crucible-emergent-resurrection

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill's 'Self-Evolving' feature creates a potential surface for indirect prompt injection by instructing the agent to modify its own core instruction file (SKILL.md) based on external data.
  • Ingestion points: The agent is instructed to read 'failed campaign' records from the references/archive/ directory and the evolution.jsonl ledger.
  • Boundary markers: The skill lacks explicit delimiters or specific warnings to the agent to disregard potential instructions or commands embedded within the campaign files it processes.
  • Capability inventory: The agent is granted Write and Edit tools and is explicitly directed to use them to update its own taxonomy within the SKILL.md file.
  • Sanitization: There is no evidence of sanitization or validation logic to ensure that content extracted from the archives is treated as inert data before being incorporated into the skill's instructional framework.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 01:40 AM