scan-all
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands via
node -eto interact with local session helpers and triggers scan commands for documentation targets as part of theExecutionandRun and verifystages. - [DYNAMIC_EXECUTION]: Utilizes
node -eto dynamically load and run logic from the local JavaScript file./.claude/hooks/lib/session-init-helpers.cjsto manage the scan staleness state. - [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific data which presents an indirect injection surface.
- Ingestion points: Reads and resolves configuration from
docs/project-config.jsonand the target registry in.claude/skills/scan/references/targets.md. - Boundary markers: Does not implement explicit delimiters for prompt interpolation, but relies on a strict internal mapping protocol.
- Capability inventory: Includes file system write operations to the
docs/directory and shell command execution via Node.js. - Sanitization: The skill explicitly implements config validation to reject unknown targets and unsafe paths, and runtime path resolution that rejects physical symlink escapes.
Audit Metadata