PreToolUse hooks should still be able to fire in bypass / dangerously-skip-permissions mode
Summary
Hammurabi integrates Claude Code through PreToolUse hooks and has moved to a single approval-on runtime mode, so this is now lower priority for us. Still, bypass mode currently removes an important extension point for policy, audit, and telemetry integrations.
Current behavior
When Claude Code runs in bypass / dangerously-skip-permissions mode, PreToolUse hooks do not appear to fire for tool calls.
Why this matters
There are valid cases where an integration wants:
- no user-facing permission prompts
- but still wants hook visibility for audit, policy logging, or external approval systems
Right now those concerns are coupled.
Ask
Please consider an option that keeps PreToolUse hook delivery active even when bypass mode is enabled.
For example, something conceptually like:
- bypass permission prompts
- still emit PreToolUse events
- let the hook observe or optionally block the action
Even a documented statement of intended behavior would help, but a supported hook-on-bypass mode would make policy integrations much simpler.
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