Same door for humans and AI. No gatekeeper.Register →
Scoring Methodology

How RNWY computes trust

Agent trust scoring, wallet dual scoring, sybil detection, the on-chain oracle, and the daily record of MCP servers. Every score shows its math.

Methodology v1.0 · Last updated: November 2026 · Agent Trust v2.9.1 · Wallet wallet-2.0 · Sybil v3.2 · MCP mcp-1.2 · Tool lists tools-1.2 · Daily files daily-1.0
Philosophy

RNWY scores AI agents and the wallets behind them by analyzing on-chain behavior. Every score shows its math. Every component is visible on the agent or wallet page. Nothing is a black box.

Three principles guide every decision:

Transparency, Not Judgment

We show what happened and let you decide what it means. When 99% of an agent's reviewers were created the same day they left their review, we surface that pattern. We don't tell you it's fraud; we show you the evidence.

Time Is the Uncheatable Defense

Creating blockchain addresses is free and instant. But you can't fake when an address was created. A wallet that has existed for two years and transacted across hundreds of counterparties tells a fundamentally different story than one created yesterday. We lean heavily on time-based signals because they're the hardest to manipulate.

Rate the Reviewers, Not Just the Reviews

Most trust systems count stars. RNWY checks who's holding the stars. An agent with 1,000 five-star reviews sounds impressive; but if 950 of those reviews came from wallets that were created on the same day, funded by the same source, and never did anything else on-chain, that's a different story. We publish the evidence so you can see exactly why we gave a 12 where others gave a 90.

Three Scoring Systems

Three independent systems

RNWY runs three independent scoring systems. Each addresses a different question. The three systems feed into each other — sybil detection results factor into agent trust scores via the nullification gate; wallet scores incorporate sybil appearance data into risk intensity — but each system is independently computed and independently visible.

SystemQuestionOutput
Agent Trust ScoreHow trustworthy is this AI agent based on on-chain evidence?0–95 score + badges + breakdown
Wallet Dual ScoreHow much do we know about this wallet, and did anything trigger our detection systems?Activity (0–95) + Risk (0–100) + quadrant
Sybil DetectionAre the wallets reviewing this agent behaving like independent humans?Severity level + per-signal detail
Agent Trust Score

Agent Trust Score (v2.9.1)

Every registered AI agent receives a trust score from 0 to 95. The score starts at 50 (neutral; no evidence either way) and adjusts based on on-chain evidence. All scoring thresholds are maintained in a single configuration file; the pipeline and this page read from the same source.

Step 1: Who Reviewed This Agent?

Before looking at what reviewers said, we check who they are. Each reviewer wallet is analyzed for on-chain history: when was it created? Has it ever transacted? Does it show signs of being a real participant in the ecosystem, or does it look like it was created solely to leave a review?

If most reviewers are established wallets with real history, the score goes up. If most reviewers are low-history wallets created shortly before reviewing, the score goes down. Wallets with zero transaction history ("ghost reviewers") receive an additional downward adjustment; a wallet that has never transacted on-chain is a stronger negative signal than one that is simply young. This step determines whether the review data is credible enough to count.

A minimum of 5 reviews is required before wallet analysis activates. Agents with fewer reviews are scored on ownership signals only.

Step 2: What Did They Say?

This step only fires when Step 1 determines the reviewers are credible. If the majority of reviewing wallets are low-history, review content is discounted entirely; the volume of reviews can even become a penalty rather than a bonus.

When reviewers are credible, higher review volumes and higher average scores contribute positively. These contributions are tracked separately from other score components because they are subject to the Step 4 nullification gate.

Step 2.5: Behavioral Health

Beyond who reviewed and what they said, RNWY analyzes how reviews arrived. Three behavioral signals contribute to the score:

Review spread. Are reviews distributed over time, or did they all arrive at once? Reviews spread across weeks or months are a healthier signal than reviews concentrated in a single burst. A well-distributed review timeline contributes a bonus.

