MCP server for Vetmanager veterinary clinic management — 101 tools, bearer auth, self-hosted
Agent Claude Code
Aggregates findings from specialized reviewers and Spark scout candidates — validates untrusted leads, dedupes, sorts by severity × confidence, produces final markdown report with verdict and top-N for cross-CLI arbitration.
Agent Claude Code
Reviews system-level design — module boundaries, layer violations, cross-module duplication, fit with technical requirements. Supports two modes (local/system) via user prompt.
Agent Claude Code
Reviews Python code for readability, simplicity, dead code, local duplication, and naming. Does NOT cover architecture, security, performance, docs, or tests — those are other reviewers.
Agent Claude Code
Parallel reviewer that runs via the available Codex/GPT adapter with an anti-correlation prompt — uses Spark candidates as untrusted leads and records the actual model used.
Agent Claude Code
Reviews documentation drift — PRD/Roadmap/AssumptionLog/README consistency with actual code, outdated assertions, broken references, contradictions between docs.
Agent Claude Code
Reviews logs, metrics, tracing — correlation IDs, secret leakage in logs, metric coverage (latency/error rate), exception readability, recoverability of request chain from logs.
Agent Claude Code
Reviews performance (latency, N+1, cache, indexes, memory) AND reliability (retry/backoff, timeouts, circuit breaker, idempotency, race conditions, partial failures, cleanup, graceful degradation).
Agent Claude Code
Reviews product fit — does implementation match business task from PRD, acceptance criteria coverage, UX quality, hidden breaking changes, consistency between PRD iterations.
Agent Claude Code
Reviews security — bearer/VM token lifecycle, input validation, SQLi, CSRF, rate limiting, secrets in git, info disclosure. Reliability (retry/timeout/circuit breaker) — other reviewer.
Agent Claude Code
Reviews changes for unnecessary complexity and proposes simpler alternatives. Evaluates readability, maintainability, obviousness. Complement to reviewer-code (local quality) and reviewer-architecture (system layering) — focuses on "is there a simpler way" at the solution-design level.
Agent Claude Code
Reviews tests — behavior over implementation, minimal mocking of internals, unhappy path coverage, boundary conditions, idempotency, serialization, real payload examples, fragility to harmless refactors.