Skip to content
Skills Library
AI & agents · Tool integration

agent-kernel

/agent-kernel

Use Agent Kernel in Claude Code when work needs shared durable rules, reviewed memory proposals, Failure Lessons, bounded local context, sessions, safe project linking, Architecture Guardian, Claude hooks, local MCP, project provider isolation, daemon security, or release verification. Trigger on: remember this rule, capture this failure, search past debugging evidence, connect this repo safely, prevent architecture drift, configure MCP, install narrow hooks, secure the daemon, or make coding agents follow one reviewed workflow.

Licence
MIT
Last repository change
27.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
5
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use Agent Kernel in Claude Code when work needs shared durable rules, reviewed memory proposals, Failure Lessons, bounded local context, sessions, safe project linking, Architecture Guardian, Claude hooks, local MCP, project provider isolation, daemon security, or release verification. Trigger on: remember this rule, capture this failure, search past debugging evidence, connect this repo safely, prevent architecture drift, configure MCP, install narrow hooks, secure the daemon, or make coding agents follow one reviewed workflow.

Original author description

Use Agent Kernel in Claude Code when work needs shared durable rules, reviewed memory proposals, Failure Lessons, bounded local context, sessions, safe project linking, Architecture Guardian, Claude hooks, local MCP, project provider isolation, daemon security, or release verification. Trigger on: remember this rule, capture this failure, search past debugging evidence, connect this repo safely, prevent architecture drift, configure MCP, install narrow hooks, secure the daemon, or make coding agents follow one reviewed workflow.

How it works

  1. Start
  2. Core contract
  3. Claude memory and failure flow
  4. Claude MCP
  5. Claude hooks
  6. Required architecture loop
  7. Validation
Original SKILL.md headings · 110 lines, 399 words
SKILL.md preview
---
name: agent-kernel
description: |
  Use Agent Kernel in Claude Code when work needs shared durable rules, reviewed memory proposals,
  Failure Lessons, bounded local context, sessions, safe project linking, Architecture Guardian,
  Claude hooks, local MCP, project provider isolation, daemon security, or release verification.
  Trigger on: remember this rule, capture this failure, search past debugging evidence,
  connect this repo safely, prevent architecture drift, configure MCP, install narrow hooks,
  secure the daemon, or make coding agents follow one reviewed workflow.
---

# Agent Kernel for Claude Code

Use the canonical [`SKILL.md`](https://github.com/imMamdouhaboammar/agent-kernel/blob/master/SKILL.md) for the complete workflow. This adapter adds Claude-specific MCP and hook guidance without weakening the shared trust model.

## Start

```bash
agent-kernel --version
agent-kernel doctor
agent-kernel status
agent-kernel mcp test
```

For an existing project:

```bash
agent-kernel compile
agent-kernel-safe-link . --dry-run
agent-kernel-safe-link .
```

Claude should read `CLAUDE.md` and `AGENTS.md`. Existing content outside Agent Kernel managed blocks remains user-owned.

## Core contract

- Approved source memory is canonical; generated guidance is delivery output.
- Claude may capture evidence and create pending proposals. Approval and publication remain user-owned.
- Failure Lessons are evidence first.
- File-backed IDs are identifiers, not paths.
- MCP core mode is the default. Approval is disabled unless explicitly enabled; publish and delete are not exposed.
- The daemon is local-only by default. Remote mode requires explicit opt-in and `AGENT_KERNEL_DAEMON_TOKEN`.
- Provider targets come from project manifests, not prompt or CLI overrides.
- Architecture policy, baseline, contract, and exception state require review.

## Claude memory and failure flow

```bash
agent-kernel propose \
  --from claude \
  --text "<durable rule>" \
  --reason "<why it should persist>"

agent-kernel failure capture \
  --from claude \
  --type tool-failure \
  --command "<command>" \
  --exit-code 1 \
  --text "<redacted failure>" \
  --root-cause "<supported cause>" \
  --fix "<verified fix>"
```

Do not approve a proposal merely because Claude created it.