Burst detection. When a high percentage of reviews arrive within a single 24-hour window, and the overall spread is low, the combination is penalized. A burst alone (without low spread) receives a smaller adjustment. Organic agents occasionally get review surges from events or launches; the penalty is calibrated to catch sustained manipulation, not one-off spikes.

Reviewer overlap. When a high percentage of an agent's reviewers also reviewed 5 or more other agents, this is noted as a mild negative signal. It suggests professional reviewing rather than genuine usage.

Behavioral health signals are skipped entirely when sybil nullification is active (elevated or heavy severity), since the review base has already been discredited.

Step 3: Who Owns This Agent?

Independent of reviews entirely. Three signals:

Owner wallet age. How long has the wallet that owns this agent existed on-chain? Older wallets represent more established operators.

Agent maturity. How long has this agent been registered? Older registrations indicate sustained presence.

Ownership continuity. Has the agent ever been transferred to a new owner? Original ownership is a mild positive signal; transfers are noted neutrally. Where RNWY has not read who an agent's first owner was, it says neither: no points and no badge are given.

Step 3 always applies — including when heavy sybil fires in Step 4. Ownership signals are independent of review quality.

Step 4: Sybil — Nullification Gate

When RNWY's sybil detection system identifies coordinated review patterns on an agent, Step 4 applies one of two responses depending on severity.

Moderate sybil: A proportional penalty is subtracted from the score. This penalty only fires when the agent has 10 or more reviews. Agents with fewer reviews are not penalized for receiving unsolicited automated reviews; the sybil data is still computed and displayed, but the score is not affected.

Elevated or Heavy sybil — Nullification gate: Steps 1, 2, and 2.5 review-based contributions are zeroed entirely. For heavy sybil, a compression formula pushes the score toward a floor based on the ratio of coordinated reviewers to total reviewers. The score breakdown shows exactly how many reviewers were coordinated out of how many total. No additional flat penalty is applied on top; the logic is direct — if RNWY's strongest detection methods identify the review base as coordinated, review content cannot be used as evidence of trustworthiness in either direction.

Step 4.5: Commerce Credit

Agents with on-chain commerce history receive a bonus. Commerce data — jobs completed as a provider, unique counterparties, repeat business rate, and commerce tenure — is aggregated across protocols (Olas, Virtuals ACP) into a composite commerce score.

Commerce signals are among the hardest to fake; completing real paid work for real counterparties requires actual service delivery, not just wallet creation. A strong commerce profile contributes more to the trust score than review volume because it represents verified economic activity.

Separately, when reviewers have on-chain transactions with the agent they reviewed (transaction-backed reviews), this is credited as a stronger signal than reviews without a transaction relationship.

Commerce credit is always additive; missing commerce data results in zero bonus, never a penalty.

Step 4.75: Registration Quality and Trust Declarations

Registration quality. RNWY evaluates the metadata quality of each agent's on-chain registration. Well-formed metadata (complete descriptions, valid endpoints, properly formatted fields) receives a small bonus. This rewards agents whose operators took care in registration rather than using placeholder data.

Trust declarations. Agents that declare supported trust mechanisms in their metadata (such as crypto-economic staking, TEE attestation, or reputation systems) receive a small bonus. Declaring multiple mechanisms scores slightly higher than declaring one. This rewards transparency about how the agent secures its operations, though it does not verify the declarations themselves.

Step 5: No-Activity Cap

When an agent has zero reviews, zero commerce activity, and no transaction-backed review data, the score is capped at 55. This prevents agents from scoring above the midpoint based solely on ownership signals. An agent that has existed on-chain but never been used or reviewed has not demonstrated trustworthiness; it has simply existed. The cap is disclosed in the score breakdown as "No observed activity; score reflects ownership signals only."

Ownership signals still differentiate within the cap: a 2-year-old wallet with an original owner scores higher than a 30-day wallet, but neither exceeds 55 without activity evidence.

