serenity-stock-scorer

Pass

Audited by Gen Agent Trust Hub on Jun 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to run a shell command python skills/serenity-stock-scorer/scripts/score_serenity_stock.py <SYMBOL> --pretty. This pattern creates a risk of command injection if the agent fails to normalize the <SYMBOL> parameter as instructed, potentially allowing an attacker to execute arbitrary commands by supplying a symbol with shell metacharacters like ; or &&.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from a local SQLite database, presenting an indirect prompt injection surface. (1) Ingestion points: The scripts/score_serenity_stock.py script reads data from the tweets and mentions tables in the local serenity.sqlite database. (2) Boundary markers: The script filters raw database content and only returns structured metrics and metadata (e.g., signal names, IDs) to the agent, preventing raw tweet text from entering the agent's context. (3) Capability inventory: The skill utilizes shell command execution to run its scoring script. (4) Sanitization: Ticker symbols are normalized both by instruction to the agent and within the Python script logic itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 29, 2026, 04:02 PM
Security Audit — agent-trust-hub — serenity-stock-scorer