Gate context injection for AI agents (CLAUDE.md / .gate awareness)

[jordanpartridge]

Idea

Every repo that uses gate should automatically inject context about gate into any AI agent working on it. When an agent (agentctl, Claude Code, etc.) encounters a gate-enabled repo, it should immediately know:

• What gate is and what it enforces
• The quality thresholds (coverage %, PHPStan level, style rules)
• How to run gate locally (`php gate certify`)
• Tips and tricks (common failure patterns, how to debug)
• The gate philosophy ("directions for completion" not just "FAIL")

Implementation options

Option A: Gate generates a CLAUDE.md section

Gate's GitHub Action or CLI could generate/inject a section into the repo's CLAUDE.md:

## Quality Gate (Synapse Sentinel)
This repo uses gate for quality enforcement.
- Run: \`php gate certify --coverage=80\`
- Checks: Tests+Coverage, Security Audit, Pest Syntax
- Coverage threshold: 80%
- Fix failures: gate posts actionable prompts on PRs

Option B: .gate/context.md convention

A `.gate/context.md` file that gate reads and agents can discover:

This project uses Synapse Sentinel Gate v2.x
Coverage requirement: 100%
PHPStan level: 6
Before committing: run \`php gate certify\`

Option C: Gate CLI command

`gate context` command that outputs agent-ready context:

php gate context > .claude-gate-context.md

Value

• Agents stop breaking gate by accident
• New contributors (human or AI) immediately understand quality bar
• Consistent gate knowledge across all repos in the ecosystem
• Reduces CI red → fix → re-push cycles

Considerations

• Should work with any AI agent tool (Claude Code, agentctl, GitHub Copilot)
• Keep context concise — agents have limited context windows
• Auto-detect config from `gate.yml` / `action.yml` / workflow files
• Include current coverage status if available