aiboarding — Claude Code plugin: onboarding, installation
The aiboarding plugin keeps AGENTS.md and CLAUDE.md aligned, tracks lifecycle changes, and audits repository guidance for AI coding agents.
MIT ★ 1,596 updated 2026-10-29
/plugin marketplace add hashgraph-online/awesome-codex-plugins
/plugin install aiboarding@awesome-codex-pluginsWhat aiboarding does
aiboarding creates a canonical AGENTS.md, a thin CLAUDE.md import, .aiboarding/state.json, and hooks behind a validation gate. Its update flow checks the changed scope, patches affected guidance, or advances the state pointer when no instruction change is needed.
The plugin can move legacy AIBOARDING.md guidance into the current layout and compress instructions while preserving commands, paths, URLs, and code. Audit mode leaves files untouched while reporting stale commands, contradictions, secrets, size risks, and instruction-chain limits.
Codex can use portable SKILL.md files copied to .agents/skills/, but that path does not install lifecycle hooks, so updates must be run manually. On Windows, the verification command bash tests/run.sh requires Git Bash.
What it helps with
- create canonical
AGENTS.mdguidance for coding agents - keep
CLAUDE.mdlinked to the main instructions - update onboarding after meaningful repository changes
- migrate guidance from
AIBOARDING.md - audit instructions for stale commands and contradictions
Example prompts for the agent
- Create the repository onboarding files with
/aiboarding:create-agent-onboarding. - Refresh the onboarding after meaningful commits using
/aiboarding:update-agent-onboarding. - Audit the guidance for stale commands, contradictions, secrets, and size risks.
How to install aiboarding
Plugin marketplace
/plugin marketplace add hashgraph-online/awesome-codex-plugins/plugin install aiboarding@awesome-codex-pluginsStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about aiboarding
How do I install aiboarding in Claude Code?
Run /plugin marketplace add hashgraph-online/awesome-codex-plugins first, then run /plugin install aiboarding@awesome-codex-plugins.
What does aiboarding do?
It manages the onboarding lifecycle for AI coding agents: creating and updating AGENTS.md, keeping CLAUDE.md as an import, migrating the legacy layout, compressing guidance, and auditing it.
Can I use aiboarding with Codex?
Codex can consume portable SKILL.md files from .agents/skills/. This route does not install lifecycle hooks, so updates are manual.
Source text
aiboarding · MIT
# AIBoarding <picture> <source media="(prefers-color-scheme: dark)" srcset="./assets/aiboarding-wordmark-dark.svg"> <img src="./assets/aiboarding-wordmark-light.svg" alt="AIBoarding — a current boarding pass for every coding agent" width="600"> </picture> **AIBoarding keeps `AGENTS.md` alive.** It is a repository-onboarding lifecycle for maintainers using AI coding agents: canonical guidance stays current across supported agents without duplicated instruction sources or broad rewrites of unchanged guidance. [](./LICENSE) [](./RELEASE-NOTES.md) ## Why Agents arrive like new engineers: they need the stack, commands, architecture, guardrails, and known failure modes before touching code. Without maintained guidance they repeatedly rediscover the repository—**onboarding drift**. AIBoarding creates a canonical `AGENTS.md`, keeps `CLAUDE.md` as a thin import, and tracks lifecycle state outside instruction files. ## See it  The example is derived from [`tests/fixtures/modern`](./tests/fixtures/modern): it demonstrates the intended file relationship, not a live install or a universal agent-quality result. ## Quick start (Claude Code) ```text /plugin marketplace add gustavo-meilus/aiboarding /plugin install aiboarding@aiboarding ``` Then create the onboarding files and lifecycle: ```text /aiboarding:create-agent-onboarding ``` Run the deterministic suite after repository changes (Git Bash is required on Windows): ```bash bash tests/run.sh ``` ## Proof and limits - The offline lifecycle contract is covered by [`bash tests/run.sh`](./tests/run.sh) and the focused test suites. - Live-runtime loading and hook delivery are opt-in protocols; their current availability and degraded paths are recorded in the [verification runbook](./docs/VERIFICATION.md). - The retained [benchmark report](./benchmarks/agent-outcomes/results/codex-command-discovery-v2/report.md) is limited to its task and runtime. It is not a general speed, token, or coding-quality claim. - Portable `SKILL.md` files can be copied to `.agents/skills/` for Codex and Copilot CLI. That route does not install lifecycle hooks, so run updates manually. ## What it does | Lifecycle step | Outcome | | --- | --- | | Create | Generates `AGENTS.md`, thin `CLAUDE.md`, sidecar state, and hooks behind a validation gate. | | Update | Triages changed scope; patches affected guidance or advances only the state pointer on a no-op. | | Migrate | Moves legacy `AIBOARDING.md` guidance to the current layout without discarding it. | | Compress | Reduces instruction bloat while byte-preserving commands, paths, URLs, and code. | | Audit | Reports stale commands, contradictions, secrets, size risks, and instruction-chain limits without editing files. | Use `/aiboarding:update-agent-onboarding` after meaningful commits, `/aiboarding:migrate-aiboarding` for the legacy layout, and the namespaced compression/audit skills when needed. ## Safety and supported surfaces `AGENTS.md` is the canonical source; `CLAUDE.md` imports it rather than duplicating it. The sync pointer remains in `.aiboarding/state.json`, preventing bookkeeping from modifying instructions. Hooks are small, deterministic, and advisory: native loading still works if hooks are unavailable, disabled, or untrusted. Older runtimes and Windows systems without Git Bash can have reduced hook behavior; see the runbook for exact boundaries. Claude Code has the tested marketplace installation above. Codex plugin installs include optional native hooks subject to `/hooks` trust review. Copilot, Cursor, and other agents consume canonical repository guidance through their native or portable instruction mechanisms; live compatibility verification is documented rather than implied. ## Architecture and documentation `skills/` owns lifecycle reasoning, `templates/` owns installed copies, and `tests/` prove deterministic behavior. Read [AGENTS.md](./AGENTS.md) for repository conventions, [loop engineering](./docs/LOOP-ENGINEERING.md) for where this fits in an agent loop, [verification](./docs/VERIFICATION.md) for manual protocols, and [release notes](./RELEASE-NOTES.md) for versioned history. ## Help, contributing, and security Report reproducible bugs or propose scoped features through [GitHub issues](https://github.com/gustavo-meilus/aiboarding/issues). Read [CONTRIBUTING.md](./CONTRIBUTING.md), follow the [Code of Conduct](./CODE_OF_CONDUCT.md), and use [SECURITY.md](./SECURITY.md) for private vulnerability reporting—do not file vulnerabilities as public issues. Licensed under [MIT](./LICENSE). Keep your agents current, then let them get on with the work.
© hashgraph-online · MIT · hashgraph-online/awesome-codex-plugins