# Canon Boundary Guard

Canonical page: https://xxyoudeadpunkxx.github.io/canon-boundary-guard-codex/
Repository: https://github.com/XxYouDeaDPunKxX/canon-boundary-guard-codex
License: CC BY-SA 4.0

Canon Boundary Guard is a Codex plugin that gives Codex a cognitive frame for
keeping project evidence, chat context, operator instructions, working
hypotheses, and model assumptions in separate layers during a session.

The bundled skill is not only a pre-edit guard. It is intended to support
reading, analysis, planning, conflict detection, and persistence decisions
across long or consequential Codex sessions.

Core files in the repository:

- .agents/plugins/marketplace.json: local marketplace entry
- plugins/canon-boundary-guard-codex/.codex-plugin/plugin.json: plugin manifest
- plugins/canon-boundary-guard-codex/hooks/hooks.json: plugin hook configuration
- plugins/canon-boundary-guard-codex/skills/canon-boundary-guard/SKILL.md: full skill instructions and operating frame
- plugins/canon-boundary-guard-codex/skills/canon-boundary-guard/references/frame.md: compact frame used by the hook
- plugins/canon-boundary-guard-codex/skills/canon-boundary-guard/scripts/inject_frame.py: hook payload generator
- plugins/canon-boundary-guard-codex/skills/canon-boundary-guard/agents/openai.yaml: Codex-facing skill metadata

Layer summary:

- L0: persistent or verified evidence
- L1: conversation shaping not approved for persistence
- L1A: operator-approved delta pending persistence
- L2: agent-control instructions
- L2A: Codex instruction-chain material
- L3: unverified model prior

Machine-readable companion:

- Manifest: https://xxyoudeadpunkxx.github.io/canon-boundary-guard-codex/raw-manifest.json
- Sitemap: https://xxyoudeadpunkxx.github.io/canon-boundary-guard-codex/sitemap.xml