penfield

What This Skill Does

Penfield is the persistent memory backbone for OpenClaw agents, designed to bridge the gap between ephemeral conversational sessions and long-term intelligence. By utilizing a hybrid storage architecture that combines BM25 keyword matching, vector embeddings, and a relational knowledge graph, Penfield enables agents to "remember" user preferences, historical decisions, and complex logical structures. This skill allows your agent to compound knowledge over time, ensuring that context isn't lost when a session ends or a task is handed off. It transforms your agent from a stateless chatbot into a reliable assistant that understands your workflow, technical constraints, and long-term project goals.

Installation

To integrate Penfield into your OpenClaw agent, execute the following command in your terminal:

clawhub install openclaw/skills/skills/dial481/penfield

Ensure your agent environment has access to the required local storage directories defined in your configuration.

Use Cases

Penfield is essential for high-fidelity agentic workflows. Use it to:

• Build Personalization Profiles: Save user preferences regarding coding style, response brevity, or specific project constraints so you don't have to repeat them.
• Manage Multi-Session Projects: Use penfield_save_context at the end of a workday and penfield_restore_context to pick up exactly where you left off, preserving variable states and conversational threads.
• Knowledge Graph Mapping: Use penfield_connect to link related findings, creating a web of data that allows the agent to reason across different technical domains.
• Long-term Research Tracking: Store artifacts and synthesis reports, then retrieve them later to avoid redundant analysis.

Example Prompts

1. "I'm going to start working on the authentication module. Please recall the security best practices we agreed upon in our last session regarding OAuth2 implementation."
2. "Reflect on our conversations from the past week. Are there any recurring bottlenecks or gaps in our project architecture that I should focus on fixing today?"
3. "Save this session as a checkpoint; we've finalized the API documentation and I want to be able to switch to the frontend task without losing this context."

Tips & Limitations

To get the most out of Penfield, quality of input is paramount. Always provide structured, descriptive memory entries. Avoid vague statements like "User likes X"; instead, document the "Why," the context, and the technical implementation details. Remember that while the knowledge graph is powerful for discovery, it requires intentional linking via penfield_connect to be effective. Keep your artifacts pruned using penfield_delete_artifact to maintain a performant and relevant memory database over the long term.