release-sandbox-tools
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to execute a local shell script,
scripts/release-sandbox-tools.sh, which performs sensitive actions such as building binaries, managing Docker containers, and uploading files to S3 using the user's ambient AWS credentials. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process code from Pull Requests, which may originate from external contributors. This creates an attack surface where malicious code in a PR could attempt to influence the agent's behavior during the build and release process.
- Ingestion points: Source code within the
src/inspect_sandbox_tools/directory of a Pull Request. - Boundary markers: The instructions do not define delimiters or specific 'ignore' directives to prevent the agent from being influenced by instructions embedded in the source files.
- Capability inventory: The agent can execute shell scripts, use Docker, perform S3 uploads via the AWS CLI, and push commits to the Git repository.
- Sanitization: The skill relies on external manual PR approval and does not specify any internal validation or sanitization steps performed by the agent itself.
Audit Metadata