tailoring-resumes
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted Job Descriptions (JD) provided by the user (Step 1).
- Ingestion points: Data is ingested in
SKILL.mdduring the "Locate inputs and get the job description" phase. - Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions when the agent processes the JD content.
- Capability inventory: The skill performs file system writes in the
tailored/directory and executes shell and Python scripts (compile_check.sh,keyword_check.py) using the agent's environment. - Sanitization: No explicit sanitization or filtering of the Job Description content is specified before the agent uses it to drive the tailoring process.
- [COMMAND_EXECUTION]: The skill workflow involves executing local scripts and the Typst compiler to perform its tasks.
- Evidence:
SKILL.md(Step 5 and Step 6) instructs the agent to runbash skills/tailoring-resumes/scripts/compile_check.shandpython3 skills/tailoring-resumes/scripts/keyword_check.py. - Analysis: These operations are legitimate parts of the skill's intended functionality to verify page counts and keyword coverage, but they represent an execution surface that relies on the agent's handling of file paths and inputs.
Audit Metadata