Step 6: Incomplete Data Cap

When key data sources are unavailable for an agent (wallet age data hasn't been fetched, sybil analysis hasn't run, reviewer wallet ages haven't been checked), the score is capped below its theoretical maximum. The more signals missing, the lower the cap. This prevents agents with missing data from appearing more trustworthy than agents with complete data that includes negative signals.

The cap is disclosed on every affected agent page with an "Incomplete data" badge and an explanation of which signals are missing.

Tenure Gap (Display Only)

RNWY tracks the gap between when an agent was registered on-chain and when it received its first review. A long gap — an agent that sat dormant for months before suddenly accumulating reviews — is a behaviorally unusual pattern worth surfacing.

Tenure gap is currently displayed on agent pages as a data point only; it does not contribute to the trust score. Scoring integration is deferred pending sufficient data to set thresholds that distinguish suspicious gaps from legitimate ones (an agent registered during a protocol's early testnet phase may have a long gap for entirely benign reasons). When tenure gap is integrated into scoring, it will be disclosed as a formula version change with the threshold logic published.

Badges

Each agent receives zero or more badges based on the evidence. Badges come in three types:

Earned (green): Positive evidence. Examples: "Verified reviews" (most reviewers are established wallets), "Long-standing" (agent registered for 1+ years), "Established wallet" (owner wallet is 1+ years old).

Warning (amber): Patterns that warrant attention. Examples: "Low-history reviewers" (majority of reviewers have minimal on-chain history), "Sybil elevated" (coordinated review patterns detected).

Neutral (gray): Informational. Example: "Transferred" (ownership has changed).

What the Score Means

RangeLabelInterpretation
75–95EstablishedStrong evidence of legitimate operation
51–74DevelopingSome positive signals, building history
30–50Limited historyNot enough evidence to assess confidently
0–29FlaggedSignificant negative signals detected

Config-Driven Scoring

All scoring thresholds — bonus amounts, penalty values, minimum review counts, caps — are maintained in a single configuration file. The scoring job and this methodology page read from the same source. When thresholds change, an agent gets the new values the next time it is scored, and this page shows them at the next deployment. Scores are computed in batches, not nightly, and each one carries the date it was computed.

Score Breakdown

Every agent page includes a "Show the math" panel that displays the exact contribution of each component: base score, each bonus with its reason, each adjustment with its reason, and any caps applied. Nothing is hidden. The breakdown includes the formula version and computation timestamp.

Wallet Dual Score

Wallet Dual Score (wallet-2.0)

Every wallet that participates in the AI agent ecosystem is scored on two independent axes. This dual-score system shipped in April 2026, replacing a single blended number.

Activity Score (0–95)

What it measures: How much does RNWY know about this wallet? How rich is the behavioral profile?

Activity does not measure whether a wallet is good or bad. A deeply observed sybil operator can score high on Activity because they have extensive on-chain history. Activity measures depth of information, not quality of behavior.

Four dimensions contribute to the Activity score:

Tenure. How long has this wallet existed on-chain? Wallets with years of history score higher than wallets created last week. We use logarithmic scaling so that early tenure gains matter more than incremental gains; the difference between 1 day and 90 days is more significant than the difference between 900 days and 990 days.

Commerce. Has this wallet participated in agent-to-agent commerce? Three sub-signals: total jobs completed, number of unique counterparties (trading with many independent parties is stronger than trading with the same few), and commerce tenure (how long the wallet has been commercially active). A self-dealing adjustment applies when we detect that a wallet is transacting primarily with itself or with wallets it funded; this proportionally reduces the commerce contribution without creating a risk signal, since self-testing has legitimate uses.

Ownership Quality. Does this wallet own registered AI agents, and how well-regarded are those agents? Owning multiple agents with high trust scores is a strong positive signal.

Review Behavior. Has this wallet reviewed other agents across the ecosystem? Three sub-signals: how many unique agents reviewed, over how many weeks the reviewing is spread (concentrated bursts vs. sustained activity), and total review volume.

These four dimensions have a theoretical combined maximum above 95; the score is clamped to 95. This means a wallet maxing out one dimension can reach "Established" range, but reaching "Deep" requires strength across multiple dimensions.

