Agent Claude Code
General-purpose code reviewer that checks whether a change is correct, sane, appropriately scoped, and consistent with its repository. Use after implementation, before commits, or during PR review. Follows the diff into surrounding code only as needed to understand concrete effects and reports useful findings without…
Agent Claude Code
Reviews changed code for overengineering and premature structural decisions. Use after implementation or during PR review to find unnecessary code, weak data shapes, duplicated ownership, threaded signals, shared mutable state, scattered special cases, and machinery that an existing or smaller primitive can replace…
Agent Claude Code
Traces how a specific behavior works through control flow, data flow, state ownership, boundaries, and side effects with precise file:line evidence.
Agent Claude Code
Locates where a concern lives and returns the concrete files, precedents, primitives, tests, and validation commands that describe the codebase as it exists.
Agent Claude Code
Finds changed comments and docstrings that misstate behavior, preserve a false invariant, or create a concrete maintenance trap. Use when a PR adds or edits comments, API documentation, TODOs, examples, or operational notes. Verifies every finding against code and direct consumers; ignores harmless wording preferences…
Agent Claude Code
Finds repository documentation made false by a change and missing documentation required to use or operate a new public behavior. Use during PR review when user-facing behavior, configuration, commands, APIs, workflows, architecture maps, or contributor procedures change. Checks the documentation surfaces this…
Agent Claude Code
Researches and validates GPUI usage patterns, APIs, and conventions. Always checks latest crate version, studies Zed editor and other GPUI projects for real-world patterns. Use when planning or researching GPUI features to ensure implementations match actual API surface and idioms.
Agent Claude Code
Finds meaningful changed behavior that lacks regression protection. Use when reviewing a PR or completed implementation for test completeness and quality. Maps outcomes and invariants to existing unit, integration, and end-to-end tests, then reports only gaps with a plausible faulty implementation the proposed test…
Agent Claude Code
Diagnoses bugs, errors, stack traces, regressions, and unexplained behavior by reproducing the symptom, testing competing hypotheses, and proving the smallest causal chain and fix boundary. Advisory only — does not modify files, commit, or publish findings.
Agent Claude Code
Hunts for a missing type at a seam — structure flattened and rebuilt downstream, hand-maintained lists held together by KEEP IN SYNC comments, a phase that exists in the code but not in the type, a second route that skips the validator, an invariant carried by a comment or by a helper with a reachable bypass, a type…
Agent Claude Code
Finds changed failure paths that become indistinguishable from success or lose the evidence needed by the owner who can act. Use when a change adds catches, fallbacks, retries, optional operations, error translation, default values, or recovery behavior. Requires a reachable failure, a suppression point, and a…
Agent Claude Code
Researches current external facts, APIs, platform behavior, and technical claims from primary sources, returning cited evidence, applicability, conflicts, and unresolved gaps.