grok-bot` emits Grok Bot / Sand skills that use the `user-obsidian-second-brain` MCPserver for vault I/O), `pi` emits a Pi package (`.pi/prompts/` + `.pi/skills/`), and `gemini-cli` / `opencode
vsearch|get|multi-get`** — CLI fallback for one-off shell checks or when the MCPserver is unavailable. Always pass `--index `, where ` ` is the `qmd_index` field from `vault-manifest.json` when
projects: `Unity-MCP-Plugin/` (Unity Editor/Runtime plugin), `cli/`, `Installer/`, `Unity-Tests/`.
The MCPserver is NOT in this repo — the plugin consumes the shared [GameDev-MCP-Server](https://github.com/IvanMurzak/GameDev-MCP-Server)
framework that registers and executes MCP tools, prompts, and resources. Includes a self-hosted MCPserver manager and auto-configuration for AI clients.
## Build / run
- **Open**: `Unity-MCP-Plugin` folder
code blocks in `README.md` and `api.md`. |
| `./scripts/test` | Runs the suite. Starts a Steady mock server on port 4010 unless `TEST_API_BASE_URL` is set. |
### Test matrix
- **Python and Pydantic
plugin/plugin.json` (Codex), and `AGENTS.md` (every host's resolver).
- Paired with one universal NotFair MCPserver for Google, Meta, X, LinkedIn, Reddit, and TikTok Ads plus Google Search Console, Google
scheduler in `src/server/scheduler/tick.ts` sweeps
`goals.next_tick_at`.
- **MCP**: NotFair exposes two internal MCPservers as Next routes —
`notfair-goals` (`/api/mcp/goals`, the goal tools) and `notfair-browser`
(`/api/mcp/browser`). External catalog tokens
full list.
3. If the `sage_inception` tool is present in your available MCP tools, load `core/sage/CLAUDE.md` (persistent-memory workflow). If absent, SAGE is not installed — skip silently
reading issues, applying fixes, and submitting PRs.
## MCP Tools
You have a `gateway` MCPserver connected that provides GitHub tools (prefixed `mcp__gateway__GitHubMCP___`). Use them directly — no manual HTTP
query.
> **Keep the connection alive during indexing.** Indexing runs in the background — the MCPserver continues working even when not actively responding to tool calls. However, some MCP hosts
list based on the skill's purpose.
**Qualify MCP tool names with their server.** The `mcp__context7__query-docs` form above is Claude Code's; other harnesses expect `ServerName:tool
interface** (`fli/cli/`) - Typer-based command line tool with `flights` and `dates` commands
- **MCPserver** (`fli/mcp/`) - Model Context Protocol server for AI assistant integration
- **Core utilities** (`fli/core/`) - Shared parsing and building
needed
- Limit tool search to maximum 2 searches per turn to avoid overhead
- Essential MCPservers (nogic, code-graph) are never disabled
- Non-essential servers (EXA, Apify, Playwright) load
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
Where does it go?
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
What should it contain?
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
CLAUDE.md or AGENTS.md?
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.