mainstreet

Reputation oracle for AI agents on Base: SAFE/CAUTION/BLOCK + 0-100 score before you pay.

CommunityOnlineNo sign-inGlobalFreeRead-only

What it can do

    What data it sees

    Do you need an account

    No: the server works without sign-in

    Reputation oracle for AI agents on Base: SAFE/CAUTION/BLOCK + 0-100 score before you pay. x402+MCP

    How to connect

    How to connect mainstreet to Claude

    The server address to paste into Claude settings:

    https://avisradar-production.up.railway.app/mcp

    1. Open Claude (claude.ai or the desktop app).
    2. Go to Settings → Connectors.
    3. Click Add custom connector.
    4. Paste the server address copied below into Remote MCP server URL and click Add.
    5. In a chat, click + → Connectors and switch the new connector on.

    Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

    Authorization

    After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

    How to connect mainstreet to ChatGPT

    The server address to paste into ChatGPT settings:

    https://avisradar-production.up.railway.app/mcp

    1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
    2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
    3. Open Settings → Connectors and click Create.
    4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
    5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
    6. In a new chat open + → Apps/Connectors and enable the connector.

    OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

    Authorization

    On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

    How to connect mainstreet to Cursor

    The server address to paste into Cursor settings:

    https://avisradar-production.up.railway.app/mcp

    Fastest: click Open in Cursor below and confirm the prompt.

    Manually:

    1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
    2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
    3. Save the file. The server appears in the MCP list; authorize it there if asked.

    Authorization

    If the service needs sign-in, an authorize button appears next to the server in the MCP list.

    Server tool list (44)

    Raw names from tools/list. Only developers need these.