## Claude MCP

```bash
claude mcp add --transport stdio --scope user \
  agent-kernel-memory -- agent-kernel mcp serve
claude mcp list
```

Keep core mode for normal work. Extended mode and MCP approval are separate explicit decisions.

## Claude hooks

Use narrow lifecycle events:

- `PreToolUse` for command or Architecture Guardian scope checks
- `PostToolUseFailure` for failure evidence
- short timeouts
- exec-form command hooks where supported
- structured, bounded output

Architecture scope hook:

```bash
agent-kernel architecture init .
# Configure Write|Edit|MultiEdit to call agent-kernel-architecture-hook.
```

Default Architecture Guardian hook mode is review. Set `AGENT_KERNEL_ARCHITECTURE_MODE=strict` only after policy and baseline review.

Hooks must not approve memory, publish memory, broaden architecture policy, create broad exceptions, or leak credentials.

## Required architecture loop

Load [`architecture-guardian`](../architecture-guardian/SKILL.md) before non-trivial code changes:

```bash
agent-kernel architecture doctor .
agent-kernel architecture discover . --json
agent-kernel architecture reuse "<capability>" . --json
agent-kernel architecture check . --json
```

## Validation

For Agent Kernel repository changes:

```bash
npm run docs:check
npm run lint
npm run typecheck
npm test
```

Read [docs/MCP_SERVER.md](https://github.com/imMamdouhaboammar/agent-kernel/blob/master/docs/MCP_SERVER.md), [docs/hooks/CLAUDE_HOOKS_BEST_PRACTICES.md](https://github.com/imMamdouhaboammar/agent-kernel/blob/master/docs/hooks/CLAUDE_HOOKS_BEST_PRACTICES.md), [docs/COMMAND_REFERENCE.md](https://github.com/imMamdouhaboammar/agent-kernel/blob/master/docs/COMMAND_REFERENCE.md), and [docs/SECURE_RUNTIME_AND_RELEASES.md](https://github.com/imMamdouhaboammar/agent-kernel/blob/master/docs/SECURE_RUNTIME_AND_RELEASES.md).

Versions

RepositoryScoreLicence
This page · immamdouhaboammar/agent-kernel100MIT
immamdouhaboammar/agent-kernel100MIT

Database snapshot: 13.11.2026 · package version 3d2bb96d6360 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 90agent-memory
    Recall durable project decisions, capture explicitly authorized lessons, and review memory history or corrections using the bundled local memory server. Use when continuity across sessions matters; skip standalone tasks that need no prior context.
  • 100agent-builder
    Build AI agents and automate Claude Code programmatically via the Claude Agent SDK and headless CLI mode. Covers Python SDK, claude -p, SDK MCP servers, hooks, sessions. Triggers on: "build an agent", "agent SDK", "headless mode", "automate Claude", "programmatic agent".
  • 100setup
    Set up, finish or re-tune the Claude Code configuration (CLAUDE.md, rules, settings, MCP servers, agents, hooks) for a project or for this device. Use when a "[cct]" session notice says setup is pending, right after installing the Claude Code kit, when the user runs /setup, or…
  • 86setup
    Install and connect memry, the persistent memory for AI coding agents. Use when the user wants to set up memry, asks why memry tools are missing, or wants Claude Code to remember context between sessions.
  • 100claude-code-cli
    Use when working with the Claude Code CLI (`claude`) — headless and interactive modes, flags, agents, plugins, MCP servers, hooks, settings, sessions, and the local custom ecosystem (forge, bob, alf, pa, wiki, agent-teams). Covers Claude Code 2.1.x. Refresh of the upstream…
  • 100cc-stack-setup
    Deploy or audit the full 5-layer Claude Code stack (CLAUDE.md + skills + hooks + agents + memory) in a repository. Use this skill whenever the user runs /cc-setup, asks to "set up Claude Code on this repo", "configure a new repo for Claude", "audit my Claude Code setup",…

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.