{
  "markdown": "# Aegis Boundary Skill\n\n> Verify what AI agents are allowed to see, do, and output.\n\n`aegis-boundary` is a free, official Aegis skill for AI coding agents. It\nperforms an **advisory review** of agent work around Aegis surfaces — code\nchanges, documents, public claims, and agent task plans — and returns\n**READY / REVIEW / STOPPED** with machine-check evidence.\n\nIt is a review layer, not an enforcer: it does **not** authorize production\naccess, enforce policy, or produce receipts. Runtime enforcement belongs to\nthe [aegis-trust SDK](https://aegisagentcontrol.com) (LITE, in-process\nhygiene) and Aegis Core (FULL, gateway enforcement with key separation).\n\n## What it checks\n\n- **Boundary**: does the work keep the SDK / Gateway / Core responsibility\n  split intact (cooperative hygiene vs. non-cooperative enforcement)?\n- **Claims**: do README / site / sales statements stay within what is\n  implemented? LITE/FULL claim separation, never-claims, certification\n  wording, and three machine-checked overclaim families.\n- **Agent plans**: is what the agent is about to see, do, and output within\n  declared purpose and scope — without treating a purpose label as\n  authorization?\n\nOutput shape:\n\n```\nDecision: READY | REVIEW | STOPPED\nMachine checks: overclaim_scan PASS|FAIL\nWhy / Allowed (advisory) / Blocked (advisory) / Required checks / Next\n```\n\n## Requirements\n\nThe overclaim scan (`scripts/check_overclaims.py`) needs **Python 3.7+**. The\nskill instructions themselves are plain markdown and need no runtime.\n\n## Install\n\n**Claude Code** (plugin marketplace):\n\n```\n/plugin marketplace add Incierge3789/aegis-boundary-skills\n/plugin install aegis-boundary@aegis-skills\n```\n\n**Codex** (agent skills):\n\n```bash\nmkdir -p ~/.agents/skills\ncp -R skills/aegis-boundary ~/.agents/skills/\n```\n\n(or `./install.sh codex`)\n\n**Gemini CLI** (extension):\n\n```bash\ngemini extensions install https://github.com/Incierge3789/aegis-boundary-skills\n```\n\n**Cursor** (plugin): this repository follows the Cursor plugin layout\n(`.cursor-plugin/plugin.json` + `skills/`); marketplace listing pending.\n\n## Layout\n\n```\nskills/aegis-boundary/\n├── SKILL.md                      # the skill (trigger spec + review procedure)\n├── references/boundary_canon.md  # vendored boundary canon (parity-tracked)\n├── references/pitfalls.md        # defaults a model gets wrong\n├── templates/boundary_review.md  # READY/REVIEW/STOPPED output template\n└── scripts/check_overclaims.py   # machine check (3 overclaim families)\n```\n\n## Honesty contract\n\nThis skill applies the same rules to itself that it reviews for: it is\ndescribed as advisory everywhere, its bundled canon is parity-tracked against\nthe internal canonical document (`docs/PARITY.md`), and its own files pass its\nown machine checks.\n\n## License\n\nMIT — see `LICENSE`. Copyright (c) 2026 Aegis Project (Incierge).\n\nThe MIT license covers the code and text of this repository. It does not\ngrant rights to the \"Aegis\" name or logos, or the right to represent forks as\nofficial Aegis distributions. The official distribution of this skill is\n`github.com/Incierge3789/aegis-boundary-skills`.\n",
  "bytes": 3120,
  "sha": "70c166a20debd38aac71b294ac633216bef3f6826ca88c11f2f65c40c8ccdabb",
  "repo_slug": "nemotek-inc/aegis-boundary-skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_nemotek_inc_aegis_boundary_skills_349994bc/readme"
}