agent-kernellisted

Local-first governance, memory, architecture conformance, and safety layer for AI coding agents. Use Agent Kernel when the user wants Claude Code, Codex, Cursor, OpenCode, Antigravity, Gemini CLI, or AGENTS.md-compatible agents to share persistent rules, preferences, workflows, Failure Lessons, episodic memory, approval inbox proposals, generated guidance files, MCP tools, hooks, deterministic command guards, or reviewed architecture boundaries. Best triggers include: remember this rule, save this workflow, propose memory, search past episodes, capture this failure, learn from this error, prevent architecture drift, search before creating a service, create a change contract, baseline existing debt, install agent guidance, safe-link this repo, configure MCP, add a guard policy, or make multiple coding agents follow the same standards without repeating context.
imMamdouhaboammar/agent-kernel · ★ 1 · AI & Automation · score 67

Install: claude install-skill imMamdouhaboammar/agent-kernel

# Agent Kernel Agent Kernel is a local-first governance kernel for AI coding agents. It gives agents one shared source of truth for durable rules, project preferences, workflows, debugging lessons, architecture policies, generated instruction files, hooks, MCP tools, and safety checks. Use this skill when the user wants coding agents to remain consistent across projects and sessions, or when working code must also respect reviewed architectural boundaries. --- ## What this skill provides 1. Shared JSON-first memory at `~/.agent-kernel/source/memories/*.json`. 2. Approval inbox where agents propose memory and the user decides what gets published. 3. Failure Lessons for turning repeated build, test, edit, and command failures into searchable evidence. 4. Episodic memory for session history, debugging context, decisions, and investigations. 5. Architecture Guardian for project maps, dependency rules, reuse search, change contracts, baselines, exceptions, reports, and scope hooks. 6. Generated guidance for Claude Code, Codex, Cursor, OpenCode, Antigravity, Gemini CLI, and AGENTS.md-compatible agents. 7. Safe project linking that preserves existing instructions outside Agent Kernel marked blocks. 8. Claude hooks, git hooks, MCP tools, deterministic command guards, and repo-local ECC scaffolding. --- ## When to activate this skill Activate when the user says or implies: - remember this rule - save this as a workflow - make all agents follow this standard - stop the agent