skills/adeonir/agent-skills/app-flow/Gen Agent Trust Hub

app-flow

Pass

Audited by Gen Agent Trust Hub on Nov 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from product documentation (PRDs), user-provided prose, and project source code to extract navigation flows. This inherent processing of external data creates a potential attack surface for indirect prompt injection, where instructions embedded in those documents could attempt to manipulate the agent's logic. The skill provides specific countermeasures by instructing the agent to treat all intake data as claims rather than authoritative instructions and enforcing a confirmation step where the agent must summarize its interpretation for user approval before drafting.
  • Ingestion points: Project source code (routing files, screen lists), PRD files, and user chat input.
  • Boundary markers: Mandatory confirmation turns where the agent proposes its read of the situation for user redirection.
  • Capability inventory: The skill utilizes the Read, Write, and Edit tools to inspect codebases and maintain the FLOWS.md artifact.
  • Sanitization: The agent is instructed to validate all findings against a specific flow rubric and notation model.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 15, 2026, 03:38 AM