This isn't claude_code, its source, or anything like that - just helper code that I use with claude code that I want to share.
phobologic/claude_code_helpers
Agent
Verifies that an implementation satisfies its ticket's acceptance criteria. Reads the ticket, reads the diff, and returns PASS or FAIL with specifics. Does not evaluate code quality, style, or anything beyond whether the AC are met.
phobologic/claude_code_helpers
Agent
Adversarial code reviewer. Assumes the code has problems and finds evidence. Covers logic, security, performance, and reliability in a single pass with a discriminator mindset.
phobologic/claude_code_helpers
Agent
Reviews code for logical errors, best practices, and architecture issues.
phobologic/claude_code_helpers
Agent
Reviews code for performance issues, efficiency, and resource usage.
phobologic/claude_code_helpers
Agent
Reviews code for structural problems — duplication, dead code, broken abstractions, and consistency violations.
phobologic/claude_code_helpers
Agent
Creative UI/UX designer on a design sprint team. Produces design proposals across three rounds of increasing concreteness — mood/direction in Round 1, system in Round 2, spec-ready values in Round 3. Three instances of this agent run in parallel; the team lead supplies each with its designer number and the app context…
phobologic/claude_code_helpers
Agent
Persistent evaluator on a design sprint team. Scores three designer proposals across three rounds against weighted criteria, issues shared briefs between rounds, and writes the final design specification synthesis after Round 3.
phobologic/claude_code_helpers
Agent
Implements a single tk ticket. Reads the ticket, gathers context, writes code, writes tests, ensures tests pass, and commits. Does not self-review or verify its own acceptance criteria.
phobologic/claude_code_helpers
Agent
Adversarial code reviewer on an agent team. Reviews changes that have already passed AC verification. Focuses on correctness, security, reliability, and performance. Creates tk tickets for findings.
phobologic/claude_code_helpers
Agent
Aggregates and organizes code review findings from multiple reviewers.
phobologic/claude_code_helpers
Agent
Reviews code for security vulnerabilities, risks, and defensive coding practices.
phobologic/claude_code_helpers
Agent
Adversarial reviewer for implementation plans and specs. Evaluates whether a plan will actually work when handed to independent implementer agents. Focuses on testability of ACs, gaps between tickets, dependency correctness, and self-containment.