account-sourcing
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external job boards via the Apify MCP. This content (job titles, descriptions) is processed by a Python script and then likely reviewed by the agent or presented to the user, creating a surface for embedded instructions.
- Ingestion points: Job data fields such as
description_textandtitleare processed inscripts/filter_score.py. - Boundary markers: The skill does not explicitly use delimiters or "ignore previous instructions" warnings when processing the job description text.
- Capability inventory: The script performs file read/write operations and calculations (score calculation, tiering).
- Sanitization: The script performs whitespace normalization and length capping for the
proof_snippet, but does not sanitize for potential prompt injection content. - [COMMAND_EXECUTION]: The skill instructions involve executing a local Python script
scripts/filter_score.pyto filter and score accounts. The script uses standard libraries and performs local file I/O to read JSON inputs and write CSV outputs.
Audit Metadata