Quality

Blame shows who last changed each line in version-controlled files and when. Use it to trace the origin of code, comments, or regressions during debugging and review.

Bench defines how mlx-serve benchmarks are run and compared, using llmprobe and bench.sh with rules for same-methodology comparisons, variance, and performance claims. Use it before running benchmarks or citing numbers.

Check And Test runs lint, type checks, and the standard pytest tiers for Python and TypeScript projects. Use it to validate changes, investigate failures, and fix application bugs instead of weakening tests.

Code Review performs a rigorous review of uncommitted changes for correctness, security, architecture, and best practices. It reports findings and a proposed change plan before any edits are made.

Code Reviewer runs lightweight checks on saved, modified, or committed code to catch style issues, bugs, and basic security patterns. It is useful for quick feedback before commits or when reviewing diffs.

Dependency Auditor checks package manifests and lockfiles for known vulnerabilities before deployments or after dependency changes. It runs audits like npm audit or pip-audit and suggests fixes and breaking-change risks.

Secret Scanner detects exposed secrets, API keys, credentials, and tokens in code. Use it before commits, on file saves, or during security checks to block accidental leaks.

Security Auditor scans code and dependencies for common vulnerabilities like SQL injection, XSS, auth issues, and exposed secrets. Use it when reviewing changes, before deployments, or after security-related edits.

Test Generator suggests basic tests for new functions and components when code is added or testing is mentioned. It creates starter cases in Jest, Vitest, or Pytest and leaves comprehensive suites to the test-engineer sub-agent.

Check runs a pre-publish quality gate on marketing content to flag hallucinations, unsupported claims, brand-voice mismatches, structure issues, and AI asset provenance problems. Use it before publishing copy when you need a pass/warn/blocked report without editing the content.

CTF Binary Exploitation covers buffer overflows, format strings, heap exploitation, ROP, ret2libc, shellcode, seccomp bypass, and sandbox escape for pwn-style CTF challenges. Use it when solving binary exploitation tasks.

CTF Cryptography provides quick references for cryptography and math-based CTF challenges, including RSA, AES, ECC, lattices, PRNGs, ZKP, and related attack patterns. Use it when solving crypto puzzles or checking which technique applies.