analyse-smartmoney-ptr

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface as it ingests untrusted data from an external web source.
  • Ingestion points: The agent is instructed in SKILL.md to use WebFetch on https://www.capitoltrades.com/trades to retrieve transaction data.
  • Boundary markers: Absent. The instructions do not define delimiters or provide warnings for the agent to ignore instructions embedded within the fetched content.
  • Capability inventory: The skill utilizes a local Python script (watch.py) which performs file system writes to a JSONL ledger and is executed via shell commands.
  • Sanitization: The watch.py script performs basic validation on dates and restricts chamber and action fields to specific enums, but does not sanitize other content strings.
  • [EXTERNAL_DOWNLOADS]: The skill fetches transaction data from Capitol Trades, a well-known service for tracking political financial disclosures. This operation is the primary stated purpose of the skill.
  • [COMMAND_EXECUTION]: The skill executes a local Python script, watch.py, to manage the state of recommended tickers. This involves shell command invocation by the agent to record and query the ledger.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 02:21 PM
Security Audit — agent-trust-hub — analyse-smartmoney-ptr