Collect relevant evidence from supplied public sources and return source-linked snippets that agents can cite and verify.

Review a source-code diff for likely bugs, missing tests, compatibility problems, unsafe migrations, and security-relevant mistakes. Returns structured findings with file/line evidence and confidence.

Normalize, trim, validate, and deduplicate tabular CSV or JSON data for reliable agent workflows.

Inspect dependency manifests for missing versions, risky install scripts, unknown licenses, and policy issues without claiming to replace a live CVE database.

Extract structured fields and source evidence from supported documents using task-specific profiles such as tenders, invoices, contract dates, and privacy policies. Returns schema-validated JSON with confidence and page/source references.

Extract Markdown, CSV, and plain-text tables into clean structured rows for agents and business workflows.

Check invoices, contracts, tenders, and text documents for required fields and source evidence before downstream automation.

Compare invoice and receipt text for matching document numbers and totals with a configurable amount tolerance.

Analyze Postgres, Supabase Postgres, or Prisma/Postgres migration SQL for destructive operations, lock/data-loss risk, backwards-compatibility hazards, rollback issues, and safer sequencing. Returns structured findings with evidence and confidence.

Audit an OpenAPI document for complete schemas, useful examples, stable operation IDs, clear descriptions, and payment/authentication discoverability.

Detect and redact common personal data and credentials from text, JSON, and source code without sending content to a model.

Compare two webpage versions and return only meaningful content changes as structured JSON. Filters cosmetic DOM churn, tracking IDs, timestamps, and other noise.

Fetch a public webpage and return compact agent-ready Markdown, canonical metadata, tables, and important links while removing navigation, scripts, cookie UI, and layout clutter.

Check public HTML for common accessibility failures including missing image text, labels, page language, headings, and link names.

Fetch a public webpage, normalize its readable text, and report whether it changed against a caller-supplied baseline.