The Agent Skills Directory

[SAFE]: The skill provides a structured analytical framework for reasoning verification. No malicious patterns were identified across the threat categories.- [DATA_EXFILTRATION]: The skill configuration limits tool access to Read, Grep, and Glob. No network-capable tools are authorized, and no patterns for accessing sensitive files or hardcoded credentials were detected.- [PROMPT_INJECTION]: The instructions do not attempt to bypass safety filters, extract system prompts, or override agent constraints. The language used is strictly task-oriented.- [REMOTE_CODE_EXECUTION]: There are no external package requirements, remote script downloads, or dynamic code execution patterns found in the skill.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted problem statements and solutions. 1. Ingestion points: Problem and solution text are ingested into the context (SKILL.md). 2. Boundary markers: Present as markdown headers (e.g., ## Original Conditions). 3. Capability inventory: No high-risk capabilities such as network access, subprocess execution, or file writing are present. 4. Sanitization: Absent; the skill relies on the agent's analytical logic to process input. The risk is low and inherent to the analytical nature of the skill.

reasoning-verifier