The Agent Skills Directory

[SAFE]: No malicious instructions, obfuscation, or security violations were found in the skill or its associated reference files.
[DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive data access or external network requests were identified; the skill interactions are confined to local development files.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and process untrusted skill files. Ingestion points: Reads files and directories provided as $ARGUMENTS in SKILL.md Phase 1. Boundary markers: None explicitly defined in the auditing instructions. Capability inventory: Uses Read, Write, and Edit tools, and has unrestricted access to Bash as allowed-tools is omitted in the frontmatter. Sanitization: No explicit sanitization or filtering of the audited skill's content is performed. This surface is evaluated as safe because it is required for the skill's primary purpose and is mitigated by a highly structured workflow.

repair-skill