customise-workflow

Pass

Audited by Gen Agent Trust Hub on Aug 5, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local Python script located at skills/customise-workflow/script.py. This script is used for loading current configurations, validating user input, and writing the final settings to the project directory.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied answers to configure settings that influence the behavior of other AI agent plugins.
  • Ingestion points: User answers are collected via the AskUserQuestion tool during the curated brainstorming phase.
  • Boundary markers: The workflow includes a distinct validation phase using the validate-answer command to ensure inputs map to allowed configuration values.
  • Capability inventory: The skill has the capability to execute a local script via Bash and write to the filesystem (Write) at .atlas-ai/config/atlas.json.
  • Sanitization: The skill explicitly incorporates a validation step (validate-answer) that verifies user input against an allowed schema before any data is committed to the configuration file.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 5, 2026, 11:13 AM
Security Audit — agent-trust-hub — customise-workflow