aeo-core
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (LOW): The skill utilizes shell commands to persist and retrieve confidence signals. Evidence: usage of
echo,tail, andjqwithin theSKILL.mdfile. - [DATA_EXFILTRATION] (LOW): The skill accesses and modifies files within a hidden directory in the home folder. Evidence: writes to and reads from
~/.claude/MEMORY/aeo-signals.jsonl. - [PROMPT_INJECTION] (LOW): The system is susceptible to indirect prompt injection where a series of crafted 'success' signals could artificially inflate confidence scores to bypass safety gates. 1. Ingestion points:
~/.claude/MEMORY/aeo-signals.jsonl. 2. Boundary markers: None. 3. Capability inventory: Shell command execution and autonomous decision-making. 4. Sanitization: None; the confidence adjustment logic directly consumes values from the log file.
Audit Metadata