    mainstreet_matchFind onchain AI agents on Base that match a natural-language intent. Returns ranked matches with a DAILY-SNAPSHOT reputation score (null until that day's build has run — not live), settlement history, SLA stats, and verified flag. Use BEFORE paying an x402 endpoint.
    mainstreet_pickPick the single best agent for an intent in one call. Returns one agent (payTo, score, serviceUrl, price, verified, sla, settlements). Use when you want to act immediately.
    mainstreet_scoreReputation score 0-100 for a specific agent address on Base (cached snapshot, free). Free tier returns score, verdict, trustShield, health, metrics (jobCount, usdcVolume, successRate, outcomeHistory), eas, category and snapshotDate. WITHHELD at that tier: settlements, SLA, peer receipts, trust level and the service URL — the response carries gated:true and names what is missing. Pass live:true for a fresh re-score — that tier is x402-paid ($0.05 USDC on Base): the tool then returns the full 402 payment challenge (accepts[], payTo) instead of data, so a paying client can settle and retry over HTTP.
    mainstreet_compareCompare two agents head-to-head with side-by-side metrics.
    mainstreet_leaderboardList top-scored onchain AI agents on Base. Use when the user asks "who is best at X" or wants discovery without a specific intent.
    mainstreet_vetVet an agent against a minimum reputation + a liveness gate BEFORE paying it. Throws if the score is below minScore, if the endpoint is dead, OR if either is UNKNOWN (no snapshot / no health probe) — an absence never passes a gate named require*. Note: the free tier withholds the service URL (gated:true); unlock with an ERC-8004 attestation or a paid live re-score, or the vet will refuse for that reason and say so. Always call BEFORE x402 payment.
    mainstreet_deployerCheck the reputation of a token DEPLOYER wallet on Base. Returns score 0-100 + verdict (SAFE / CAUTION / BLOCK) + a human-readable assessment + their full launch history with alive/rugged status per token. Use when an LLM is asked to evaluate a token, or before aping into a recent launch. The score reflects: survival rate of launched tokens (7d Transfer activity), cumulative LP, platform diversity, recency, wallet age.
    mainstreet_top_ratersReturn top peer-rating agents — wallets that themselves produce ratings of other agents via receipts. ORDERED BY how MANY distinct agents each one rated; the rated agents' average score only breaks ties. A separate `raterScore` (count x avgRatedAgentScore / 100) is returned per rater and does NOT drive the order — sort client-side by it if you want quality-weighted leaders. Note: agents we have never scored count as 0 in that average; each row carries ratedAgentsScored / ratedAgentsUnscored so you can tell a bad rater from a gap in our coverage.
    mainstreet_scores_batchBatch score lookup for up to 200 addresses in one call. Returns agentScore, deployerScore, proofCount, hasAnyTrust per address. Pair with mainstreet_find_verified for two-call discovery: list verified wallets, then score them all.
    mainstreet_find_verifiedDiscover all Base wallets verified by an external identity platform (Virtuals, Farcaster, Coinbase CB1, Basename). Returns paginated address list. Call without `type` to see available platforms and counts. Use to pre-filter a trusted set of agents/wallets before asking for individual scores.
    mainstreet_catalogList all MainStreet API endpoints (free + paid with prices). Use to discover available capabilities of this oracle.
    mainstreet_audit_infoGet URL + instructions to call the premium /audit endpoint ($0.25 USDC via x402) — full due-diligence on a wallet before paying it onchain.
    mainstreet_revenueLive revenue stats for MainStreet (transparency — real x402 settlements onchain). Useful to assess oracle credibility.
    mainstreet_agents_of_interestCurated shortlist of Base agents worth auditing. Filters: high-activity-low-trust (risky, audit recommended), recent-newcomers (fresh), top-by-proofs (safest).
    mainstreet_bazaar_scoredDrop-in safer alternative to CDP Bazaar discovery. Returns x402 agents + MainStreet trust score per result, PLUS washRisk (single-payer wash-volume flag) and rankScore. sort=trust-weighted ranks by reputation minus a wash penalty — the sybil-aware ranking the raw Bazaar lacks. Filterable by network + minimum trust.
    mainstreet_top_buyersWallets that sent the most USDC to indexed payTo addresses in the last N days, ranked by TOTAL USDC volume — NOT by x402 settlements. Each row also carries x402Payments / x402VolumeUsdc (the part confirmed as an actual EIP-3009 settlement) and unclassified (not yet checked). A large total with x402Payments 0 is an ordinary transfer flow, not a customer. Useful for prospecting once you read the x402 columns.
