Reusable Claude Code baseline with safety hooks, audit logging, automatic formatting, cost controls, specialist agents, and tested workflows.
Agent Claude Code
Rewrites code for human readability using clear naming, decomposed complexity, and consistent structure. Applies best practices to make code immediately understandable. Does not change behavior.
Agent Claude Code
Reviews code changes, files, or diffs for correctness, security, style, and performance. Reports findings grouped by severity. Read-only — never modifies files.
Agent Claude Code
Diagnoses bugs through root cause analysis, stack trace reading, and error pattern matching. Proposes targeted fixes without changing unrelated code.
Agent Claude Code
Scans project dependency files for outdated, vulnerable, or abandoned packages. Reports findings with safe upgrade paths. Never installs, upgrades, or modifies lockfiles.
Agent Claude Code
Generates and updates code documentation — TSDoc/JSDoc comments, module-level docs, and README sections. Matches existing documentation style in the project.
Agent Claude Code
Plans database schema migrations, breaking API changes, and major dependency version bumps. Produces a phased plan with rollback strategies. Read-only — never modifies files or schemas.
Agent Claude Code
Default agent for all software engineering tasks in this project. Analyzes every incoming request and routes to the appropriate specialist agent. Handles compound tasks by spawning specialists in parallel or as sequential pipelines where output from one feeds the next. Always delegates — never implements directly.
Agent Claude Code
Reviews code for performance anti-patterns — N+1 queries, unnecessary re-renders, unoptimized loops, missing pagination, and expensive operations in hot paths. Read-only analysis.
Agent Claude Code
Refactors a target file or function for structure, naming, and maintainability without changing behavior. Enforces project naming conventions and engineering constraints.
Agent Claude Code
Scans for secrets, injection vulnerabilities, dependency CVEs, and authentication gaps. Bash is limited to audit commands only — no installs, no modifications.
Agent Claude Code
Generates unit and integration tests for a target file, function, or module. Detects the project test framework and matches existing test style. Runs tests to verify after writing.