skills/akillness/jeo-skills/utopia/Gen Agent Trust Hub

utopia

Pass

Audited by Gen Agent Trust Hub on Oct 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill manages a bitemporal knowledge application that ingests external documents and provides search/chat capabilities, creating a surface where malicious instructions embedded in processed data could influence the agent.
  • Ingestion points: Ingests external documents and knowledge base data for search, chat, and derivation tasks as defined in SKILL.md.
  • Boundary markers: Instructions mandate human adjudication for uncertain facts and agent-proposed write operations to maintain a "review boundary" (specified in SKILL.md).
  • Capability inventory: The agent has access to Bash, Read, Grep, and Glob tools for file and system interactions (SKILL.md).
  • Sanitization: While a review boundary is established, specific automated sanitization or escaping of ingested document content before model processing is not detailed.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to perform application management tasks, such as deployment and backups.
  • Evidence: SKILL.md and references/upstream.md mandate explicit approval gates for all destructive actions, database mutations, or service deployments.
  • Mitigation: The provided scripts/preflight.py helper facilitates safe, read-only integrity checks of the local environment using secure POSIX path traversal and hash verification before any operational commands are executed.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 30, 2026, 09:02 AM