crossing-the-chasm

Warn

Audited by Gen Agent Trust Hub on Mar 21, 2026

Risk Level: MEDIUMPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Presence of deceptive and contradictory instructions in the main instructional file (SKILL.md). The skill is defined as a technology adoption framework, but multiple sections (Prerequisites, Instructions, Error Handling) contain logic specifically for managing 'logging configuration' and 'authentication' via 'CLI tools'. This represents an attempt to redefine agent behavior for tasks unrelated to the skill's stated intent.\n- [PROMPT_INJECTION]: Indirect Prompt Injection vulnerability surface. The skill ingests untrusted user content for strategic analysis without defining delimiters.\n
  • Ingestion points: Contextual analysis of product lifecycles and market segments within SKILL.md.\n
  • Boundary markers: Absent.\n
  • Capability inventory: Authorized to use file system tools (Read, Glob, Grep).\n
  • Sanitization: Absent; no instructions provided to sanitize or ignore embedded control sequences in analyzed data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 21, 2026, 12:38 AM