skill-forge-convert
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (
scripts/convert_skill.pyandscripts/validate_skill.py) to perform analysis, conversion, and validation tasks. While this is the intended functionality, it grants the agent the ability to run shell commands based on the skill's instructions. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a processing engine for external instruction files (SKILL.md, .mcp.json). This creates a vulnerability surface where a malicious skill being converted could contain instructions intended to override the agent's behavior during the conversion process.
- Ingestion points: Reads source
SKILL.md, parsed frontmatter,.mcp.jsonconfiguration, and other project files (scripts/, references/, agents/). - Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings to isolate the content of the source skill from the agent's current instructions.
- Capability inventory: The skill uses shell command execution via Python to run its conversion and validation logic.
- Sanitization: There is no mention of sanitization, escaping, or validation of the source skill's content before it is processed or analyzed.
Audit Metadata