qodo-codebase-wisdom
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes content from external repositories, pull requests, and commit logs, which creates a potential surface for indirect prompt injection attacks.
- Ingestion points: Untrusted data enters the agent context from codebase files, PR descriptions, and commit history via tools such as
qodo read codebaseandqodo read pull-request. - Boundary markers: The instructions do not specify the use of clear delimiters or explicit "ignore embedded instructions" warnings when the agent processes retrieved codebase content.
- Capability inventory: The skill uses the
qodoCLI to execute shell commands and perform searches, file reads, and metadata retrieval, which are then used to inform the agent's reasoning. - Sanitization: There is no explicit requirement or methodology provided for the agent to sanitize, escape, or validate external content before it is included in the reasoning process or final output.
- [COMMAND_EXECUTION]: The skill relies on executing a series of commands through the
qodoCLI and MCP tools. This includes authentication checks (qodo read whoami), codebase exploration (qodo read codebase grep), and potentially managing its own updates via the shell.
Audit Metadata