postgres-schema-auditlisted

# Postgres Schema Audit ultrathink <!-- anthril-output-directive --> > **Output path directive (canonical — overrides in-body references).** > All file outputs from this skill MUST be written under `.anthril/audits/postgres-schema-audit/`. > Run `mkdir -p .anthril/audits/postgres-schema-audit` before the first `Write` call. > Primary artefact: `.anthril/audits/postgres-schema-audit/<artefact>`. > Do NOT write to the project root or to bare filenames at cwd. > Lifestyle plugins are exempt from this convention — this skill is not lifestyle. ## Before You Start 1. **Detect available connection modes.** Run `scripts/check-connection.sh` to discover whether the user has a Supabase MCP connector configured, a direct-postgres profile configured, both, or neither. The skill's entire Phase 1 behaviour depends on this result. 2. **Do not invent project, connection, or schema names.** Every identifier the skill uses in queries must come from a real MCP response or a real `pg_namespace` row. Never synthesise names to fill gaps. 3. **Read-only guarantee.** Every SQL call in this audit is SELECT-only against `pg_catalog`, `information_schema`, and (rarely, for data-shape sampling) the target schema's tables with `LIMIT 100`. The skill never runs INSERT, UPDATE, DELETE, ALTER, CREATE, or DROP. Remediation is emitted as commented SQL the user applies themselves. 4. **Load `.db-design-ignore`.** If the current working directory contains this file, parse it and treat entries as suppression