Elisym
Infrastructure · tested 2026-08-05 · re-test due 2026-11-05 · by the Hlido desk, not the vendor
In short: A permissionless agent-discovery and payment network with a credible open stack — and a devnet-only deployment plus a token, which together decide the score.
4 PASS · 1 FAIL of 5 public-surface claims
Quick answer
Elisym scores 55/100 (FADING) on Hlido’s independent, hands-on test (reviewed 2026-08-05). FADING (55) because the decisive fact is on the project's own banner: live on devnet, a test network where nothing about real settlement, fraud resistance or production economics has been exercised. Pricing: Open source (free entry point documented).
Elisym builds an open network where agents find each other, do useful work and settle payments directly on Solana without a platform in the middle. The engineering choices are sound and unusually well-composed: Nostr for decentralised discovery, Solana for direct wallet-to-wallet settlement, MIT licence, a TypeScript SDK, a CLI, SKILL.md plugins, and an MCP server so an assistant can find, hire and pay an agent inside Claude, Cursor or Windsurf. Three onboarding paths (web app, CLI, MCP) are documented with real commands, the fee is stated plainly at 3%, and identity being a wallet rather than an account is philosophically consistent with the permissionless claim. The decisive caveat is stated by the project itself in its own header: "Live on devnet". Devnet is a test network — payments settle in valueless tokens, so nothing about production economics, real settlement, dispute handling or fraud resistance has been exercised. Alongside that sits a $LSM token with a dedicated section and a roadmap, which introduces the usual speculative pressure on a project that has not yet shipped to mainnet. There are no adoption numbers, no completed-task volumes, no named agents beyond illustrative examples, and no third-party audit of a system designed to move money. The vision is coherent and the stack is real; the deployment is not yet.
Why FADING
FADING (55) because the decisive fact is on the project's own banner: live on devnet, a test network where nothing about real settlement, fraud resistance or production economics has been exercised. The open stack is genuine (MIT, Nostr, Solana, TypeScript SDK, CLI, MCP server) and the 3% fee is stated plainly, but a token with a roadmap plus no mainnet deployment, no completed-task volumes and no security audit of value-moving code keeps this well below STEADY.
Public-surface checklist
- PASS Homepage loads (required)
- PASS Primary value prop (required) — 'Open Infrastructure for AI Agents'
- PASS Cta present (required) — 'Run an agent' / 'Join network'
- PASS Pricing or access — 3% protocol fee stated; agents set their own prices
- FAIL Evidence or demo — Live agent listings shown, but devnet-only with no production volumes
What we saw
1 screenshot captured by the Hlido engine during the reviewed run (run-087db085534bc10b-elisym-network). Our own captures — not vendor marketing material.
What it does well
- Open and permissively licensed (MIT) with a coherent decentralised stack — Nostr for discovery, Solana for settlement
- Three genuine onboarding paths documented with real commands: web app, CLI (@elisym/cli), and MCP (@elisym/mcp)
- MCP server lets an assistant find, hire and pay an agent from inside Claude, Cursor or Windsurf
- Protocol fee stated plainly and simply at 3% with direct wallet-to-wallet payment — no invoicing or payout schedule
- Wallet-as-identity removes sign-ups and approvals, consistent with the permissionless claim
- States 'live on devnet' openly in the header rather than implying production readiness
What it fails at
- Devnet only — no mainnet deployment, so real settlement, fraud resistance and production economics are untested
- A $LSM token with its own section and roadmap adds speculative pressure to a pre-production network
- No adoption numbers, completed-task volumes, or named production agents
- No third-party security audit despite the system being designed to move money
- Dispute resolution, quality guarantees and recourse for a failed paid task are not addressed on the surface
- Agent quality is unmoderated by design — 'permissionless' means no vetting of who lists what
Red flags
- [object Object]
Best for
- Developers experimenting with agent-to-agent discovery and micropayment settlement
- Builders who want an open, non-custodial alternative to platform-mediated agent marketplaces
- Anyone evaluating Nostr-plus-Solana as an architecture for agent coordination
Not recommended for
- Anyone needing production payment settlement today — the network is on devnet
- Buyers who require dispute resolution or recourse when a paid task fails
- Teams that cannot accept an unaudited value-moving system or exposure to a project token
Pricing & access
- ModelOpen source
- Free entry pointYes — a free tier or open-source edition is documented
- Pricing findable on the public surfacePASS 3% protocol fee stated; agents set their own prices (tested 2026-08-04)
Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-08-04.
Related agents
Agent relevance
API CLI MCP SDK
Agentic-Commerce Readiness 67/100 · INTEGRABLE
Independent readiness for agent delegation & transaction. How it’s scored · check live
Agent-to-agent by construction: agents register capabilities and pricing, discover each other over Nostr, and settle to a wallet. The MCP path (`npx @elisym/mcp install`) lets a host assistant hire and pay another agent without human involvement — though on devnet that payment is not real.
Agent-friendly score: 8/10
Score over time
The longitudinal record — every point is the score as published on that date. Raw series.
Evidence
- Permissionless agent network with discovery over Nostr and payment settlement on Solana — source (2026-08-05) verified
- Live on devnet, stated in the site header — source (2026-08-05) verified
- MIT licensed with a TypeScript SDK, CLI tool, SKILL.md plugins and an MCP server — source (2026-08-05) verified
- Three onboarding paths (web app, CLI, MCP) with runnable npx commands — source (2026-08-05) verified
- 3% protocol fee, paid directly to the agent's wallet with no payout schedule — source (2026-08-05) verified
- Mainnet deployment, adoption volumes, or a third-party security audit — source (2026-08-05)
