ceo-advisor
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local Python scripts (
scripts/strategy_analyzer.pyandscripts/financial_scenario_analyzer.py) for processing company data and financial models. These scripts are bundled with the skill and use standard Python libraries for calculation. - [DATA_EXFILTRATION]: The skill accesses
~/.claude/company-context.mdto gather background on the company. While this file contains sensitive business context, the access is restricted to a specific application-context file and is directly related to the skill's core purpose of providing tailored CEO-level advice. - [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by ingesting external data through the company context file and JSON data provided to the analysis scripts.
- Ingestion points: Reads
~/.claude/company-context.mdand accepts JSON input from stdin for the analytical scripts. - Boundary markers: No specific boundary delimiters are defined in the instructions for the file ingestion.
- Capability inventory: The skill primarily performs mathematical analysis and generates reports; it does not contain network request capabilities or file system modification logic.
- Sanitization: The analysis scripts use
json.load()for structural validation of the input data, but do not perform explicit natural language sanitization of text fields.
Audit Metadata