title	L9 Governance
version	2.0.0
created	2025-01-27
updated	2026-07-19
owner	Igor Beylin
source	Post-Suite-6, Graphiti-native governance
tags	governance skills commands rules ops graphiti
domain	system-governance
type	documentation
production_ready	true

L9 Governance

🎯 Purpose

Centralized, IDE-agnostic governance system for L9/Quantum-L9 repos. The clone at ~/.cursor-governance/ is the governance root — there is no nested GlobalCommands/ subfolder. Every coding workspace exposes it through a single symlink: .cursor-commands → ~/.cursor-governance/.

See CANONICAL_LAW.md for the authoritative, binding contract this README summarizes, and AGENTS.md for exactly how a session activates this governance layer.

⚡ Activation (TL;DR)

Activation is automatic — one hook, no manual step:

ops/hooks/session_start_bootstrap.sh → installed at ~/.cursor/hooks/session-start-bootstrap.sh → registered in ~/.cursor/hooks.json under sessionStart. It syncs this clone, reconciles the declared Claude Code plugin set (ops/scripts/setup_claude_code_plugins.sh), reconciles the IDE profile (environment/ide/ — Biome/Ruff/Pyright extensions plus a managed-key merge into .vscode/settings.json), auto-wires symlinks, checks Graphiti (activated and round-trip verified as of 2026-07-27 — see AGENTS.md §2.3), and reads memory-bank/activeContext.md. See AGENTS.md §2 for the full activation contract and manual/repair commands. start-session.yaml (the old YAML "protocol") was retired 2026-07-19 — the .sh hook above is now the sole activation mechanism.

To trigger that same pipeline by hand — synchronously, with visible output:

make -C "$HOME/.cursor-governance" start WS="$(pwd)"

📁 Directory Structure

~/.cursor-governance/            (this repo)
├── skills/            # l9-* agent skills (SKILL.md per skill)
├── commands/          # Slash commands (/gmp, /plan, /end-session, ...)
├── rules/             # Global .mdc rules, symlinked as @.cursor-commands/rules
├── workflows/         # DAG definitions + executors
├── ops/
│   ├── scripts/       # Active automation (setup, validation, backup, sync)
│   ├── scripts/_archived/  # Retired pre-Graphiti / Suite-6 scripts (do not depend on)
│   ├── hooks/         # sessionStart / sessionEnd hooks
│   ├── graphiti/       # Graphiti memory client + activation runbooks
│   └── logs/          # Runtime logs
├── intelligence/      # Active signal corpus — chat exports, distillation, mining
├── environment/       # Runtime environment adapters (IDE-neutral policy + per-target renderers)
│   ├── ide/           # Editor profile: policy.json + render.cursor.json (Cursor/VS Code)
│   └── claude-code/   # Claude Code environment (CLI · Web · Mobile) — committed .claude/ + account env
├── profiles/          # DEPRECATED — content ported into skills/ + rules/; pending retirement
├── learning/          # Curated lessons, repeated-mistakes, quick-fixes
├── protocols/         # GMP protocol contracts and templates
├── security/          # Security governance docs
├── integrity/         # Integrity verification docs/scripts
├── pipeline/          # Pipeline orchestration & validation docs
├── reports/           # GMP execution reports
├── C_GOV_FILES/       # Legacy duplicate tree — pending removal (see hygiene PRs)
├── ORG_INVARIANTS.yaml # Canonical Quantum-L9 org policy
├── CANONICAL_LAW.md   # Authoritative governance contract (read first)
├── AGENTS.md          # Activation contract + agent operating rules (read second)
└── README.md          # This file

execution-governance/, foundation/, operations/, environment/, telemetry/, key components/, prompts/, current_work/, and logs/ hold supporting docs, in-progress notes, and legacy scaffolding; treat CANONICAL_LAW.md and skills/*/SKILL.md as the sources of truth over any directory listing, including this one.

🔗 Access Methods

In Cursor workspaces

• .cursor-commands/ — the sole symlink target in every coding repo
• @.cursor-commands/skills/..., @.cursor-commands/rules/..., @.cursor-commands/commands/...

Direct path

• ~/.cursor-governance

📋 Key Files

Governance contract

• CANONICAL_LAW.md — SSOT, symlink law, memory layer, anti-patterns
• AGENTS.md — activation contract, change policy, agent operating rules
• ORG_INVARIANTS.yaml — canonical Quantum-L9 org policy

Skills

• skills/l9-gmp-protocol/SKILL.md — locked phase-0–6 execution
• skills/l9-structured-reasoning/SKILL.md — planning/debugging reasoning stack
• skills/l9-graphiti-memory/SKILL.md — Graphiti memory wiring

Learning

• learning/repeated-mistakes.md — critical mistakes to never repeat
• learning/quick-fixes.md — fast solution patterns

Operations

• ops/scripts/setup_workspace_symlinks.sh — install symlinks in a workspace
• ops/scripts/validate_governance_symlinks.sh — verify symlink wiring
• ops/scripts/backup_to_github.sh — commit + push SSOT to GitHub
• ops/scripts/backup_gate.sh — gate the sessionEnd backup (reason filter, debounce, activity guard)

🚀 Usage

Reference in prompts

@.cursor-commands/learning/repeated-mistakes.md
@.cursor-commands/skills/l9-structured-reasoning/SKILL.md
@.cursor-commands/rules/00-global.mdc

Wire a new workspace

bash .cursor-commands/ops/scripts/setup_workspace_symlinks.sh
bash .cursor-commands/ops/scripts/validate_governance_symlinks.sh

📊 Status

Origin: Quantum-L9/Cursor-Governance (GitHub is the SSOT remote; this clone at ~/.cursor-governance/ is the SSOT working copy — see CANONICAL_LAW.md §1, §5).

---

Last Updated: 2026-07-19 Version: 2.0.0

Program Execution

environment/program-execution/ contains the sealed Program Execution core, replaceable execution adapters, conformance contracts, routing policy, and bridges to existing Cursor-Governance runtimes. Mutable program state remains outside Git under $HOME/.l9/programs/.