- Home
- Skills
- Agents & Orchestration
- Project Nexus: Multi-Agent Scoping
Works with the AI Agents you already use
Project Nexus: Multi-Agent Scoping
Transform raw client briefs into technical PRDs through a multi-agent debate between Sales, Engineering, and Product.
$9
Project Nexus: Multi-Agent Scoping
Example session with this skill installed
Scope a new real-time client analytics dashboard for 'FinTechFlow.com'. Sales promised it in 4 weeks for a major Q4 enterprise demo with live CSV/PDF exports, custom RBAC permissions, and sub-second chart updates ($50k fixed budget), but our primary PostgreSQL instance is already running at 80% CPU load during peak hours.
- Read your context and instructions
- Compiled the project nexus
I have initiated the
Project Nexus Scoping War Room for FinTechFlow.com. First, I enriched the domain profile via Agensi tools (Series B FinTech, ~140 employees, high compliance requirements, AWS/PostgreSQL stack) and executed the 3-persona debate between Sales, Engineering, and Product.
Multi-Agent Debate Summary
| Persona | Core Arguments & Stress-Test Findings |
|---|---|
| Pass 1: Sales Lead | Deal is worth $50k upfront + $180k ARR expansion. Client needs a live "Real-Time Analytics" narrative for their Q4 board demo in 4 weeks. Dropping visual polish or export proof will kill the expansion deal. |
| Pass 2: Systems Engineer | CRITICAL BLOCKER: Primary PostgreSQL is at 80% load. True sub-second streaming queries against production tables will trigger a database outage. Full custom RBAC + PDF rendering engine cannot be safely built and penetration-tested in 4 weeks. |
| Pass 3: Staff PM (Arbiter) | We protect production stability while saving the Q4 demo. We will ship a Read-Replica + Redis Cached Dashboard (60-second refresh) instead of live WebSocket streaming, use existing workspace roles instead of net-new RBAC, and ship async CSV exports while deferring PDF generation to Phase 2. |
Scoped Constraints & Technical PRD
1. Executive Summary & Business Context
Core Problem Statement: FinTechFlow requires an enterprise analytics dashboard in 4 weeks for a Q4 demo without overloading an already strained (80% load) primary PostgreSQL database.
*
Target Commercial Outcome: Secure the $50k milestone and unlock $180k ARR expansion by proving high-speed analytics capabilities safely.
- Agreed Delivery Horizon: 4 Weeks (2 two-week sprints; hard code freeze at Day 24).
2. War Room Resolution Log (Sales vs. Engineering)
Conflict 1: Sub-second real-time queries (Sales) vs. 80% Postgres CPU saturation (Engineering) ->
PM Resolution: Implement a dedicated PostgreSQL Read Replica with a Redis caching layer (60s TTL) and a manual "Refresh Now" button for the live demo.
*
Conflict 2: Custom granular RBAC & branded PDF engine in 4 weeks (Sales) vs. Timeline/Security risk (Engineering) ->
PM Resolution: Inherit existing Admin vs. Viewer JWT claims for Sprint 1; ship CSV export now and move custom PDF templating to Phase 2.
3. Non-Negotiable Scoped Constraints
| Category | Scoped Constraint | Trade-Off / Rationale |
|---|---|---|
| Architecture & Stack | Zero direct analytics queries on Primary DB; all reads routed to Read Replica + Redis. | Prevents production downtime during peak trading hours. |
| Data & Security | Enforce tenant isolation via existing JWT org_id middleware; no net-new permission tables. | Saves 1.5 weeks of auth engineering and security auditing. |
| API & Rate Limits | Dashboard API rate-limited to 12 requests/min per user; exports capped at 50,000 rows. | Protects memory overhead on backend workers. |
| Timeline Cut-Off | Hard scope lock after Day 5; final 4 days reserved strictly for load testing and demo rehearsal. | Guarantees reliable demo environment for Sales. |
4. Feature Prioritization Matrix
- Sprint 1 (MVP — Must Ship by Week 4):
- Read-replica pipeline + Redis 60s cached aggregation endpoints.
- 4 core KPI visualizations (Transaction Volume, Latency, Settlement Rate, Anomaly Count).
- Date-range filtering (24h, 7d, 30d) + Async CSV export.
- Phase 2 (Fast-Follow — Weeks 5–8):
- Automated branded PDF report generation.
- Custom field-level RBAC permission builder.
- Out of Scope (Explicitly Excluded):
- Sub-second WebSocket raw event streaming.
- Custom SQL query builder for end-users.
5. Testable Acceptance Criteria
- Dashboard endpoints return p95 response times
< 250msunder simulated 100 concurrent demo users. - Primary PostgreSQL CPU utilization increases by
0%during dashboard load testing (100% read-replica routing verified). - CSV export for 50,000 transaction records completes within
< 5 secondswithout blocking the main event loop. - Cross-tenant data leakage test passes with
100%isolation across mockorg_idtokens.
Recommended Next Steps
Generate Database Migration & Redis Config: Ask me to scaffold the read-replica connection pool and Redis caching middleware.
2. Export Linear / Jira Tickets: Ask me to break Sprint 1 down into story-pointed engineering tasks.
Connects securely to your tools. The creator never sees your data.
About this skill
Turn messy client briefs, discovery notes, and feature requests into scoped technical constraints and structured PRDs inside Cursor, Claude Code, or Codex.
What this skill does for you: Instead of generating generic one-pass specs, Project Nexus simulates a 3-agent war room directly in your coding agent:
- Sales Persona: Extracts commercial goals, client promises, and timeline pressure.
- Engineering Persona: Stress-tests the brief for architectural bottlenecks, API limits, security risks, and tech debt.
- PM Arbiter Persona: Resolves conflicts between Sales and Engineering to lock down an achievable MVP.
What you get on every run:
- Optional client company enrichment (via Agensi's built-in Apollo/Hunter tools) when a client domain is provided.
- A transparent Sales vs. Engineering debate log highlighting hidden project risks.
- A Non-Negotiable Constraints Table covering stack, security, and delivery boundaries.
- A prioritized MVP vs. Phase 2 vs. Out-of-Scope feature matrix and testable acceptance criteria.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
- Client domain enrichment (via Apollo, Hunter, or ContactOut) is optional and requires approving Agensi tool calls; if skipped or unavailable, the 3-agent debate runs directly on the text provided in your brief.
- Architectural constraint accuracy scales with input detail—providing your current tech stack, database load, or timeline in the prompt produces much sharper engineering trade-offs than a single-sentence idea.
- Uses the bundled
./references/prd_template.mdfile for formatting; keep the skill's folder structure intact when running locally.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
Nothing to install. It runs in your AI agent.
Agensi enrichment tools (Apollo, Hunter, ContactOut) for client domain lookups. Runs directly in your AI agent with zero setup. Optionally uses Agensi's built-in enrichment tools when a client domain is provided.
Permissions
File Scopes
Read and Write Files allow the agent to read the bundled PRD template and save generated specs. Network Access is used only for optional client domain enrichment.