raise-ambition
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an analysis pipeline that ingests untrusted user content from the conversation or existing work products to 'raise ambition'.
- Ingestion points: The skill instructions in
SKILL.mdexplicitly direct the agent to 'Start from the active request and anything already produced'. - Boundary markers: There are no instructions for the agent to use delimiters or ignore embedded instructions within the user's task description.
- Capability inventory: The skill encourages the use of 'tools you can install or run', 'database', 'statistical model', 'profiler', and 'benchmarks'.
- Sanitization: No sanitization or validation of the input task content is specified before the agent attempts to execute the 'ambitious' version of the task.
Audit Metadata