security-basics

Security basics

Use this skill when

• Code under review or being written touches authentication, session handling, secrets, or credential material.
• Request handling, input parsing, header trust, or CORS/cookie behaviour is changing.
• A sensitive endpoint (admin, metrics, debug) is being added, exposed, or reworked.
• Logging or telemetry may incidentally capture credentials, tokens, or PII.

Do not use this skill when

• The task is writing a full threat model, pen-test plan, or formal compliance artefact.
• The change has no security-sensitive surface (pure UI copy, doc-only edits).
• A specialised skill already covers the narrower concern (for example, configuration-env for env/secret plumbing).

Inputs to gather

• The concrete diff or file set touching auth, request handling, or sensitive endpoints.
• How the service sits behind proxies or load balancers (for header trust decisions).
• Whether the environment is production or a lower tier (for CORS and cookie posture).

First move

• Identify every auth, secret, input-trust, endpoint-exposure, and logging touchpoint in the change.
• Run the review checklist below against each touchpoint before proposing fixes.

Outputs

• A concrete security review or fix list covering auth, secrets, input validation, endpoint exposure, CORS/cookies, and logging touchpoints.
• Checklist-backed confirmation that no hardcoded secrets, unsafe trust boundaries, leaked internals, or unprotected sensitive endpoints remain.
• Targeted validation evidence for authorization failures, invalid input handling, and scrubbed logs or responses.

Authentication & secrets

• Auth tokens, passwords, and API keys must come from environment variables or a secrets manager — never hardcode them.
• Do not log request headers containing auth, session cookies, or full URLs that may embed credentials.
• Fail fast at startup if required auth config is missing.

Input trust

• Trust nothing from client input without validation (schema, type, length, format).
• If behind a reverse proxy, trust X-Forwarded-For / X-Real-IP only when the proxy is known and controlled.
• Strip or overwrite forwarding headers at the proxy before forwarding to the app to prevent IP spoofing.
• Rate-limit by the real client IP, not by spoofable headers.

Sensitive endpoint exposure

• Admin, metrics, and debug endpoints must be protected: auth middleware, network ACL, or both.
• /health and public status endpoints can remain open but must not reveal internal state, versions, or stack traces.

CORS & cookies

• Set Allowed-Origins to explicit origins in production — never * when cookies or Authorization headers are in use.
• Enable secure cookies (SameSite=Strict + Secure flag) when the app is behind HTTPS.

Review checklist

Before merging code that touches auth, request handling, or data exposure:

• No hardcoded secrets in source code or test fixtures
• All inputs validated (shape, type, length, format)
• Error responses do not leak internals (stack traces, DB errors, file paths)
• CORS origins are explicitly restricted for production
• Sensitive endpoints are protected
• Structured logs are scrubbed of credentials and PII

Workflow

See the body and references for security and privacy guardrail steps.

Guardrails

• Never hardcode secrets, tokens, or API keys in source or test fixtures.
• Never trust spoofable request headers without a known, controlled proxy boundary.
• Never let error responses leak stack traces, DB errors, or internal file paths.
• Never default production CORS to * when cookies or Authorization headers are in play.
• Never log raw credentials, tokens, or PII — scrub at the structured-logging boundary.

Validation

• Run targeted tests or checks for the changed auth, request-handling, logging, or exposure surface.
• Confirm no secrets, tokens, credentials, private data, or sensitive endpoint outputs were added to logs or public responses.
• Confirm authorization failures, invalid input, and cross-origin/cookie behavior are covered where relevant.

Support files

• Read references/examples.md when you need concrete user utterances, expected behaviour, or a model answer shape to mirror.
• Read references/edge-cases.md when the request is a near miss, partially matches this skill, or the first attempt fails.