allure-agent-mode
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it requires the agent to ingest and act upon test artifacts generated during runtime, which may contain data from untrusted sources.
- Ingestion points: Instructions in
SKILL.md(Workflow step 8) direct the agent to review various manifest files (run.json,test-events.jsonl,tests.jsonl,findings.jsonl) and test markdown files. - Boundary markers: The skill does not provide explicit delimiters or instructions for the agent to treat these specific manifest contents as non-instructional data.
- Capability inventory: The skill uses
npx allure agent,allure agent inspect, andnpm test, which involve shell command execution based on the analyzed environment. - Sanitization: The
references/allure-evidence.mdfile provides extensive guidance on redacting sensitive data (credentials, API keys, etc.) from reports, which effectively mitigates data exposure but does not address instruction injection within the evidence. - [EXTERNAL_DOWNLOADS]: The skill utilizes
npx allure agent, which downloads and executes the Allure CLI from the npm registry. This is the primary functional component of the skill and originates from the vendor's own software ecosystem.
Audit Metadata