keek-setup
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to read external, untrusted content from the repository (README, lockfiles, scripts, etc.) to determine the project's setup and startup commands. This creates a surface for indirect prompt injection where malicious instructions in the repository files could influence the agent's profile generation.
- Ingestion points:
package.json,README.md,Makefile,.env.example, and various lockfiles (e.g.,package-lock.json,Cargo.lock) within the target repository. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are provided to the agent to distinguish between trusted instructions and data from the repository.
- Capability inventory: The skill uses
set_project_profileto write configuration andprepare_projectto execute shell commands in a sandbox environment. It also modifies repository files likeAGENTS.mdorCLAUDE.md. - Sanitization: The instructions do not specify any validation or sanitization steps for the data extracted from repository files before it is used to construct shell commands.
- [COMMAND_EXECUTION]: The skill facilitates the execution of arbitrary shell commands through the
installandstartfields in the project profile. While these commands are executed within a sandbox that resets the environment after build, the skill provides the mechanism for their definition and execution based on repository content.
Audit Metadata