{ "markdown": "```\n ▄▀█ ▀▄▀ █▀▄▀█\n █▀█ █ █ █ ▀ █\n Agent Extension Manager by AgentXM\n\n https://axm.sh | https://agentxm.ai\n```\n\n![Status: Alpha](https://img.shields.io/badge/status-alpha-orange)\n[![npm version](https://img.shields.io/npm/v/axm.sh.svg)](https://www.npmjs.com/package/axm.sh)\n[![CI](https://img.shields.io/github/actions/workflow/status/agentxm/axm/ci.yml?branch=main&label=CI)](https://github.com/agentxm/axm/actions/workflows/ci.yml)\n[![License: FSL-1.1-MIT](https://img.shields.io/badge/license-FSL--1.1--MIT-blue.svg)](./LICENSE)\n\n# Extension manager for coding agents and AI assistants\n\n- Manage agent skills, subagents, MCP servers, rules, hooks, knowledge, and packs.\n- Bundle extensions with packs to share workflows across projects, teams, and others.\n- Switch effortlessly between coding agents, bringing your workflow with you. Avoid lock-in.\n\n> [!WARNING]\n> AXM is in **early alpha** testing. Capabilities and APIs may change. To be\n> included in the test user group, reach out to\n> [hello@agentxm.ai](mailto:hello@agentxm.ai).\n\n- **Multi-tool** — a unified context-engineering toolkit for managing skills,\n subagents, MCP servers, rules, hooks, knowledge, and packs\n- **Multi-agent** — share workflows across Claude Code, Codex, Cursor,\n Gemini CLI, GitHub Copilot, OpenCode, and\n [nearly 40 others](#supported-agents)\n- **Cross-platform** — discover and distribute agent skills and extensions\n across package ecosystems including JavaScript, Python, Java, .NET, Go,\n Rust, Ruby, and more\n\n## Installation\n\n**Ask your agent to install AXM:**\n\n```\nFollow these install instructions to set up AXM: https://axm.sh/install.md\n```\n\n**macOS / Linux:**\n\n```bash\ncurl -fsSL https://axm.sh/install.sh | sh\n```\n\n**Windows (PowerShell):**\n\n```powershell\nirm https://axm.sh/install.ps1 | iex\n```\n\n**Windows (CMD):**\n\n```cmd\ncurl -fsSL -o install.cmd https://axm.sh/install.cmd && install.cmd\n```\n\n**Homebrew:**\n\n```bash\nbrew install agentxm/tap/axm\n```\n\n**npm:**\n\n```bash\nnpm install -g axm.sh\n```\n\n## Getting started\n\nInitialize AXM in your project. AXM detects your installed agents and creates\nroot `axm.json`; acquired packages and accepted external resolutions are kept\nin `agent_extensions/` and `axm-lock.yaml`, while `.axm/` is runtime scratch.\nAXM leaves Git attributes and formatter policy to the repository; exclude\n`agent_extensions/` from tools that rewrite files.\n\n```bash\naxm setup\n```\n\nSee what's recommended for your project:\n\n```bash\naxm discover\n```\n\nInstall an extension and AXM wires it into every agent in your workspace:\n\n```bash\naxm install @acme/skills/code-review\n```\n\n## Extension types\n\n\n\n| Type | What it is | Governing standard |\n| --------------- | ---------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |\n| **Skills** | Package reusable agent skills with SKILL.md metadata and instructions. | [Agent Skills](https://agentskills.io) |\n| **MCP Servers** | Configure Model Context Protocol servers for agents. | [Model Context Protocol](https://modelcontextprotocol.io) |\n| **Subagents** | Install specialized agent profiles into an agent's native subagent system. | — |\n| **Rules** | Sync instruction files and distribute rule extensions that inject into them. | [AGENTS.md](https://agents.md) |\n| **Hooks** | Install lifecycle hook extensions into an agent's native hook system. | — |\n| **Knowledge** | Package portable Open Knowledge Format concept bundles. | [Open Knowledge Format 0.2](https://github.com/GoogleCloudPlatform/knowledge-catalog/blob/main/okf/SPEC.md) |\n\n\n\n**Packs** are the container type: a pack curates any of the above into one\nbundle teams install and keep in sync, such as\n`@acme/packs/frontend-tools` — the frontend guild's standard skills and\ncommands.\n\n\n\nEvery type has its own subcommand namespace (`axm skills`, `axm mcps`, `axm subagents`, `axm rules`, `axm hooks`, `axm knowledge`, `axm packs`) sharing a common shape: `install`, `uninstall`, `list`, `update`, `new`, `publish`, plus `enable`/`disable` where it applies.\n\n\n\n## Subcommands\n\nExamples below use `skills`; the same shape applies to every type.\n\n```bash\naxm skills new my-skill # Scaffold a new skill\naxm skills install @acme/skills/code-review\naxm skills list\naxm skills disable my-skill # Turn off without uninstalling\naxm skills enable my-skill\naxm skills import ./external-skill @acme/skills/my-skill\naxm fork @acme/skills/code-review @me/skills/code-review-custom\naxm skills publish # Publish authored skills\n```\n\nInstalled and enabled skills are always materialized in `.agents/skills/` for\nthe agentskills.io format, plus any declared agent-native skill directories.\n\n`axm packs` adds bundling commands:\n\n```bash\naxm packs add my-pack @acme/skills/code-review\naxm packs add my-pack \"effect-*\" # Add all matching extensions\n```\n\nRun `axm --help` for the top-level command list, or `axm --help`\nfor options on any subcommand.