audit-skilly-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/audit.mjsexecutesgitcommands usingchild_process.execFileSyncto retrieve repository metadata, file lists, and history for the audit process. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external instruction files which could contain malicious instructions.
- Ingestion points: The skill identifies and reads instruction files from the repository and the user's home directory (e.g.,
.claude/rules/*.md,~/.claude/CLAUDE.md,~/.codex/AGENTS.md). - Boundary markers: Absent; the instructions in
SKILL.mdguide the agent to read every source in full without explicit delimiters or sanitization instructions. - Capability inventory: The skill uses a Node.js script to execute shell commands, read/write local files, and the agent is instructed to fix identified configuration issues.
- Sanitization: Absent; the agent directly processes the content of instruction files to extract claims and judge contradictions.
Audit Metadata