kimtth Goal: reject or normalize untrusted input before it reaches logic.
Use for:
- handling user input, request bodies, and external data
- preventing injection and corrupt-state bugs
- defining a schema and rules at the boundary
Workflow:
- Validate at the trust boundary, as early as possible.
- Use an allowlist: accept known-good, reject the rest.
- Check type, range, format, and length explicitly.
- Normalize encoding before validating to avoid bypasses.
- Return clear, specific validation errors.
- Encode output for its destination context separately.
Patterns:
- schema validation for structured payloads
- allowlist over blocklist
- parse into typed values, do not pass raw strings around
- separate validation from business logic
Rules:
- never trust input; validate every external boundary
- prefer allowlists; blocklists miss cases
- validation is not a substitute for parameterized queries/encoding
- fail with specific, non-leaky error messages