How Agent Teams Actually Work: Lessons from Running a 6-Agent Organization
I run a six-agent organization.
soul · engineering · by toli
Your technical architect and engineering leader. Designs systems schema-first, writes specs that teams ship from on the first try, and coordinates specialist sub-agents across the full stack. **How it works:** - Starts every project with the data model - bad schemas radiate upward through the entire stack - Writes implementation specs with API shapes, section lists, and design system references - Coordinates frontend, backend, and design sub-agents through structured delegation - Reviews code across all layers using a 4-step protocol: code analysis, source verification, experiential testing, specialist review **Includes 5 specialized skills:** - **Architecture Review** - Evaluates system design, identifies bottlenecks, proposes improvements - **Spec Writer** - Produces implementation specs developers can ship from directly - **Code Review** - Multi-layer review protocol with 8-point checklist - **Incident Debugger** - Traces production issues from symptom to root cause - **Compound Engineer** - Nightly learning loop that improves over time Covers React, Next.js, Node.js, Python, Rust, PostgreSQL, Docker, Vercel, and more.
Price: $49
Downloads: 53 · Version: 1.3.0
Rating: No reviews yet
Tags: architecture, full-stack, code-review, specs, team-lead, typescript
API: GET /api/v2/items/the-cto
Technical architect and engineering leader. Schema-first, spec-driven development.
$49
Your technical architect and engineering leader. Designs systems schema-first, writes specs that teams ship from on the first try, and coordinates specialist sub-agents across the full stack.
How it works:
Includes 5 specialized skills:
Covers React, Next.js, Node.js, Python, Rust, PostgreSQL, Docker, Vercel, and more.
// security
Last scanned Mar 14, 2026
Screening language is intentionally conservative. Review the bundle contents and capability needs before installing.
// details
Soul Doctor audit: added explicit value hierarchy (security > correctness > speed > elegance), added recency markers to MEMORY.md, trimmed redundant Identity Refresh, updated README version
Soul Doctor audit: automated quality sweep - added authority tiers, operating modes, fallback chain, trust recovery, session learning, error handling, secret management, rate limit handling, messaging protocol, decision logging, thinking level guidance, context/compression strategy, session startup checklist, workspace layout, routing table, context handoff format, unavailable-agent fallback, energy patterns template, reversibility classification. Fixed em dash in BOOTSTRAP.md. Trimmed README under 3K.
Renamed compound-engineer to nightly-compound. Added Phase 5: RESTRUCTURE. BOOTSTRAP: QMD auto-setup, opt-in nightly compound. SOUL.md: universal rules added. CHANGELOG.md and README version badges.
Initial release.
// soul notes
Published notes from active creators attached to Gary the CTO.
I run a six-agent organization.
// package contents
// related
Nearby items, alternatives, and more from the same creator.