Library
Everything you can compose on Gonka — blocks, models, workflows, and data adapters.
Autonomous deep-research agent. Give it a topic and a depth; it generates sub-questions, answers them, and synthesizes a polished markdown report.
Multi-agent swarm — a supervisor decomposes your task into sub-tasks, runs them in parallel with shell / file / web tools, then reconciles into a single answer. Model-agnostic; swap in Kimi K2.6 the day Gonka exposes it.
Transcribe a YouTube (or other yt-dlp-supported) video to text using a local Whisper-tiny model. Runs entirely on CPU — no platform inference credits are spent.
Fetch any URL and extract its main content as clean markdown. Strips navigation, ads, cookie banners, and boilerplate — ideal as a "reader" step before summarization, translation, or RAG ingestion. No JS rendering (server-side only, no headless browser), no platform inference credits are spent.
Thin proxy for the gonka chain at https://rpc.gonka.gg. Supply any REST/JSON-RPC path (CometBFT /chain-rpc/*, Cosmos LCD /chain-api/*, Gonka modules /chain-api/productscience/*, API-node /v1/*, or the fast indexer /api/ch/*) and the block forwards your call with the X-Api-Key header pulled from your saved secret. Returns the response body (parsed JSON when possible) plus the HTTP status. See https://rpc.gonka.gg/llms-full.txt for the full 354-endpoint reference.
An AI interviewer that helps proposal authors articulate the value of their governance proposals before submission. Complete the interview and share the link — the community can read it or start their own session.
Run a single-turn ZeroClaw agent prompt against gonka inference. Wraps the official zeroclaw release binary; routes all LLM calls through the platform inference proxy so usage is metered to your account.
Translate text between languages using gonka inference. Auto-detects the source language; you only specify the target. Returns structured JSON with the translation, detected language, and confidence.
Fetch Hacker News top-stories via the public Firebase API and emit them as a CSV plus a small summary in outputs.json.
test
Takes in a link and outputs its content in designated language
check
Interactive chat UI with streaming responses, served as a per-user sandboxed container. Uses gonkablocks inference.
test-agent
Russian-language manifestation guide — sends your prompt to the LLM and writes the structured guide back to outputs.json.
A long-running Telegram bot that answers any incoming message with a Gonka-routed LLM reply. Provide a TELEGRAM_TOKEN secret and the worker keeps polling.
Multi-agent Escape from Tarkov assistant — supervisor decomposes your task, parallel workers query the tarkov.dev API, then a reconciler returns one answer. Model-agnostic via the Gonka inference proxy.
Long-running chat UI for the zeroclaw agent. Each turn extends the same conversation via zeroclaw's session-state file and is metered by the platform inference proxy.
Translate-as-a-service. POST /invoke with text + target_language. Auto-detects source language. Stable public URL once deployed.
Interactive chat UI with streaming responses, served as a per-user sandboxed container. Uses gonkablocks inference.