An automated session telemetry hook that captures turns, tool calls, and model costs into searchable Markdown reports.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install collect-session
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install collect-session using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is collect-session?
The collect-session skill is a powerful telemetry and archival tool designed for the OpenClaw ecosystem. It bridges the gap between ephemeral AI interactions and persistent knowledge management by automatically capturing session data whenever a user issues a /new or /reset command. By leveraging LiteLLM, it enriches raw logs with human-readable summaries and precise cost tracking.
This skill is essential for developers and teams who need to maintain a verifiable audit trail of agent work. It transforms messy session histories into a structured library of Markdown files, a machine-readable JSONL log, and a centralized index, ensuring that every insight generated by your AI agent is findable and accountable within Openclaw Skills.
collect-session Use Cases
- Archiving AI research sessions into a permanent personal knowledge base.
- Monitoring LLM API expenditures and token usage across different models.
- Generating automated documentation for agentic workflows and tool execution.
- Performing bulk backfills of historical session data for analytics.
- Troubleshooting agent behavior by reviewing detailed turn-by-turn telemetry.
How collect-session Works
- The user triggers a /new or /reset command within the OpenClaw interface.
- The collect-session hook intercepts the command and identifies the most recent session telemetry.
- The script communicates with a local LiteLLM instance to calculate costs and generate an LLM-powered session name and summary.
- A rich Markdown report is generated containing metadata, tool calls, and model usage statistics.
- The skill updates the global SESSION-INDEX.md and appends the entry to a structured session-log.jsonl file for future data analysis.
collect-session Setup
1. Script Deployment
Copy the core logic to your workspace and ensure execution permissions:
cp collect-session.mjs <workspace-dir>/scripts/collect-session.mjs
chmod +x <workspace-dir>/scripts/collect-session.mjs
2. Hook Configuration
Register the hook by creating the handler directory and the required metadata file at ~/.openclaw/hooks/collect-session/HOOK.md with the appropriate events (command:new, command:reset).
3. Enable in OpenClaw
Update your openclaw.json to activate the internal hook:
"hooks": {
"internal": {
"entries": {
"collect-session": {
"enabled": true
}
}
}
}
collect-session Data Schema & Taxonomy
The skill maintains a structured hierarchy within your designated memory directory to ensure compatibility with other Openclaw Skills:
| File / Path | Format | Purpose |
|---|---|---|
/sessions/YYYY-MM-DD-<slug>.md |
Markdown | Individual session reports with full telemetry and enrichment. |
SESSION-INDEX.md |
Markdown Table | A high-level overview of all recorded sessions for quick navigation. |
session-log.jsonl |
JSON Lines | Machine-readable records including cost, model, and turn counts. |
collect-session Advanced Features
- LLM-Powered Summarization: Uses Gemini Flash or other models via LiteLLM to create context-aware titles and summaries.
- Heuristic Fallbacks: Operates gracefully without an LLM by using metadata-based naming if the API is unavailable.
- Batch Processing: Supports a
--sweepflag to process and archive all historical, uncollected sessions at once. - Custom Cost Calculation: Includes a
MODEL_PRICINGtable to derive costs even when providers return zero-value telemetry. - Flexible Environment Support: Configurable via CLI flags, environment variables, or direct script edits for seamless CI/CD integration.
SKILL.md
Loading
Related Openclaw Skills
DevOps Platform Skill
A comprehensive interface for managing DevOps lifecycles, from application tracking to automated release window monitoring.
Official Account Content Expert
A professional AI content creation skill that crafts high-engagement articles for social media while eliminating robotic AI traces.
Excel/WPS Table Diagnosis
A professional diagnostic tool for AI agents to analyze spreadsheet structures and recommend stable, compatible formulas for Excel and WPS Office.
Know Your Owner
An automated data collection and analysis skill that builds a precise personal profile for your AI agent by indexing your logged-in social media activity.
Taobao E-commerce System (Dropshipping 2026)
A data-driven automation suite for Taobao dropshipping focusing on intelligent selection, marketing optimization, and order fulfillment.
Yabbie Net
A high-performance safety proxy that intercepts and validates AI tool calls in real-time to prevent malicious or accidental actions.
Featured*
