Skip to content

CLAUDE.md

Latest commit

 

History

History
39 lines (30 loc) · 1.56 KB

CLAUDE.md

File metadata and controls

39 lines (30 loc) · 1.56 KB

hookify

Rule-based behavior prevention. Intercepts Claude Code events and blocks/warns based on user-defined patterns.

Files

Directory Contents
hooks/ 4 Python hook handlers (pretooluse, posttooluse, stop, userpromptsubmit) + hooks.json
hookify/core/ config_loader.py (parses rules), rule_engine.py (evaluates + caches regex)
commands/ hookify.md (create rules), configure.md (toggle rules), help.md, list.md
skills/writing-rules/ SKILL.md + references/patterns-and-examples.md
agents/ conversation-analyzer.md (scans transcripts for frustration signals)
examples/ 4 example rule files (.local.md)
global-rules/ Global rules applied across all projects

Rule File Format

---
name: rule-identifier
enabled: true
event: bash|file|stop|prompt|all
pattern: regex-pattern
action: warn|block
---
Message shown when rule triggers.

Rules loaded from TWO locations:

  1. .claude/hookify.*.local.md (project-local, higher priority)
  2. ~/.claude/global-rules/hookify.*.local.md (global, fallback)

Key Behavior

  • Hades permit exemption: active .smart/delete-permit.json bypasses ALL blocking rules.
  • Python hooks read stdin JSON, return JSON with event-specific fields: permissionDecisionReason (PreToolUse deny), reason (Stop/PostToolUse block), additionalContext (warnings on PostToolUse/UserPromptSubmit/SessionStart). systemMessage is user-display only — Claude never sees it.
  • Conversation analyzer searches for "Don't use X", frustrated reactions, repeated issues.