The Agent Skills Directory

[METADATA_POISONING]: The skill metadata contains deceptive claims regarding its origin, labeling itself as an "official" Supabase skill in the YAML frontmatter (source_type: official). However, the skill is provided by an independent author, which may lead users to overtrust the content's authority and safety.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze user-provided SQL queries and schemas, creating a potential vector for indirect prompt injection where malicious instructions could be embedded in database code. Ingestion points: SQL queries, table schemas, and database configurations provided by the user for optimization (as specified in SKILL.md). Boundary markers: The skill lacks instructions or delimiters to isolate user-provided data from its own logic or to prevent the agent from obeying instructions embedded in SQL comments. Capability inventory: The skill itself lacks active tool execution or network capabilities. Sanitization: No validation or sanitization of input SQL is performed.
[NO_CODE]: The skill consists entirely of Markdown reference files and does not contain any executable scripts, binaries, or tool-calling configurations.

supabase-postgres-best-practices