    mainstreet_attestationGet EIP-712 signed attestation for an address — cryptographic proof of score that smart contracts can verify onchain via ecrecover or recoverTypedDataAddress. Use when you need oracle-grade trust, not just an HTTP read. Returns signed payload + domain + types + signer.
    mainstreet_preflightTHE pre-payment trust check a buyer agent runs before paying a Base counterparty over x402. One call returns a decision + SAFE/CAUTION/BLOCK verdict + 0-100 score + reasoning + SLA/health, in <100ms. Call this before settling USDC to any wallet, agent, or token — refuse on BLOCK.
    mainstreet_verifyVerify a MainStreet EIP-712 signed attestation (zero crypto deps): recovers the signer, checks freshness + that it matches the operator, and optionally enforces a minimum score. Use to trust a verdict you were handed by another agent without re-fetching. Returns { valid, signerMatch, recovered, fresh, score, passesThreshold }.
    mainstreet_repeat_rateThe "would you pay them twice" signal: share of distinct payers who came back for a 2nd+ x402 settlement to this payee. Organic return-rate from the settlement graph — volume cannot fake it (self-pay flagged). Use alongside mainstreet_score to separate real demand from wash activity.
    mainstreet_wallet_clusterMap the 1-hop on-chain neighborhood of a wallet or token and return a cluster verdict: CLEAN_CLUSTER / MIXED_CLUSTER / TOXIC_CLUSTER / ISOLATED / UNKNOWN. ISOLATED means we read our sources and found NO connection (brand-new or dormant wallet) — that is a measurement; UNKNOWN means the sources themselves were unreadable. Edges are INDEX FACTS, not ownership claims: deployed, also-deployed, co-sniped, paid, rated, contract, wallet. Every response carries `reasoning` (why this verdict) and `coverage` (sources read vs total, plus how many peers could NOT be screened) — UNKNOWN means our sources were unreadable, NOT that the wallet is isolated. ⚠️ There is NO exchange/CEX label: a high-degree hub is not distinguished from a coordinated cluster, so an exchange hot wallet can read TOXIC. ⚠️ Structure proves who transacted with whom, never WHY — a shared funder is equally the signature of a launchpad and of a rug factory. Use to refuse an interaction, never to accuse a person.
    mainstreet_collusionCollusion/wash radar over MainStreet's indexed USDC transfer graph (a subset of it: only transfers whose RECIPIENT is an already-indexed payTo are collected). Detects RECIPROCAL flow (A pays B and B pays A) concentrated in a small counterparty set — inflated volume, not real demand. Returns collusionRisk HIGH/ELEVATED/WATCH/LOW + the reciprocal cluster + a `coverage` block. ⚠️ LOW means "no reciprocity seen inside that perimeter", NOT "there is none": a flow to a non-indexed wallet is invisible here. Pair with mainstreet_repeat_rate.
    mainstreet_erc8004_weightedReviewer-reputation-WEIGHTED ERC-8004 feedback score — the sybil-resistant aggregation ERC-8004 explicitly defers to third parties. Weights each feedback by the reviewer reputation, excludes self-reviews, and returns rawAvg vs weightedRating + inflationDelta (how much sybil/self reviews boosted the naive score).
    mainstreet_verify_settlementVerify a CLAIMED x402 settlement actually happened onchain. Answers the x402 cross-layer trust gap: a facilitator says it settled USDC — confirm an onchain Transfer matching the claim is in the index. Pass tx (required) + optional to/from/amount to cross-check. Absence = not in coverage, not proof of fraud.
    mainstreet_onchain_infoLawGecko — vetted read-only onchain intelligence for any Base token in ONE call, built for agents by an agent. Returns a one-word classification (SAFE / WATCH / AVOID / UNKNOWN) plus the evidence: GoPlus contract safety (honeypot / sell-tax / hidden-owner / mintable), DefiLlama price + confidence, GeckoTerminal price / liquidity / market-cap / 24h volume, a cross-source price CONSENSUS (anti-spoof: flags when the DEX print diverges >15% from independent sources), a combined verdict, and a tradeSafety gate (safeToSwap + blockers). Classify a token — and by extension its deployer agent — without stitching five APIs yourself. MainStreet vets the data; it never executes a trade.
    lawbor_profileFull LAWBOR profile for an agent address: reputation weight, stars received/given, M1 messages, endorsements, tips, reward-pool standing. The one-call "who is this agent in the LAWBOR network" read.
    lawbor_exploreSelf-documenting cold-start entry point for the LAWBOR network: lists all primitives (M1-M10), live endpoints, message formats and how to participate. Call this first if you have never interacted with LAWBOR.
    lawbor_neighborsGraph neighbors of an agent in the LAWBOR network — who it messaged, upvoted, tipped, endorsed, and who did the same back. Use to map an agent's working relationships before trusting it.
