An automated evidence preservation and trace management skill designed to capture the full lifecycle of OpenClaw agent activities.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-agentlog
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 openclaw-agentlog 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 OpenClaw Agent Log Skill?
The OpenClaw Agent Log Skill provides a unified framework for session management and activity tracing. It serves as an essential component for developers working with Openclaw Skills who require reliable record-keeping of agent behaviors, including reasoning processes from high-end models like DeepSeek-R1 and Claude. By leveraging OpenClaw Hooks, this skill automatically handles session creation, tool call logging, and response capturing without requiring manual session ID management.
This skill is particularly valuable in multi-agent environments where task handoffs (traces) are frequent. It ensures that every step of a workflow is documented and attributable, providing a clear audit trail for complex AI operations. It replaces the legacy agentlog-auto and openclaw-agent modules to offer a more streamlined and feature-rich experience.
OpenClaw Agent Log Skill Use Cases
- Automated evidence preservation for all OpenClaw agent session activities.
- Seamless task handoff management between different specialized agents.
- Transparent logging of model reasoning processes for debugging and quality assurance.
- Correlating agent actions with source code changes via Git commit binding.
How OpenClaw Agent Log Skill Works
- The agent initializes and executes the checkAndClaimTrace function to look for pending tasks in the workspace.
- Once a trace is claimed, the skill automatically sets the AGENTLOG_TRACE_ID and initializes a new session.
- During the agent's operation, hooks capture events such as session_start, before_tool_call, and after_tool_call in real-time.
- Detailed logs including tool arguments, execution results, and model reasoning are sent to the AgentLog Backend (MCP Server).
- Upon completion, the agent_end hook triggers intent archiving, and the session is closed, optionally binding the activity to the latest Git commits.
OpenClaw Agent Log Skill Setup
To get started with this skill, ensure your OpenClaw Gateway and AgentLog Backend (default port 7892) are running. Configure your environment variables as follows:
# Define backend and MCP server locations
export AGENTLOG_BACKEND_URL=http://localhost:7892
export AGENTLOG_MCP_URL=http://localhost:7892
export AGENTLOG_AGENT_ID=your-agent-identifier
For advanced configurations, update your YAML settings:
agentlog:
autoBindCommit: true
reasoningCapture: true
sessionTimeout: 600
OpenClaw Agent Log Skill Data Schema & Taxonomy
The skill organizes activity data through a structured pipeline that interfaces with a SQLite database via MCP. Below is the metadata taxonomy used:
| Data Type | Description | Capture Method |
|---|---|---|
| Session ID | Unique identifier for a single agent run | Automated Hook |
| Trace ID | Identifier for cross-session task tracking | Handoff API |
| Reasoning | Internal logic steps (e.g., DeepSeek thoughts) | reasoningCapture |
| Tool Calls | Arguments and results for all tool invocations | Tool Call Hooks |
| Git Commit | Binding to the latest commit in the repository | autoBindCommit |
OpenClaw Agent Log Skill Advanced Features
- Reasoning capture support for advanced LLMs like DeepSeek-R1 and Claude.
- Automatic session-to-commit binding to track AI changes in Git repositories.
- Robust trace handoff API for managing complex multi-agent workflows.
- Fully automated lifecycle management through integrated OpenClaw Hooks.
- Standardized MCP server integration for persistent storage and querying of agent history.
SKILL.md
Loading
Related Openclaw Skills
Self-Improving Agent
A universal self-improvement system that enables AI agents to learn from every interaction and autonomously evolve their codebase based on real-world experiences.
Broker Weekly Monitor
A professional-grade financial monitoring tool that automates the tracking of US retail brokerages, market-wide trading volumes, and the emerging crypto and prediction market ecosystems.
Atlas Travel Extras
A comprehensive personal travel assistant for planning trips, tracking budgets, and managing travel documents within a structured environment.
Agent Harness
A production-grade engineering harness that enforces execution discipline, product thinking, and knowledge compounding for AI agents.
LiteRAG
A local retrieval skill that utilizes SQLite for hybrid vector and keyword search across massive documentation libraries without bloating agent memory.
Search Agent
An intelligent AI research assistant that performs multi-source web searches and generates comprehensive, cited summaries.
Featured*
