cocoindex
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from user-specified source directories, which could contain malicious instructions designed to influence the agent's behavior during the ETL pipeline.
- Ingestion points: Source files and directories specified by the user for processing via the
cocoindex updatecommand as described inSKILL.md. - Boundary markers: The instructions advise the agent to "Map the flow before running it" but do not define strict delimiters or automated sanitization for the data being processed.
- Capability inventory: The agent can execute shell commands via the
Bashtool and thecocoindexCLI, which can interact with local files and external databases/vector stores. - Sanitization: Instructions in
SKILL.mdexplicitly mandate manual inspection of source connectors, transforms, and environment variables before execution to mitigate risks. - [COMMAND_EXECUTION]: The skill requires the execution of local Python scripts and CLI tools to function.
- Evidence: The skill uses
python3 .agent-skills/cocoindex/scripts/preflight.pyfor inventory checks and thecocoindexCLI for pipeline operations, as demonstrated in the Examples section ofSKILL.md.
Audit Metadata