dstack-beads-project-alignment-review
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a binary file located at
{baseDir}/../../bin/dstack. This uses a relative path to execute a local controller tool for project alignment tasks. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze a wide variety of untrusted data from the repository environment, including source code, tests, documentation, and 'Beads' work records.
- Ingestion points: Reads repository source, durable documentation, architecture patterns, code, tests, and Beads records (SKILL.md, Step 2).
- Boundary markers: None explicitly defined in the instructions to separate untrusted repository content from the agent's logic.
- Capability inventory: The skill has the capability to execute the
dstackCLI tool to initialize audits, add corrections, and finish plans (SKILL.md, Steps 1, 3, 6). - Sanitization: There is no evidence of sanitization or filtering for the instructions that might be embedded within the processed code or documentation.
Audit Metadata