audit-security

target = $ARGUMENTS

If target provided, audit that path. Otherwise, files changed since the default branch. Full-codebase audit requires explicit user request.

Find and fix vulnerabilities where untrusted input or weak controls let an attacker change behavior, exfiltrate data, or escalate privilege. Tag each fix with a CWE so the change is traceable.

The patterns the model under-weights without prompting: shell or subprocess called with string args instead of array (command injection); SQL via string concatenation, template literals, or ORM escape hatches (.raw(), .extra(), RawSQL) — parameterized queries are non-negotiable; deserialization of untrusted input (pickle, yaml.load, JSON-into-class hydrators) and XML parsers at default settings (XXE); template engines or DOM APIs rendering user input unescaped (dangerouslySetInnerHTML, innerHTML, unescaped template vars); IDOR — endpoints fetching a resource by user-supplied ID without an ownership check; JWT misuse — alg:none accepted, key/algorithm confusion, missing exp/iss/aud validation; weak primitives — md5/sha1 for passwords, Math.random for tokens, ECB mode, missing salt; secrets in source, logs, or error responses; trust-boundary gaps — server relying on client-side validation, parsers invoked on raw input without size/depth limits; SSRF / path traversal where user input becomes a URL or filesystem path without allowlist; LLM output trusted as code, SQL, or shell (insecure output handling).

Filter aggressively before acting. Not a vulnerability: server-controlled config (env vars, constants, settings.*), framework-mitigated sinks (React {x}, parameterized ORM, prepared statements), UUID identifiers, or client-side-only validation when a server check also exists. Fixes need a concrete attack path from an untrusted source to a sink — theoretical "could be vulnerable if" is noise.

Apply the fix directly when the safe pattern is already established in the codebase or by the framework: parameterize the query, switch subprocess to array form, escape the rendered output, add the ownership check, swap md5 for argon2, scrub the secret from logs. Run tests after the batch; report git diff --stat and per-fix the CWE, the attack path closed, and the remediation. Auth/session/crypto changes that rotate secrets, migrate algorithms, or change token format are a separate decision — sketch the change and surface for sign-off, since the migration affects active sessions and stored credentials. Hard-coded secrets discovered in source: stop, surface immediately, and do not commit a "fix" that only deletes the line — the secret is already in history and needs rotation.