agent-context-system
A persistent local-only memory system for AI coding agents. Two files, one idea — AGENTS.md (committed, shared) + .agents.local.md (gitignored, personal). Agents read both at session start, update the scratchpad at session end, and promote stable patterns over time. Works across Claude Code, Cursor, Copilot, Windsurf. Subagent-ready. No plugins, no infrastructure, no background processes.
Why use this skill?
Stop the cold start problem. Use agent-context-system to give your AI coding agent persistent, local memory across sessions with shared and private context files.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/andreagriffiths11/agent-contextWhat This Skill Does
The agent-context-system is a lightweight, persistent memory solution designed for AI coding agents. It solves the 'cold start' problem by implementing a dual-file context architecture: AGENTS.md for shared, long-term project knowledge, and .agents.local.md for personal, evolving scratchpad notes. By leveraging these files, agents can maintain continuity across sessions, remembering project-specific patterns, architecture, and previous debugging efforts without requiring complex infrastructure, plugins, or background processes. It acts as a standardized interface across various environments like Claude Code, Cursor, Windsurf, and GitHub Copilot.
Installation
To install this skill, use the ClawHub CLI tool in your terminal:
clawhub install openclaw/skills/skills/andreagriffiths11/agent-context
After installation, execute the setup script to initialize your configuration files:
./scripts/init-agent-context.sh
This script will create your .agents.local.md file, ensure it is ignored by git, and symlink it to your agent’s specific configuration rules file (e.g., .cursorrules, .windsurfrules).
Use Cases
This skill is perfect for complex, long-running projects where maintaining context is difficult. Use it to store API integration nuances, persistent architectural decisions, and specific coding conventions that your team or individual development workflow requires. It is especially useful for developers who switch between multiple AI tools, as it provides a unified source of truth that every agent can read and understand.
Example Prompts
- "Analyze our current progress in the session log within .agents.local.md and propose a plan for today's work based on the 'Ready to Promote' patterns."
- "Review the notes from our last three sessions in .agents.local.md and compress them, moving any stable patterns into AGENTS.md."
- "Please update the session log in .agents.local.md with a summary of the authentication refactor we just completed, noting the specific gotchas we encountered with the OAuth flow."
Tips & Limitations
To get the most out of the system, keep AGENTS.md concise—ideally under 120 lines—to avoid token bloat. The system relies on the agent's ability to follow instructions; therefore, maintain clear, descriptive language in your logs. Remember that the system is local-only; files are not synced to the cloud automatically unless your git workflow includes them. If you work in a team, encourage colleagues to maintain AGENTS.md as the primary source of truth, ensuring the shared knowledge base evolves alongside the codebase. Always perform periodic cleanup of the scratchpad to prevent it from becoming disorganized.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-andreagriffiths11-agent-context": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-write, file-read
Related Skills
claw-relay-openclaw
Control a remote browser through Claw Relay using the CLI client. Use when you need to navigate authenticated websites, click buttons, fill forms, take screenshots, or read page content on a user's real browser — and your platform doesn't support MCP. Triggers on remote browser control, authenticated browsing, real browser, cookie-based access, browser relay.
proof-agent
Adversarial verification of AI-generated work. Spawns an independent verifier to check for false claims, broken code, and security issues.
agent-context-system
A persistent local-only memory system for AI coding agents. Two files, one idea — AGENTS.md (committed, shared) + .agents.local.md (gitignored, personal). Agents read both at session start, update the scratchpad at session end, and promote stable patterns over time. Works across Claude Code, Cursor, Copilot, Windsurf. Subagent-ready. No plugins, no infrastructure, no background processes.