readme-i18n
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text from repository README files, creating a surface for indirect prompt injection attacks. Any instructions hidden within a source README could potentially influence the agent's behavior during the translation process.
- Ingestion points: The agent reads source
README.mdand existing localized siblings (e.g.,README.zh.md) from the user's repository (identified inSKILL.mdStep 1 and 2). - Boundary markers: The skill does not explicitly instruct the agent to ignore instructions embedded within the README text, although the workflow constraints and preservation checklist (referenced in
references/preservation-checklist.md) serve as a structural guardrail. - Capability inventory: The skill utilizes file system read and write capabilities to process and create README variants.
- Sanitization: There is no explicit sanitization or filtering of external prose before it is processed by the agent.
Audit Metadata