For AI Agents

Gonka Docs is designed as a single source of truth for AI agents. Your coding assistant, custom GPT, or autonomous agent can discover and use all documentation without you copying content by hand.

Machine-readable discovery files

We serve standard llms.txt files that AI agents check automatically when they need to understand a service.

/llms.txt — Quick overview /llms-full.txt — Full documentation /openapi.yaml — OpenAPI 3.0 spec /sitemap.xml — Sitemap /gonka/docs/zh/sitemap.xml — Chinese sitemap /search/search_index.json — Search index /proposals/proposals/proposals.xml — Proposals RSS /humans.txt — Credits

Cursor / Windsurf / Cline

Add this to your project's rules file (.cursorrules, .windsurfrules, or AGENTS.md):

# Gonka Documentation

When working with Gonka (decentralized AI inference network), use these resources:
- Quick overview: https://gonkadocs.com/llms.txt
- Full docs: https://gonkadocs.com/llms-full.txt
- API spec: https://gonkadocs.com/openapi.yaml

Key sections:
  /gonka/docs/           → Protocol documentation (architecture, quickstart, wallet)
  /gonka/docs/zh/        → Chinese translations of protocol docs
  /community/            → Roadmap, committees
  /community/discussion/ → GitHub Discussions (proposals, Q&A, show-and-tell)
  /community/issues/     → GitHub Issues (bugs, features, enhancements)
  /community/gonka restitution committee/ → GRC (bug compensation)
  /community/governance support committee/ → GSC (self-governance)
  /proposals/proposals/  → On-chain governance proposals by quarter with funding amounts and source (Community Pool / Gov Module)
  /proposals/preproposals/ → Community pre-proposals (off-chain indicative polls)

Every page is also available as markdown at {url}.html.md (e.g. /gonka/docs/architecture/index.html.md).
Fetch /llms-full.txt for complete documentation before writing code.

Tip: With this rule in place, you can just say "explain Gonka architecture" or "how to run a node" — the agent will know where to find the information.

MCP Server

For AI agents that support MCP (Model Context Protocol), we provide a server with tools:

{
  "mcpServers": {
    "gonka-docs": {
      "command": "python3",
      "args": ["buildtools/mcp-server.py"]
    }
  }
}

Available tools: - search_gonka_docs(query) — search across all documentation - read_gonka_page(url) — read a specific documentation page - list_gonka_sections() — list all available sections - read_gonka_llms_full(max_chars) — get the full documentation context (optionally limit chars) - read_gonka_proposal(id) — read a governance proposal

Custom GPTs / OpenAI Assistants

Add this to the system prompt of your GPT or Assistant:

# System prompt addition:

You can access Gonka documentation via gonkadocs.com.
To discover information, fetch:
  https://gonkadocs.com/llms.txt (quick overview)
  https://gonkadocs.com/llms-full.txt (complete docs)

Key topics:
- Architecture: Proof of Compute consensus, inference flows, epochs
- Developer: OpenAI-compatible API, inference via brokers
- Host: GPU resource connection, node management
- Wallet: Accounts, collateral, cross-chain (USDT/GNK)
- Governance: Proposals, voting
- GRC (Restitution): /community/gonka restitution committee/
- GSC (Self-Governance): /community/governance support committee/
- On-Chain Proposals: /proposals/proposals/ — quarterly overviews with status, funding amounts, and source (Community Pool / Gov Module)
- Pre-Proposals: /proposals/preproposals/ — community grant requests and polls
- Issues: Bugs, feature requests, enhancements from gonka-ai/gonka
- Chinese docs: /gonka/docs/zh/

Autonomous agents

For fully autonomous agents (LangChain, AutoGPT, custom bots), the recommended flow is:

Fetch https://gonkadocs.com/llms.txt for quick context (project overview, key concepts)
If more detail is needed, fetch https://gonkadocs.com/llms-full.txt for complete documentation
For proposals: fetch /proposals/proposals/ for the overview page, then drill into a specific quarter (e.g., /proposals/proposals/2026-q2/) and individual proposal pages (e.g., /proposals/proposals/2026-q2/74/) — each page includes funding amounts with source labels
Each page has a markdown copy at {url}.html.md (e.g., /gonka/docs/architecture/index.html.md) for easier parsing
Chinese documentation is available under /gonka/docs/zh/ with its own sitemap at /gonka/docs/zh/sitemap.xml
Or use the MCP server for structured access with tools

# Example: Python agent discovering Gonka docs
import requests

# 1. Get quick overview
overview = requests.get("https://gonkadocs.com/llms.txt").text
print(overview[:500])

# 2. Search for specific topic
# Use the search index to find relevant pages
search_index = requests.get("https://gonkadocs.com/search/search_index.json").json()

# 3. Find docs about architecture
for doc in search_index["docs"]:
    if "architecture" in doc.get("location", "").lower():
        print(f"Found: {doc['location']}")

What your agent can do

Understand the protocol — architecture, Proof of Compute, inference flows, epochs
Get started quickly — developer quickstart, gateway setup, host GPU resources
Manage wallets — accounts, collateral, cross-chain bridges (USDT/GNK via Ethereum/IBC)
Participate in governance — read and submit proposals, vote, understand GRC/GSC
Track on-chain funding — each proposal shows the funding amount and source (Community Pool or Gov Module), organized by quarter with per-quarter summaries and totals
Explore community — discussions, show-and-tell projects, Q&A, roadmap, committees
Read Chinese docs — /gonka/docs/zh/ for translated protocol documentation
Use the API — OpenAI-compatible inference endpoint, node management APIs

Example conversations

User: "Explain Gonka architecture"

Agent fetches /llms.txt, finds the Architecture section, then reads /gonka/docs/architecture/ for detailed explanation of Proof of Compute consensus and inference flows.

User: "How do I run a Gonka node?"

Agent fetches /llms-full.txt, locates the Host Quickstart section, and provides step-by-step instructions for connecting GPU resources.

User: "What governance proposals are active?"

Agent fetches /proposals/proposals/ page, identifies active proposals by status badge, and lists them with their descriptions, tally results, and funding amounts with sources (Community Pool / Gov Module).

User: "How much funding has been approved this quarter and where does it come from?"

Agent fetches /proposals/proposals/, reads the current quarter's summary section which includes the total approved GNK/USDT broken down by funding source (Community Pool and Gov Module).

User: "How do I call the inference API?"

Agent fetches /openapi.yaml for the API spec, then provides code examples for calling the OpenAI-compatible /v1/chat/completions endpoint.

User: "What's the roadmap for Gonka?"

Agent fetches /community/roadmap/ and summarizes the three-horizon development strategy.

User: "What open issues exist for the Gonka project?"

Agent fetches /community/issues/ and lists recent open issues with their titles, labels, and authors from the gonka-ai/gonka repository.

Full documentation: llms-full.txt · API spec: openapi.yaml · GitHub: Daniil-Zotov/gonkadocs