landing-prs
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local scripts (
preflight.sh,ci-gate.mjs,merge.sh) that use thegitandghCLI tools to manage repository state and pull requests. These scripts include regex-based validation for critical inputs such as PR numbers and commit OIDs to prevent command injection. - [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to identify and execute "project's agent instructions" as a pre-merge step. This creates an attack surface where instructions found in repository files could influence the agent's behavior during the landing process.
- Ingestion points: Project instructions found within the repository and PR metadata like titles.
- Boundary markers: No specific boundary markers or "ignore instructions" delimiters are defined for the project-level instructions.
- Capability inventory: The agent has permissions for file system access, shell execution, and git push/merge operations.
- Sanitization: While CLI parameters are validated, the content of the project-specific instructions is not sanitized or filtered by the skill logic.
Audit Metadata