electrobun-sdlc
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a multi-step agent orchestration pipeline which is susceptible to injection sequences where malicious instructions in a feature request could influence downstream code generation agents.
- Ingestion points: Untrusted data enters the agent context through the
/electrobun-sdlc <feature description>command trigger inSKILL.md. - Boundary markers: The handoff protocol lacks explicit delimiters or instructions for the orchestrator to sanitize or isolate embedded instructions within the documents passed between stages.
- Capability inventory: The pipeline includes agents (
electrobun-ui-agent,electrobun-backend-agent,electrobun-alignment-agent) that have permission to modify source code insrc/, configuration files likeelectrobun.config.ts, and test suites. - Sanitization: The instructions explicitly specify that documents (Research Reports, Architecture Specs, etc.) are passed "verbatim" between agents, which facilitates the propagation of potential injection payloads through the entire SDLC chain.
Audit Metadata