openclaw_workspace
Summary
This workspace provides a modular environment for building, customizing, and extending autonomous agent workflows with OpenClaw. It integrates curated skills, configurable agent memory, operational tools, and documentation to streamline agent development and day-to-day operation—optimized for real-world productivity and safety.
Included Assets
-
Agent Configuration & Identity
AGENTS.md— Workspace overview and agent operating proceduresSOUL.md— Defines agent personality, tone, and principlesMEMORY.md— Durable, curated long-term memory for agent continuity- Daily/contextual logs in
memory/(e.g.,YYYY-MM-DD.md,learnings.md,subreddits.md)
-
Operational Documentation
TOOLS.md— Tech stack, installed skills, API credentials, workflow tips
-
Skills Directory
skills/agent-autonomy-kit/— Enables proactive, self-directed agent operationskills/humanizer/— Cleans AI-generated text for natural, human toneskills/llm-supervisor/— Handles seamless LLM switching and prevents silent model downgradesskills/mcp-skill/,skills/youtube/, and others — Extend agent integration and data access
-
Template & Sample Files
- Skill-specific READMEs, configuration files, and usage docs
How to Use
-
Workspace Orientation
- Read
AGENTS.mdfor daily operations and session routines. - Review
SOUL.mdto understand the agent’s expected tone and behavioral guidelines.
- Read
-
Agent Operation
- Maintain daily memory logs (
memory/YYYY-MM-DD.md) and updateMEMORY.mdfor curated facts and preferences. - Reference
TOOLS.mdfor environment setup, tech stack guidance, and API key management.
- Maintain daily memory logs (
-
Integrating Skills
- Each skill in
skills/includes its own README and setup instructions. - Popular integrations:
- Agent Autonomy Kit — Enables agents to act continuously and proactively without waiting for prompts.
- Humanizer — Processes and rewrites text to remove detectable AI writing patterns.
- LLM Supervisor — Monitors rate limits and switches between cloud/local LLMs transparently.
- YouTube — Fetches video details, transcripts, and runs channel searches.
- Each skill in
-
Session Management
- Before each session, check relevant memory files and preferences.
- In main sessions (direct user interaction), load
MEMORY.md; avoid loading it in group/shared contexts for privacy.
Notes
- This workspace is optimized for a specific user context (str4007) but can be adapted to other OpenClaw setups.
- Modular skill integration allows for rapid extension—review each skill’s documentation and requirements before enabling.
- Direct access to personal or sensitive data is strictly controlled; always follow outlined security notes in documentation.
- For best results, regularly maintain and prune memory files and update configurations as the operational environment evolves.
还没有评论。