skillify
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill processes session history and user-provided arguments to generate a new
SKILL.mdfile. This creates a surface where instructions from the conversation history could be incorporated into the generated skill file. - Ingestion points: Session history and
$ARGUMENTS(SKILL.md). - Boundary markers: None explicitly defined for the interpolation of captured content.
- Capability inventory:
write_fileandbash(SKILL.md). - Sanitization: Not explicitly performed on captured content before writing to file.
- Local Command Execution: The skill uses the
bashtool to run a Python snippet (python3 -c ...) to validate the length of the YAML frontmatter description in the newly created skill. This is a local validation step performed on the generated file. - File Persistence: The skill is designed to write new instruction files to project-local directories (
.amplifier/skills/) or the user's personal home directory (~/.amplifier/skills/), which allows it to save workflows for use across different sessions.
Audit Metadata