EVOKORE-MCP Documentation Portal
EVOKORE-MCP is a TypeScript-based MCP router and multi-server aggregator. It gives an AI client a single MCP endpoint that combines EVOKORE's native workflow tools with proxied child servers, while adding namespace isolation, dynamic tool discovery, role-based permissions, rate limiting, and human-in-the-loop approval. This portal is the entry point for everything documented about that system: setup, day-to-day usage, runtime architecture, the orchestration and review frameworks, and the cross-cutting reading paths for new users, engineers, and decision-makers.
Quick start
- New to EVOKORE? Read Executive Summary for a non-technical overview, then Setup for install and client registration.
- Day-to-day operation: Usage and Use cases and walkthroughs.
- Engineer / integrator: Technical Analysis and Architecture.
- Bringing it into a real environment for the first time: New User Brutal Honesty Kit — sharp-edges checklist before you commit to anything.
Reference (alphabetical)
| Document | What it covers |
|---|---|
| All Skills Crib Sheet | Flat lookup table of every skill in the library with one-line descriptions and wiki links. |
| Architecture | Runtime shape, layers, tool populations, discovery modes, startup lifecycle, RBAC, rate limiting, session/manifest state. |
| Architecture: AEP System | The Align-Execute-Prove engineering cycle: phase planning, ADR/CONTEXT.md discipline, test matrix, evaluation harness, verification log. |
| CLI integration | Registering EVOKORE with Claude Code, Gemini CLI, Cursor, Copilot CLI, Codex CLI, and status-line surfaces. |
| Executive summary | Plain-English overview of what EVOKORE is and who it is for. |
| HTTP deployment | Running EVOKORE over the MCP StreamableHTTP transport for remote and multi-client access. |
| Migration v2 to v3 | What changed between v2 and v3, breaking-change notes, and the upgrade procedure. |
| New User BHK | Brutal-honesty kit with severity-scored sharp edges and gotchas for new operators. |
| OAuth setup | Bearer-token / JWT authentication with JWKS rotation on the HTTP transport. |
| Panel of Experts | The multi-persona panel framework — 10-step cycle, challenge depths, persona schema, persistence, worked examples. |
| Plugin authoring | Writing and shipping plugins that add custom tools and resources, including hot-reload. |
| Setup | Install, build, env variables, mcp.config.json, RBAC, rate limiting, validation. |
| Skills overview | Narrative tour of all 16 active skill categories with code / non-code workflow tags. |
| Technical analysis | Engineer-facing deep dive: runtime layers, discovery model, continuity, security, deployment shapes. |
| Testing and validation | Validation surface, test layout, regression commands, release gating. |
| Tool discovery profiles | Named profiles for shaping the visible tool surface and their measured token budgets. |
| Tools and discovery | Native vs proxied tools, prefixing, discovery modes, benchmark contract. |
| Training and use cases | Pointer into the category-level training material. |
| Troubleshooting | Common failure modes and their fixes. |
| Usage | Day-to-day operation: tools, sessions, hooks, voice, dashboard, governance. |
| Use cases and walkthroughs | Concrete walkthroughs of representative operator scenarios. |
| Voice and hooks | The three voice surfaces (proxied ElevenLabs, VoiceMode, VoiceSidecar) and the hook pipeline. |
| Webhook envelope v1 | Authoritative payload spec for the v1 webhook envelope. |
| Webhook guide | End-to-end guide for configuring, signing, and consuming webhook events. |
Guides
Step-by-step operator guides live under guides/:
- Rate Limiting Guide — token-bucket configuration, per-server and per-tool limits.
- RBAC Guide —
admin,developer, andreadonlyroles, per-tool overrides, and the resolution order. - Repo Audit Hook Guide — pre-session checks for branch drift, worktree pressure, and control-plane drift.
- Voice Sidecar Guide — running the WebSocket voice runtime, persona hot-reload, hook wiring.
Category narratives
Per-category narrative tours of the skill library live under categories/:
- Anthropic Cookbook
- Automation and Productivity
- Awesome Claude Code Resources
- Developer Tools
- General Coding Workflows
- Hive Framework
- MCP Wrappers
- Official MCP Servers
- Research and Content
- wshobson Plugins
For a higher-level narrative across all 16 active categories, see Skills Overview. For a flat lookup table of every SKILL.md, see All Skills Crib Sheet.
Wiki
A rendered, browsable per-skill wiki is generated as a build artifact. To produce or refresh it, run npm run wiki:build from the repo root. The output appears at wiki/skills-index.html and is intended to be opened locally in a browser. The wiki is not committed to the repository because it is fully derived from the SKILLS/**/SKILL.md corpus.
Presentations
A small set of standalone, self-contained presentation pages ships under presentations/. Open presentations/index.html in a browser for the executive summary and the technical analysis as slide decks. The presentations are inline HTML with no external CDN dependencies, so they work fully offline.
Reading paths
Non-technical reviewer
New operator
Engineer or integrator
Orchestration and review
- Panel of Experts
- Architecture: AEP System
- Usage (orchestration sections)
Platform integrator
Last verified: 2026-05-20