fable-anti-churn
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill manages a local ledger of agent actions in
.fable/anti-churn/receipts.ndjson. It employs SHA-256 hashing for identifiers like task IDs and attempt IDs to prevent the exposure of sensitive data in the ledger. - [REMOTE_CODE_EXECUTION]: The skill includes internal testing and validation utilities (
tests/test_churn_control.pyandtools/validate_release_candidate.py) that usesubprocessto execute its own logic for verification and concurrency testing. These are developer-focused tools and do not represent arbitrary command execution. - [DYNAMIC_EXECUTION]: The test suite uses
importlibto dynamically load the core logic script for unit testing, which is a standard Python practice for testing local modules. - [INDIRECT_PROMPT_INJECTION]: While the skill processes external action events, it uses strict schema validation and cryptographic hashing of inputs to minimize the risk of malicious data influencing agent behavior.
Audit Metadata