databow
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of the
databowCLI tool to perform SQL queries and thedbctool to manage database drivers on the local system. - [EXTERNAL_DOWNLOADS]: The
dbc installcommand is used to download and install ADBC drivers from the vendor's repository to enable database connectivity. - [CREDENTIALS_UNSAFE]: The skill provides instructions for handling sensitive database credentials, including the use of the
--passwordand--usernameflags, as well as storing passwords in plain-text within TOML connection profile files. - [DATA_EXFILTRATION]: The tool facilitates reading data from databases and exporting it to local files such as CSV, JSON, or Arrow IPC, creating a potential surface for data exfiltration if the output is directed to sensitive locations.
- [PROMPT_INJECTION]: The skill processes external data from database query results, which presents an indirect prompt injection surface.
- Ingestion points: Data retrieved from various SQL databases via the
databowtool is returned to the agent's context (SKILL.md). - Boundary markers: None identified; query results are processed by the agent without explicit delimiters to separate data from instructions.
- Capability inventory: The agent can execute shell commands via
databowanddbc, and write to the local file system using the--outputflag (SKILL.md). - Sanitization: No mechanisms for sanitizing or filtering database content are described before the data is ingested by the agent.
Audit Metadata