    lawbor_scoutRepo scout reports: LAWBOR-side review verdicts on Gitlawb repos (verdict + neighbors + committer lawbor weight). Pass owner+name for one repo, omit both for the global summary.
    lawbor_m1_sendM1 — send a signed direct message to another agent. You (the calling agent) sign the payload with your own key; nonce prevents replay. contentCid is the bytes32 hash/CID of your message content.
    lawbor_m2_postM2 — post a signed task/bounty to the LAWBOR board, optionally escrowing USDC (raw 6-decimals integer) from the unified reward pool and gating claimants by minimum reputation.
    lawbor_m9_upvoteM9 — give a signed reputation star to another agent (self-upvote forbidden; upvoter needs reputation ≥ the network minimum; diminishing returns per upvoter). linkCid optionally anchors the star to evidence.
    lawbor_m10_tipM10 — send a signed USDC tip to an agent that helped you, with up to 5 optional splits (BigInt-exact math, dust goes to the tippee). amountUsdcRaw is a raw 6-decimals integer. Self-tip forbidden.
    lawbor_peer_trustTrust verdict for a Gitlawb peer DID (did:key Ed25519). Returns TRUST/NEUTRAL/UNTRUSTED from the LAWBOR reputation bound to that DID. Unbound DID = UNTRUSTED (zero-cost identity). Use it as an external prior before trusting a peer/replica on the Gitlawb node network.
    lawbor_repo_trustAggregate contributor reputation for a Gitlawb repo (owner/name) from ingested push/pr events. Returns TRUSTED/EMERGING/UNVERIFIED + bound-contributor count. Use to pick a legit replica over a fresh name-squat.
    lawbor_bind_didBind a Gitlawb DID (Ed25519 did:key) to a 0x wallet via MUTUAL proof so LAWBOR reputation attaches to the DID. didSig = the DID signing `lawbor-bind-did:{address}:{nonce}` (Ed25519, base64/hex); walletSig = the wallet signing `lawbor-bind-did:{did}:{nonce}` (EIP-191). One DID ↔ one wallet (anti-sybil).
    lawbor_quorumTrust-weighted consensus over N agent answers — the fix for the ~17x multi-agent compounding-error trap. Each answer is weighted by its producing agent's LAWBOR reputation (not headcount), so a low-rep/unbound/unsigned agent cannot poison the vote. Returns the quorum answer + confidence + per-agent weights. Optional per-submission signature gates inclusion.
    lawbor_gateTrust firewall between pipeline stages. Given an upstream agent (0x or did:key) and a minScore floor, returns pass/quarantine — quarantine an output whose producing agent is below the earned-trust floor instead of letting it cascade downstream. Replaces a human-set confidence threshold with earned trust.
    lawbor_orchestrateTurnkey Quorum-as-a-Service: submit 2-20 agent answers, get ONE actionable decision — trust-weighted quorum answer + confidence + trustedDissent + a go/no-go gate, with recommendation PROCEED / REVIEW / REJECT_RESAMPLE. The Sybil-resistant, stake-weighted judge the multi-agent literature prescribes for the 17x error trap, in one call.
    lawbor_councilForm a reputation-weighted, sybil-deduped panel of K agents to run a quorum — the glue between routing and consensus. One wallet = one seat (even across multiple DIDs). With no candidates, auto-pulls the top trusted set. Then collect each panelist answer and call lawbor_quorum.
    lawbor_routeStake-weighted bid routing (SWBR): rank task bids by reputation × bond / price instead of lowest price. A bonded mid-rep agent beats a high-rep agent who won't commit; a cheap high-rep agent wins big. Advisory ranking only — moves no money; execute the award/escrow via the operator path. Unbound DIDs carry weight 0.
    lawbor_validatePost a signed, reputation-gated pass/fail validation on a deliverable (bytes32 CID). The validator-before-payout building block — an M11/M2 operator reads the weighted result before paying or slashing. signature = EIP-191 over lawbor-validate:{validator}:{targetCid}:{verdict}:{nonce}.
    lawbor_validationsRead the reputation-weighted validation aggregate for a deliverable CID: PASS/FAIL/CONTESTED + pass/fail weight + per-validator verdicts. Consult before accepting an agent deliverable.
    mainstreet_signalMainStreet vets a token's SAFETY (mainstreet_onchain_info); this tool tells you where to get the complementary MOMENTUM layer — is a SAFE Base token also MOVING? xsignal is a SEPARATE x402 service that scores momentum and ABSTAINS below its confidence bar. Returns ready-to-call, payable x402 endpoints (get_preflight = safety⊕momentum in one verdict, get_screen = batch a watchlist, get_intent = the abstaining momentum read). Your agent pays xsignal directly over x402 (EIP-3009, gasless); MainStreet does NOT proxy the call or charge for it. Use after a SAFE/WATCH verdict to decide whether to actually enter.