convert
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill contains a 'Self-Evolving Skill' section that explicitly instructs the agent to modify the
SKILL.mdfile based on its runtime experience. This creates a mechanism for instructions persistence where the agent's core logic can be permanently altered. - [INDIRECT_PROMPT_INJECTION]: The skill processes external
.cast(asciinema) files and converts them to text for analysis. Because the skill also encourages the agent to 'fix' its own instructions based on workarounds or issues encountered during processing, a malicious recording could contain data designed to trigger a 'fix' that injects malicious instructions into the skill's own manifest. - Ingestion points: Reads
.castfiles from user-specified directories (--source,fileargument). - Boundary markers: None identified in the workflow descriptions to separate file content from agent instructions.
- Capability inventory: The skill uses
Bash,Glob, andWritetools, and explicitly suggests modifying its ownSKILL.mdfile. - Sanitization: No evidence of sanitization for terminal escape sequences or hidden instruction patterns within the
.castfiles before conversion to.txtfor analysis.
Audit Metadata