breaking-change
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions specify the use of the
oasdiffCLI, suggesting installation viago install github.com/oasdiff/oasdiff@latestor from its official GitHub releases. This involves downloading and executing a third-party binary provided by the oasdiff project. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted OpenAPI specification files to detect changes. The output of the
oasdifftool is then interpreted by the agent, which could be exploited if malicious content is embedded in the spec files. - Ingestion points: OpenAPI spec files (e.g.,
baseline.yaml,candidate.yaml) processed as described in Step 1 and 2 ofSKILL.md. - Boundary markers: Absent. There are no specific delimiters used to separate the tool's output from the agent's internal instructions when generating the final report.
- Capability inventory: The agent executes the
oasdiffcommand line tool on the local filesystem. - Sanitization: The skill mentions
oasdiff validateas a best practice in the text but does not include it as a required step in the automated execution flow to sanitize inputs before comparison.
Audit Metadata