makepad-platform
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill metadata (description field) begins with 'CRITICAL:', a marker often used to override agent dispatch logic or force priority. While functionally a trigger here, it utilizes patterns associated with instruction overrides.
- [INDIRECT_PROMPT_INJECTION]: The instructions mandate that the agent read and incorporate content from an external file path (
./references/platform-support.md) before answering. - Ingestion points: The agent is directed to read the file
./references/platform-support.mdat runtime. - Boundary markers: Absent. There are no instructions to treat the external file content as untrusted or to use delimiters to prevent embedded instructions from being executed by the model.
- Capability inventory: The skill provides Rust code generation and suggests a synchronization command (
/sync-crate-skills) to the user. - Sanitization: Absent. The content from the external reference is incorporated directly into the agent's context without filtering or validation.
Audit Metadata