prompts

1. "OpenClaw, pull the recent summary from the #engineering-updates channel to see what was decided about the API migration."
2. "Search through past Slack conversations to find any discussions regarding the email newsletter filtering bugs reported last month."
3. "Summarize the last 50 messages in #marketing and identify any open action items that need follow-up."

Tips & Limitations

To maximize effectiveness, run the detection and compaction scripts periodically via a cron job or scheduled task, as the agent only knows what has been processed. Note that while compaction saves massive amounts of tokens, it is lossy by design; deep technical debugging might still require accessing raw logs if the summary omits specific minor details. Ensure your vector dimensions match the 1024-dim configuration required by the schema for optimal search performance.