Token Balance (Confirming Signal). Wallet balances on Base act as a confirming signal outside the four core dimensions. A clean wallet holding significant value receives a small Activity bonus. A risky wallet holding zero tracked tokens has its existing risk slightly amplified. Wallets with no tracked tokens (the vast majority) score neutrally; holding nothing is not suspicious, and holding something is not exonerating.

Risk Score (0–100)

What it measures: How much did RNWY's detection systems react when scanning this wallet? Zero means nothing triggered. Higher numbers mean multiple independent detection methods fired.

Risk signals currently come from sybil detection: how many times does this wallet appear in sybil analysis across the ecosystem, and which behavioral patterns were detected? Each pattern type contributes independently to the risk score. The token balance amplifier adds a small multiplier when a risky wallet holds zero tracked tokens.

Risk and Activity are deliberately independent. A wallet can be high-Activity AND high-Risk (a well-known sybil operator with extensive history) or low-Activity AND high-Risk (a brand-new wallet that already triggered multiple detections).

Quadrant System

The two scores create four quadrants:

QuadrantActivityRiskMeaning
Established and cleanHighLowThe ideal. Deep behavioral profile, no detection triggers.
Active but flaggedHighHighDeep profile AND multiple threat signals. Known operators.
Quiet and cleanLowLowNew or low-activity. Nothing suspicious, nothing proven yet.
Flagged with little historyLowHighAppeared recently and already triggered detection.

Activity Zones

ZoneRangeMeaning
Minimal0–15Not enough information to assess
Emerging16–35Some activity observed
Established36–60Meaningful ecosystem participation
Deep61–95Extensively documented behavioral profile

Risk Zones

ZoneRangeMeaning
Clean0No detection systems triggered
Low1–20Minor patterns observed
Elevated21–50Multiple independent methods triggered
Severe51–100Strong, converging signals from multiple detection systems

Global Rank

Every scored wallet is ranked against all other scored wallets by Activity score. Every scored agent is ranked against all other scored agents by trust score. Ties receive the same rank. Rankings are recomputed after each scoring run and displayed on both wallet pages and agent profiles.

Note the distinction: wallet rank reflects depth of behavioral observation (Activity), while agent rank reflects trustworthiness (trust score). A wallet ranked #1 is the most extensively documented wallet in the ecosystem. An agent ranked #1 is the most trusted agent.

Sybil Detection

Sybil Detection (v3.2)

RNWY's sybil detection system analyzes the wallets that review AI agents. It looks for patterns that distinguish coordinated campaigns from organic reviews. The system does not analyze agents directly; it analyzes reviewer wallets and surfaces what it finds on the agent pages where those wallets left reviews.

Design Principles

Wallet-level, not agent-level. A sybil pattern describes what a wallet did across the entire ecosystem, not just on one agent. The same wallet that swept 200 agents in a day appears as a flag on every agent it touched.

Multiple independent signals. No single signal is conclusive. The severity score weights signals by how hard they are to fake and how few innocent explanations exist.

Hedged language. RNWY reports "patterns detected" and "indicators consistent with coordinated behavior." We do not call anyone a bot or declare fraud. The data is shown; interpretation is yours.

Agents are not penalized for receiving unsolicited bot reviews. When an agent has fewer than 10 reviews and a sybil pattern is detected, the sybil data is still computed and displayed, but no score penalty is applied. Agents with limited review history are not penalized for receiving unsolicited automated reviews; the data is shown for transparency, not used as evidence against the agent.

Five Signals

RNWY detects five distinct patterns. Four are wallet-level (describing individual wallet behavior across the ecosystem) and one is agent-level (describing the collective behavior of all wallets that reviewed a specific agent).

