Six AI agent personas, each with a defined role, a set of skill files, and strict handoff contracts. All six are orches…
Six AI agent personas, each with a defined role, a set of skill files, and strict handoff contracts. All six are orchestrated via a **deterministic LangGraph pipeline** inside the `SEagenthub` MCP server — no LLM routing, no API key, no hook scripts. | Agent | Role | Activated By | |---|---|---| | `@techLead` | Orchestrator — decomposes goals, delegates, audits results | `@techLead`, `INIT_PROJECT`, `DELEGATE`, `AUDIT_RESULT`, `CHANGE_REQUEST` | | `@architect` | Infrastructure designer — CDK, IAM, observability, cost | `DELEGATE [architect]` | | `@codeCrafter` | Implementation — business logic, UI, resilience patterns | `DELEGATE [codeCrafter]`, `Cleared for implementation` | | `@codeReviewer` | Quality gatekeeper — complexity, naming, CVEs, docs | `Handing off to @codeReviewer` (automatic) | | `@qualityGuard` | Testing & security — unit / integration / load / pen testing | `Handing off to @qualityGuard` (automatic) | | `@devOps` | Deployment — CI/CD, environment promotion, verification | automatic after `AUDIT_RESULT` | The **signal phrases** defined in each skill file's OUTPUT CONTRACT are the transition conditions LangGraph evaluates at every node boundary. When used as prompt-driven agents in an IDE (Copilot, Claude Code), the same phrases drive the conversational handoff — the contract is identical in both layers.
oneofakindtech0
mcp
free
Others in the same category, ranked by how often they are opened.
