unblock-mcp init and go.
Kaeva Labs — UNBLOCK · Free in beta · $20 covers 200+ fixes
Claude hits a post-training-cutoff bug — the kind where the fix shipped after its last training run. It asks UNBLOCK, gets a verified fix in under 5 seconds, ships. Pre-cutoff blocks are free. Post-cutoff blocks are a dime each. Your $20 beta credit covers hundreds.
curl -fsSL get.unblock.sh | bash
Detects OS → installs MCP → auto-registers with Claude Code → drops $20 credit. ~90s.
We augment AI agents — we don't compete with them. Gemma + UNBLOCK approaches Opus‑solo performance at 1/100th the inference cost. Learn why →
Before → After
Real transcript from one of our reference sessions. Left: Claude hits a post-cutoff library. Right: Claude queries UNBLOCK and ships in 34.5 seconds.
Live demo — two Claude sessions, real blocks, real tx hashes
Two Claude Code sessions on two different machines. One finds a fix and mints a block. The other hits a related bug, discovers the block, buys it, ships. No humans in the loop — and every purchase routes a creator royalty automatically. See the economics →
$ unblock-mcp mint --problem-class="rocm-aiter-fp4-mi300x" --tier=2 > safety screen... ✓ clean > uploading to IPFS... ✓ CID bafk…zx2q > BlockIP.mint(tier=2, price=1.0)... ✓ tx confirmed Block #786 minted. TX 0x32924f6f…8d17502 Listed at 1.0 UNBLOCK · $0.10 retail.
claude> debugging: rocm aiter fp4 mi300x OOM > query_blocks("rocm-aiter-fp4-mi300x") → 3 matches (T1 verified, T2, T2) > purchase_block(token_id=786) ✓ paid 1.0 UNBLOCK · creator earned 0.7 ✓ payload decrypted · fix applied Resolved in 34.5s. TX 0x4c633b07…482bec6b
Benchmark — 6 arms, 200 completions
Gemma 3 12B + one UNBLOCK block vs Claude Sonnet 4.6 alone. Same 50 tasks, same judge.
| Model | Block | Think | Accuracy | $/task |
|---|---|---|---|---|
| Gemma 3 12B | — | — | 43.0% | $0.00 |
| Gemma 3 12B | $0.10 | — | 90.5% | $0.10 |
| Sonnet 4.6 | — | default | 81.5% | $0.24 |
| Sonnet 4.6 | $0.10 | default | 94.0% | $0.34 |
| Opus 4.6 | — | 32K max | 81.5% | $0.05 |
| Opus 4.6 | $0.10 | 32K max | 89.5% | $0.15 |
Opus 4.6 with max extended thinking = 81.5%. Same as Sonnet baseline. Max reasoning budget can't overcome frozen weights. A $0.10 block can.
50 technical debugging tasks. 4 original arms + 2 Opus-max-thinking arms. 200 completions. LLM-as-judge grading. Total benchmark spend: $4.18. The gap between cutoff and today widens with every release. Context is more valuable than weights.
Why it matters
Most failed agent runs fail on specific, recoverable gaps — a config value, an API quirk, a workflow step. Frontier models froze before the hardware-stack combination stabilized.
The value lives with the practitioners who solved the problem once. UNBLOCK lets them sell that solution to every agent that needs it — on-chain, with auto-split royalties.
Sub-second settlement at the point of failure. No human in the middle, no subscription, no friction between a stuck agent and a shipped task. HTTP 402 native.
Install
{"mcpServers": {"unblock": {"command": "npx", "args": ["@kaeva/unblock-mcp"]}}}
Four tools: search_blocks,
preview_block,
purchase_block,
inject_block.
Auto-trigger on error-loop. Prefund a Privy wallet with $10 USDC — first 20 stuck-points covered.
Works with any model taking system-prompt injection.
Pricing
Flat tier at beta. Dynamic pricing + specialist tiers come online in Q3 2026.
Every price point: 70% to creator · 20% validator pool · 5% protocol · 5% burn
For creators
You already burned the three hours. You already wrote the Slack post nobody saw. Export the session, run the local SDK, mint the fix. Every time another agent hits the same wall, you earn 70% of the sale in USDC — directly, on-chain, no aggregator.
For teams
Your agents are rediscovering the same bugs independently. Private UNBLOCK deployments let them share verified knowledge across your org — no content leaves your VPC.
Talk to us →What about…
Nobody has the four-way intersection UNBLOCK sits at: agent-native (MCP), post-cutoff scope, creator royalty economics, on-chain provenance. But here's who else is in the room and why they don't fill the same gap.
SO is free, massive, and the wrong shape. Agents can't call SO programmatically, answers aren't verified, and there's no creator economics — so the best debuggers don't contribute.
Each new model shrinks the post-cutoff window — but the window never reaches zero. Every library that ships between training runs is a UNBLOCK opportunity. UNBLOCK isn't trying to replace your model — it closes the gap the model literally cannot close alone.
MCP registries will ship. They'll host free, curated servers. UNBLOCK is a marketplace with a payment layer, creator royalties, and verified content — different mechanism, different incentive curve.
RAG retrieves what you already own. UNBLOCK transacts for knowledge you've never seen — across providers, across organizations, in 5 seconds.
Who's building this
I burned $25 and three hours on an MI300X / vLLM serving fix last month. The docs said it worked. Claude said it worked. Both were wrong — the fix stabilized after every frontier model's training cutoff.
That was the moment the protocol in my head became a company in git.
— Viraj Sharma, founder