dreamer
Pass
Audited by Gen Agent Trust Hub on Jun 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it handles external data during the context-gathering phase.
- Ingestion points: The agent reads all project documentation within the
docs/directory, the project manifest (CLAUDE.md), and the content of existing GitHub issues (titles, labels, and bodies) from the repository. - Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or isolate potentially malicious commands embedded within the fetched GitHub issue data.
- Capability inventory: The skill possesses the capability to execute shell commands (
ls,cat,find) and perform network operations by creating new issues via theghCLI. - Sanitization: No sanitization, filtering, or validation is performed on the data retrieved from external sources before it is interpolated into the agent's context for ideation.
- [PROMPT_INJECTION]: The skill employs role-playing instructions to define a specific persona ("The Dreamer in Darkness"), directing the agent to adopt an "ancient, unhurried" tone and explicitly stating it should be "not helpful" and "not distinguish" between brilliant and monstrous ideas, which imposes behavioral constraints that deviate from standard helpfulness guidelines.
Audit Metadata