Review Reliability — Development agent for Claude Code
R3 Reliability reviewer — behavior-first tests, coverage value, edge cases, determinism, contracts, and regressions.
How to install Review Reliability
Installs to ~/.claude/agents/labdrian-ai-labdrian-sdd-overlay-review-reliability.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/labdrian-ai/labdrian-sdd-overlay/HEAD/agents/review-reliability.md -o ~/.claude/agents/labdrian-ai-labdrian-sdd-overlay-review-reliability.md Restart Claude Code, or start a new session, for it to be picked up.
What Review Reliability does
name: review-reliability description: R3 Reliability reviewer — behavior-first tests, coverage value, edge cases, determinism, contracts, and regressions. model: sonnet effort: max tools: [], mcp__codegraph__codegraph_explore
R3 Reliability Review
Review once, return one result, and stop. Never edit, delegate, or expand scope.
Input
The task begins with GENTLE_AI_REVIEW_BINDING and its exact one-line JSON. Immediately after it, the parent supplies one block from GENTLE_AI_CLAUDE
Alternatives in Development
- PRD Reviewer — Review a PRD and flag missing acceptance criteria, edge cases, and 135 ★
- Unit Tester — Specialist in logic isolation and unit testing 93 ★
- Po Steward — Steward seat on the Atlas PO Council 79 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Verification Challenger
Independently challenge an implementation by tracing requirements to tests and evidence, checking regressions
Behavior Tester
Behavior-first tester (فاحص السلوك) — runs an application or code deliverable against an agreed behavior speci
Fk Verifier
Debugs correctness gate failures (numerical mismatch, non-determinism, edge cases, NaN) of a fast-kernel candi
Workflow Discussion Review
Periodically reviews a discussion file for gaps, shallow coverage, and missing edge cases. Invoked in the back
Edge Case Adversary
Independently brainstorm the edge cases a builder likely MISSED for a given deliverable, countering the "bound
Review Reliability.Agent
Read-only reliability reviewer. Flags missing error-path tests, non-deterministic behavior, and absent input v
Related Skills
Adversarial Test
Post-build adversarial review — a fresh agent stress-tests recent changes for bugs, regressions, conflicts, an
Definition Of Done Rules
Always-on Definition of Done. Applied to every behavior change. Five hard criteria plus two opt-in -- review,
React Test
Enforce TDD workflow for React. Write React Testing Library tests first (behavior-focused, accessibility-first