What
A new feature that analyzes tool use requests and displays risk assessments (LOW/MEDIUM/HIGH) with explanations when Claude requests permissions.
Details
- Evaluates commands before execution and categorizes risk levels
- LOW risk: Standard dev workflows (git operations, tests, installs, builds)
- MEDIUM risk: Bulk deletions, complex shell pipelines, config modifications
- HIGH risk: Dangerous or irreversible operations, system-level changes
- Only displays warnings for MEDIUM and HIGH risk actions
- Controlled by
tengu_permission_explainerfeature flag - Can be enabled via
PERMISSION_EXPLAINER_ENABLED=trueenvironment variable
Evidence
Hu2() at line 417436 (permission explainer generator, contains "tengu_permission_explainer")
Strings lifted out of the shipped bundle, so the claim above can be checked against them.