\n\n## Workspace\n\nTop-level commands work across every extension type and infer the target from\nyour input:\n\n```bash\naxm install # Sync extensions from axm.json\naxm install @acme/skills/code-review # Install a single extension\naxm update # Pull latest versions\naxm list # Inventory extensions across all types\naxm list --outdated # Show extensions with available updates\naxm list --deprecated # Show deprecated installed extensions\naxm uninstall @acme/skills/code-review\naxm adopt @acme/skills/retained-package # Make a canonical package authoritative\naxm demote @acme/skills/review ./upstream # Explicitly return to external source management\naxm lint # Report intrinsic workspace facts\naxm sync --preview # Preview reconciliation without writing\naxm sync # Reconcile desired, accepted, and observed state\naxm upgrade # Update axm itself\n```\n\n## Publishing\n\nExtensions publish to the registry in four steps.\n\n```bash\naxm skills new my-skill # 1. Scaffold\n# 2. Author content in the scaffolded directory\naxm lint # 3. Check the publish gate locally\naxm publish # 4. Publish new authored versions; verify existing ones\n```\n\nAuthorship is derived from the exact intrinsic `workspace` settings source,\nthe project `owner`, the settings map key, the extension type, and the matching\nmanifest in that type's authored root. Bare, filtered, and explicit selections\npublish only workspace-authored packages. Adopt a retained canonical package\nwhen this workspace should own it, or fork an installed package to publish it\nunder a new identity.\n\nAXM builds a deterministic archive from each selected authored package. For an\nexisting immutable version, `--on-existing verify` rebuilds that archive and\nrequires its SHA-512 digest to match the Registry release before reporting a\nsuccessful no-op. Installed external packages represent the source identity recorded\nin `axm-lock.yaml`. `axm lint` checks their current contents against that accepted\nidentity. Publication reads workspace-authored packages.\n\n`axm lint` checks the same rules the registry enforces — see\n[Lint](#lint) for details.\n\n## Lint\n\n`axm lint` evaluates workspace and per-extension invariants against a shared\nrule catalog — the same one that drives the registry publish gate.\n\n```bash\naxm lint # Report findings against the current project\naxm lint --fix # Apply every autofixable finding non-interactively\naxm lint --scope user # Lint the user-scope workspace\naxm lint --strict # Exit non-zero on warnings as well as errors\naxm lint --json # Machine-readable findings envelope\n```\n\nProject scope is the default. Local `lint.rules` overrides in\n`axm.json` affect `axm lint` only — the registry publish gate\nremains authoritative.\n\n## Authentication\n\nSign in to a registry to publish, install private extensions, or attribute\nusage to your account:\n\n```bash\naxm login # Sign in to the default registry\naxm login --device-code # Sign in from SSH or a headless machine\naxm whoami # Show the current identity\naxm logout\naxm token # Print the current token (for scripting)\n```\n\n`axm login` starts a local loopback PKCE flow, prints a manual authorization\nURL, and then tries to open your browser. SSH, CI, and Codespaces automatically\nuse device-code sign-in; pass `--device-code` to select that flow explicitly.\nDevice-code sign-in opens a complete authorization link when possible and also\nshows the clean fallback page plus the one-time code. Retrying the command\nreuses the same unexpired request; use `--restart` only when you intentionally\nwant to replace it. Never enter a code that another person or website gave you.\n\n## Supported agents\n\nAXM supports nearly 40 agents — Claude Code, Codex, Cursor,\nGemini CLI, GitHub Copilot, OpenCode, Windsurf, Cline, Continue, Roo, Goose,\nand more. If your agent isn't there,\n[open an issue](https://github.com/agentxm/axm/issues) and we'll add it.\n\n## Development\n\nThis is an Nx monorepo.\n\n```bash\npnpm install # Install dependencies\npnpm build # Build all packages\npnpm test # Run tests\npnpm typecheck # Type check\npnpm lint # Lint\n```\n\nUse `pnpm build:affected`, `pnpm test:affected`, or `pnpm lint:affected` to\nonly operate on packages changed since `main`.\n\nSee [CONTRIBUTING.md](./CONTRIBUTING.md) for the full development workflow.\n\n---\n\n**Open Agents, Open Standards, Open Source.**\n\n## License\n\nAXM is licensed under the [Functional Source License](https://fsl.software/) (FSL), a [Fair Source](https://fair.io) license that converts to a MIT license after two years.\n\nFSL-1.1-MIT © 2025-2026 AgentXM, Inc. — see [LICENSE](./LICENSE).\n", "bytes": 10846, "sha": "b57778ff997e376e41685a31dc1474030d09576b20c935637ea14a69a8cc6186", "repo_slug": "agentxm/axm", "fonte": "repo", "truncated": false, "api": "https://agentalog.com/api/listings/okf_agentxm_axm_agent_extensions_agentxm_cra_33ab880a/readme" }