Project: pal-e-config

project-pal-e-config Project Page

active project-page

Project

Project: pal-e-config

Vision

The enforcement and compliance pillar of the four-pillar operating model. pal-e-config is the mechanism that makes Agency's rules real — hooks that block unauthorized actions, agent frontmatter that constrains capabilities, session injection that provides context, and the development SOP that keeps the enforcement layer itself safe. If Agency says "agents must do X," Config makes it impossible not to.

User Stories

Role	Story	Success Metric
Agent (Dev/QA)	When I try to violate an SOP, the hook blocks me and tells me why	Zero SOP violations in merged PRs
Betty Sue	When I spawn an agent, it gets the right context injected automatically	No manual context pasting; SessionStart hook provides everything
Lucas	I can trust that enforcement rules are versioned, reviewed, and testable	All hook changes go through PR review in claude-custom
Dottie	Write operations to pal-e-docs are blocked for unauthorized agents	block-docs-writes.sh covers all 17+ write ops

Plan

No active plan. The enforcement layer is stable — hooks and agent configs are maintained as part of other project plans (primarily pal-e-agency). A dedicated plan will be created when there's a roadmap for Config-specific features (e.g., hook testing framework, enforcement coverage metrics).

Completed plans (historical):

plan-2026-02-24-sop-enforcement — SOP Enforcement via Hooks
plan-2026-02-24-enforcement-hooks-mvp — Enforcement Hooks MVP
plan-2026-02-28-agent-skill-frontmatter — Agent & Skill Frontmatter Fix

Board

See board-pal-e-config (created 2026-03-14).

Status

As of 2026-03-14:

Hooks LIVE — 7+ PreToolUse hooks active via hardlinks from ~/claude-custom/hooks/ to ~/.claude/hooks/. Auto-deploy on git pull.
SessionStart hook LIVE — Injects platform context, SOPs, active plans, MCP health check at session start.
Ruff auto-format hook LIVE — PreToolUse hook runs ruff format on staged .py files before every git commit. Prevents #1 CI failure cause.
block-docs-writes.sh LIVE — Blocks all 17 pal-e-docs write operations for unauthorized agents.
check-pr-template.sh LIVE — Blocks PR submission without Closes #N.
check-mcp-servers.sh LIVE — SessionStart MCP health check (precursor for Phase 8).
Agent configs — 5 agent .md files in ~/.claude/agents/ (betty-sue, dev, qa, dottie, penny).
Commands — /update-docs command live. Commands require manual copy to ~/.claude/commands/ after merge.

Milestones

No milestone notes yet. Key historical milestones:

2026-02-24 — First enforcement hooks deployed (SOP enforcement MVP)
2026-02-28 — Agent & skill frontmatter fix + end-to-end verification
2026-03-09 — Ruff auto-format hook deployed (PR #81, claude-custom)
2026-03-13 — Hook security hardened (PR #70, all 17 write ops blocked)
2026-03-14 — MCP health check hook deployed (PR #88, claude-custom)

Architecture

Core architecture notes:

enforcement-architecture — the four enforcement pillars (PreToolUse hooks, SessionStart injection, agent frontmatter, block-first conventions). See sop-hook-block-recovery for procedures.
hook-events-reference — complete reference of Claude Code hook events and their parameters.

Architecture diagram notes (arch-domain, arch-dataflow, arch-deployment) not yet created for this project.

Repos

Repo	Platform	Role	Status
<code>claude-custom</code>	Forgejo	Hooks, agent configs, skills, commands, settings	Active

Chicken-and-egg: Hooks in this repo can't enforce changes to themselves. QA review is the primary enforcement gate. See sop-claude-config-development.

Inbox

Query: list_notes(project="pal-e-config", note_type="todo", status="open")

project-pal-e-agency — Agency defines the rules, Config enforces them
sop-claude-config-development — development workflow for this repo
convention-cross-pillar-triggers — Agency changes trigger Config reviews and vice versa
convention-arch-sop-pairing — enforcement-architecture must pair with sop-hook-block-recovery

Project: pal-e-config