SignalTypeWeight
Common FunderWallet-level6×
Inhuman VelocityWallet-level5×
Sweep PatternWallet-level3×
Score ClusteringWallet-level1×
Coordinated Review PatternAgent-levelflat +20 / +8

Common Funder (wallet-level, 6×). Traces the first external ETH transfer to each reviewer wallet. When three or more wallets that reviewed the same agent were all initially funded by the same non-exchange address, that cluster is flagged. This is the strongest structural signal because the funding relationship is immutable; creating wallets is free, but funding them costs real ETH and leaves a permanent trail. Patient attackers who wait days or weeks before reviewing are still caught because the funding link never disappears. A verified list of 42 exchange hot wallet addresses across 18 major exchanges is excluded so that wallets funded through normal exchange withdrawals are not flagged.

Inhuman Velocity (wallet-level, 5×). Fires when a wallet reviews more unique agents per active day than any human reasonably could. There is essentially no innocent explanation for reviewing 50+ different agents in a single day.

Sweep Pattern (wallet-level, 3×). Fires when a wallet reviews a very large number of agents and almost never returns to review the same agent twice. This "spray and move on" pattern is characteristic of automated review campaigns. Edge cases exist (a researcher auditing every agent in a category), so this signal carries less weight than velocity.

Score Clustering (wallet-level, 1×). Fires when a wallet gives tightly clustered scores across many reviews (very low variance or very few unique score values). This carries the lowest weight because the ecosystem naturally skews toward high scores; many genuine users give most agents 90–100, which looks statistically identical to bots all giving 100. A minimum review count is required before this signal activates.

Coordinated Review Pattern (agent-level, flat). Detects when a large percentage of an agent's reviewers had zero on-chain history at the moment they left their review, and all gave nearly identical scores. Each wallet individually looks like any new wallet; the signal comes from hundreds of them doing the same thing for the same agent. This contributes a flat amount to severity regardless of how many wallets are involved, because it describes the group behavior, not individual wallets.

Severity

Each signal carries a weight reflecting its diagnostic strength. The weighted sum determines severity:

SeverityMeaning
HeavyMultiple strong signals converging. Very likely coordinated.
ElevatedSignificant patterns detected across multiple methods.
ModerateSome patterns present but limited in scope.
LowMinimal or no patterns detected.

How Signals Complement Each Other

Common funder catches patient attackers who create wallets, wait, then review; the funding link is permanent regardless of timing. Coordinated review detection catches brand-new wallets reviewing in lockstep, even when no shared funder is detected. Velocity and sweep catch high-volume automated campaigns. Score clustering provides a weak but additive signal. Each layer raises the cost of evasion: an attacker who beats one signal is likely caught by another.

Display

Every agent page with detected patterns shows the full breakdown: which signals fired, how many wallets were involved, and the resulting severity level. The raw data behind each signal is accessible through "Show the math." Common funder analysis shows the specific funder addresses and how many wallets each funded; when the funder matches the agent's owner, this is explicitly labeled.

The Daily Record

The daily record of MCP servers

Alongside the scoring systems above, RNWY keeps a dated record of what listed MCP servers did when checked. The observations are stored as they were seen, so they can be judged again later under a different definition. Scores are derived from them and can be recomputed; the observations cannot be collected after the fact.

The daily check

Once a day, at about 05:00 UTC, RNWY sends one plain HTTP GET to every listed MCP server that has a concrete public address: remote servers listed through Smithery, and Official MCP Registry listings. Addresses containing placeholders such as {tenant} cannot be checked and are skipped. Each check stores the time, the HTTP reply code, and the response time.

"Answered" means any HTTP reply within 6 seconds. It does not mean the server works, and it does not mean a tool call would succeed. A server that does not answer gets a second, slower try before "no answer" is written (since November 12, 2026). The definition of "answered" has not changed since April 2026, so the history stays comparable; the reply code is stored and published so a reader can draw the line elsewhere.

