nooa-context-and-state

Warn

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill documents a method for creating dynamic context blocks that evaluate Python expression strings during every LLM turn. This mechanism utilizes runtime evaluation to maintain live agent state.
  • Evidence: self.context.set_dynamic("progress", "self.format_project_state()") in SKILL.md.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of event history and context blocks into the agent's system prompt, creating a potential surface for indirect prompt injection if those sources contain untrusted data.
  • Ingestion points: self.context and agent.event_manager described in SKILL.md.
  • Boundary markers: The skill notes that blocks appear as "labelled SYSTEM sections."
  • Capability inventory: The event history includes "code executions" and "LLM calls."
  • Sanitization: The documentation does not specify sanitization or filtering logic for data stored in context blocks or event history.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 13, 2026, 08:15 AM
Security Audit — agent-trust-hub — nooa-context-and-state