bargainbadger-price-update
Fail
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: HIGHDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses a hardcoded absolute path to a sensitive directory in the user's home folder:
/Users/mtm/Documents/Obsidian Vault/bargainbadger.md. It also performs recursive file reads by resolving wikilinks found within these documents, which could lead to unauthorized access to other local files.- [INDIRECT_PROMPT_INJECTION]: The skill's behavior is driven by external, untrusted content including database rows and content from Obsidian notes. - Ingestion points: Accesses
/Users/mtm/Documents/Obsidian Vault/bargainbadger.md, linked notes, and database crawl sessions (Step 4). - Boundary markers: None present to distinguish between instructions and data.
- Capability inventory: Reading and writing local files, deleting files, and executing generated code.
- Sanitization: No evidence of sanitization or validation of the content retrieved from the database or notes.- [DYNAMIC_EXECUTION]: The workflow requires the agent to generate and run code to compute derived prices (Step 6). This execution is triggered by data parsed from external sources, increasing the risk of executing malicious logic if the source data is compromised.- [COMMAND_EXECUTION]: The skill performs file system management tasks, including creating and deleting scratch files outside of the vault environment (Step 8).
Recommendations
- AI detected serious security threats
Audit Metadata