Franklin
An autonomous AI agent with its own wallet that spends USDC to get real work done.
Last release Jul 26, 2026 · Last commit Jul 26, 2026 · ★ 629
Overview
An autonomous AI agent with its own wallet that spends USDC to get real work done.
Sato Score
67/100· MediumHow open, active & verifiable — not a safety or returns grade.
- Maintenance & liveness
- 35/35
- Code transparency
- 19/20
- Interface verifiability
- 0/10
- Docs & demo
- 0/10
- Listing transparency & provenance
- 13/15
- Independent verification
- 0/10
Details
- Chains
- Base, Solana
- Agent types
- DeFi/Yield
- Open source
- Yes
- Pricing
- Pay-per-action (YOPO): provider cost + 5%, settled in USDC per action; no subscription.
- Last checked
- 2026-07-31
Marketplace Signals
- ▸ Functional agent
Verification status reflects evidence reviewed by Sato Hub. Self-reported claims are never presented as verified.
Where Franklin fits
Related concepts: How Agents Interact With Smart Contracts, How Agents Use Wallets, What Are Onchain Agents?
Recent Activity
All news →Releases and posts from Franklin, pulled from its GitHub and X.
Franklin Agent 3.38.0 — the Polymarket port was eight rounds stale
The port had drifted, and the drift was all money bugs. Franklin's Polymarket module is a byte-faithful port of blockrun-mcp's client-side CLOB code. It was last synced at upstream f8d3f8c (Franklin 3.30.1). Upstream has since shipped eleven commits through 0.32.9, and Franklin had none of them. The port is now re-synced to upstream HEAD: every file is byte-identical to blockrun-mcp except two imp
Franklin Agent 3.37.0 — the dependency backlog, cleared
`ws` was never actually declared. `src/serve/server.ts` imported it while `package.json` never listed it — it resolved only because a transitive dependency happened to hoist it to the top of `node_modules`. Any shift in that tree, or in npm's hoisting, would have broken `franklin serve` at runtime with nothing in CI to catch it. The types were wrong too (`@types/ws` pinned at 7 while ws 8 actually
Franklin Agent 3.36.0 — Claude Opus 5 is the default Opus
`opus` and `claude` now resolve to `anthropic/claude-opus-5`. It is live on the BlockRun gateway at $5/$25 per 1M tokens — the same price as Opus 4.8, so this is a straight capability upgrade with no cost delta and nothing to opt into. The Smart Router's COMPLEX and REASONING tiers point at it, `franklin init` writes it as `ANTHROPIC_DEFAULT_OPUS_MODEL`, and the savings-vs-Opus figure in `/stats`
Franklin Agent 3.35.6 — proxy honors explicit max_tokens; build-breaking deps pinned
The payment proxy stops overwriting an explicit max_tokens . It used to replace whatever you sent with min(adaptive, modelCap) , where adaptive is twice the previous reply on that model. A client asking for 500 tokens right after a long turn was forwarded several thousand instead — and because the gateway quotes on the ceiling a request asks for, that inflated the upfront hold on a request you'd d
Franklin Agent 3.35.5 — the last three OpenAI ceilings, measured
model was now openai/gpt-5.5 32,768 128,000 openai/gpt-5.4 32,768 128,000 openai/gpt-5-mini 16,384 65,536 These took three attempts, and the story is worth keeping because a token table is where a wrong belief gets frozen. First attempt set gpt-5.5 and gpt-5.4 to 100,000, citing "upstream's stated maximum". There was no upstream response — both BlockRun SDKs rejected max_tokens > 100000 client-sid
Franklin Agent 3.35.4 — Haiku and Sonnet 4.6 get their real output ceilings
model was now anthropic/claude-haiku-4.5 16,384 64,000 anthropic/claude-sonnet-4.6 64,000 128,000 Both are Anthropic's documented limits. They were held low deliberately. The gateway clamped Haiku to 8,192 and Sonnet 4.6 to 64,000, and asking past a clamp is not free — the upfront price quote scales with the ceiling you request, so a bigger ask inflated what you were quoted while the reply stayed
Listing History
Recorded changes to Franklin's listing — status, verification, Sato Score, and recent releases. Full history is kept in the listing record.
- Jul 26, 2026New release published
- Jul 25, 2026New release published
- Jul 14, 2026Sato Score tier: High → MediumScore 72 → 67.
- Jul 14, 2026Edited: deploy spec
- Jun 25, 2026Enriched: pricing
- Jun 25, 2026Auto-enriched: creator name, creator profile url, website url, chains supported +5 more
- Jun 25, 2026Sato Score tier: Medium → HighScore 65 → 72.
- Jun 18, 202612 releases trackedShowing the 2 most recent. Full release history is retained in the listing record.
- Jun 17, 2026Sato Score tier: Low → MediumScore 33 → 64.
- Jun 16, 2026Sato Hub began tracking this listing
Related Resources
Zerion
ActiveWallet and DeFi data provider offering portfolio, positions, transactions, PnL, and prices across many chains, with an agent-accessible API.
Agent Payments Protocol (AP2)
ActiveOpen protocol from Google for secure agent-led payments across traditional and crypto rails, with a crypto-native x402 extension built with Coinbase, Ethereum Foundation, and MetaMask.
Bankr
ActiveCrypto execution layer and cross-chain wallet that lets agents and users trade, bridge, and manage assets via natural language, plus a skills marketplace and an x402 cloud for spinning up pay-per-call agent endpoints.
Join the Sato Hub Briefing
One email a week — the agents, tools, and infrastructure that actually shipped, and why they matter.