job-search-run

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted job descriptions retrieved from external URLs which creates an attack surface for indirect prompt injection.\n- Ingestion points: External job content is fetched by scripts/fetch-posting.sh and persisted to the local workspace via scripts/record-api-response.sh.\n- Boundary markers: Instructions in SKILL.md structure the evaluation subagent prompts to provide context and limit the influence of the untrusted job data.\n- Capability inventory: The system executes shell scripts and awk programs to manage a local JSONL log, interacts with the vendor API via the agent-data tool, and generates reports.\n- Sanitization: The esc function implemented in multiple awk-based scripts (scripts/record-judgment.awk, scripts/record-api-response.sh, etc.) ensures control characters and structural JSON tokens are escaped before untrusted data is written to the log.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 01:53 PM