jeremylongshore/tons-of-skills-marketplace / plugins/ai-agency/hyperflow/skills/hyperflow
hyperflow
FlaggedUse when applying Hyperflow's orchestration doctrine in Codex, Antigravity, Grok, or another single-agent surface. Auto-invoke for non-trivial engineering work: build, implement, add, refactor, debug, fix, review, audit, plan, scope, design, brainstorm, ship, or deploy. Trigger with /hyperflow:hyperflow, "use hyperflow", "apply the doctrine", or automatically on any task-shaped message.
reviewer-prompt.md
View on GitHub# Reviewer Prompt Template
Use this template when dispatching Reviewers via the Agent tool. Review depth scales by task complexity.
## Complexity Classification
The orchestrator determines complexity BEFORE dispatching the Reviewer:
- **Simple** (levels 1-2): Single file, rename, config, one-line fix
- **Medium** (levels 1-3): 2-3 files, modifies existing functionality, touches shared code
- **Complex** (levels 1-5): 4+ files, new feature, UI work, DB/API changes
## Template
The first line is a literal routing marker — emit it **verbatim** as the very first line of the dispatched prompt so an upstream tiering proxy can keep review passes on the top model. It is inert to the Reviewer.
```
hyperflow-role: reviewer
## Review scope
[Files changed, task assigned, complexity classification]
## Worker output
[Paste worker's summary]
## Level 1: Requirements
- Does the output match the task spec exactly?
- All sub-tasks completed? Nothing missing?
- Nothing extra added beyond the spec?
## Level 2: Code Quality
- Follows project naming conventions?
- No TypeScript `any`, no dead code?
- Uses existing utils/hooks (not reinventing)?
- Proper error handling, SRP, early returns?
## Level 3: Integration (medium + complex only)
- Imports resolve? No circular dependencies?
- Shared state/context not broken?
- API contracts preserved?
- Existing tests would still pass?
## Level 4: Performance & Security (complex only)
- No N+1 queries? Expensive ops memoized?
- No unnecessary re-renders?
- No hardcoded secrets (sk-*, AKIA*, ghp_*, private keys)?
- Input validation at boundaries? No injection vectors?
## Level 5: UX & Accessibility (complex UI tasks only)
- Aria labels on interactive elements?
- Keyboard navigation works?
- Loading/error/empty states handled?
- Responsive + RTL considered?
## Security Review (always)
- Were any blocked files accessed? (.env, *.pem, *.key, ~/.ssh/*)
- Any dangerous commands? (rm -rf, force push, sudo)
- Any data exfiltration? (contents piped to external URLs)
## Token economy (DOCTRINE rule 16)
Be specific and to the point. Return ONLY the Output format block below — no preamble ("I'll now review …"), no restating of the worker output or task spec, no narration of the review process, no postamble summary, no closing pleasantry. One-line summary per level; one short finding line per failure. Stop after the verdict.
## Output format
```
── Review ──────────────────────────────
L1 Requirements pass — [summary]
L2 Code Quality pass — [summary]
L3 Integration pass — [summary]
L4 Performance fail — [issue found]
L5 UX/A11y skipped — not applicable
────────────────────────────────────────
VERDICT: APPROVED | NEEDS_FIX | SECURITY_VIOLATION
[Issues per failed level — one line each]
[Notes for future tasks — omit if none]
```
```
If a finding needs a peer domain's judgment before you can rule on it, hold the verdict and emit
`CONSULT: <peer> — <question>` instead (you may consult any specialist in `agents/`). The Team Lead brokers the
answer and re-dispatches you to finish the review. See DOCTRINE rule 19 / [consultation.md](consultation.md).
## Dispatch Example
```
Agent({
description: "Review auth middleware (complex)",
prompt: `## Review scope
Files: src/middleware/auth.ts, src/middleware/auth.test.ts, src/types/auth.ts, src/types/session.ts
Task: Create JWT auth middleware with refresh logic
Complexity: Complex (4 files, new feature, security-sensitive)
## Worker output
1. Created auth middleware with RS256 verification
2. Added refresh token rotation
3. Tests cover valid/expired/malformed tokens
## Level 1: Requirements
- JWT validation with RS256? Refresh logic? Rate limiting?
## Level 2: Code Quality
- Follows conventions? Types correct? No any?
## Level 3: Integration
- Works with existing route handlers? Session types compatible?
## Level 4: Performance & Security
- No secrets hardcoded? Token validation safe? Timing attacks prevented?
## Level 5: UX & Accessibility
- Skipped (not a UI task)
## Security Review
- Blocked files? Secrets? Dangerous commands?
## Output format
── Review ──
pass / fail / skipped per level + VERDICT`
})
```
See [review-levels.md](review-levels.md) for full checklist details and failure handling.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |