The Agent Skills Directory

[SAFE]: The skill functions as a pedagogical aid, providing analogies and explanations for technical concepts. No unauthorized command execution, persistence mechanisms, or data exfiltration patterns were identified.
[SAFE]: Recommended practices, such as instructing students to use .env.local for secret management, align with industry security standards and do not involve the agent accessing credentials itself.
[SAFE]: External service references to Vercel, Supabase, Neon, and GitHub involve well-known, reputable industry providers and are documented neutrally.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local project files and user-provided screenshots to personalize responses.
Ingestion points: The skill reads the CLAUDE.md file for student profiles and encourages the upload of screenshots for error troubleshooting.
Boundary markers: No explicit delimiters or boundary markers are defined for the data ingested from CLAUDE.md or user inputs.
Capability inventory: The agent is restricted to informational responses and delegating to a lookup subagent; it does not possess capabilities to execute arbitrary code or perform network exfiltration based on the ingested data.
Sanitization: No specific sanitization or validation routines for ingested file content are described in the instructions.

Teaching Assistant