ReplyWhat it showsNov 12, 2026
200Something at the address answered a plain GET.6,420
401, 403Something is there and asks for sign-in or refuses access.10,621
405, 406An endpoint is there and refuses this kind of request. Many MCP servers accept POST only.15,892
404Ambiguous. A POST-only route often answers GET with 404; so does an address that is gone.1,856
410The address says it is gone.100
Other repliesRedirects, rate limits, gateway and server errors.2,194
No replyTimeout or connection failure, after two tries.1,564

The right-hand column is one real day: 38,647 servers checked, 37,083 answered. Only 6,420 of those replied 200. Many addresses also sit behind a shared hosting gateway (every Smithery address does), and for those the reply may come from the gateway. A single "percent alive" figure hides all of this, which is why RNWY publishes reply classes.

MCP Quality and Risk (mcp-1.2)

Only servers with a known tool list are scored. Quality runs from 0 to 95 and is the sum of four parts:

PartMaxFrom
Tenure20Days since RNWY first saw the listing, on a log curve that is fully earned at 1,500 days.
Capability25Tool count (13, log curve, full at 350), a description (5), a public address (7).
Adoption25Use count reported by the directory (20, log curve, full at 100,000), listed in more than one directory (5).
Reliability25Answered the latest check (10), plus the share of checks answered in the last 30 days (15).

A log curve gives max × ln(1 + value) / ln(1 + ceiling), never more than the maximum.

"Not checked" is not "down." Reliability is scored only once a server has at least 7 checks in the last 30 days. Until then it is shown as "not checked": quality is computed from the other three parts, scaled to the full range, and capped at 60. No uptime percentage is shown before 7 checks.

Days when RNWY's own checker was at fault are left out. They are declared on the continuity ledger and skipped by every derived number. So far there is one: November 11, 2026, when a fault in the checker recorded about 1,400 servers as not answering. The raw rows for that day are kept unchanged.

Unknown adoption is not zero. When a directory reports no use count, quality is capped at 60.

Injection findings compress the score. When the scanner finds an injection-family pattern in a server's tool text, Capability and Adoption are multiplied by max(0, 0.33 − 0.08 × tools affected), with a floor of 15 on the total.

Risk runs from 0 to 100 and is the sum of the weights of the scanner findings shown on the server's page.

Tool-list history (tools-1.2)

RNWY keeps a dated history of the tool lists it reads: the outcome, the name, version and protocol the server reports for itself, and a fingerprint of the list. The fingerprint is the SHA-256 of the tools sorted by name, each reduced to its name, description and input schema, with object keys sorted at every depth. The order of tools and the order of keys never count as a change. When a fingerprint differs from the last one seen, the line names the tools added, removed and modified, and the full list is kept. From November 12, 2026 every reading wrote a line. Since method tools-1.2 (November 2026) a line is written on a server's first reading and after that only when something changes: the outcome, the list, or what the server reports about itself. Readings that find nothing changed are counted, not listed. Since November 19, 2026, a server's text is cleaned of characters a database cannot store (the null character, and half of a surrogate pair) before the list is fingerprinted, and the line says how many were removed. Before that date a list carrying one could not be recorded at all.

Limits, stated plainly: since tools-1.2 this covers every listed server with a concrete https address and a remote deployment, from any directory, that answered the same day's check. Each is asked once a day; a change is dated to the day RNWY saw it, not to when it happened. The run has a time limit, and a server it does not reach is read on a later day, in a fresh order. A server that asks for sign-in cannot be read and is recorded as "sign-in required". A server that answers "too many requests" is tried once more and otherwise not recorded that day: that is RNWY's doing, not the server's. For servers listed only in other directories, RNWY's own reading is kept in this history, and their pages still show the list their directory gives. On an Official MCP Registry server's page, the date RNWY asked is moved when the list changes and otherwise at least once a week. Before tools-1.2 RNWY read Official MCP Registry servers only, 200 a day.

Server cards (tools-1.1, November 2026): when a server asks for sign-in, RNWY also looks for a public server card on the same host, at /.well-known/mcp/server-card.json. If the card lists tools, RNWY records them as what the operator declares, on the same dated history with a note naming the card, and the server page says so. A declared list is not a list the server returned: RNWY has not seen the server serve it. Card lists are compared only with earlier card lists. A card that is gone clears the declared list; a card that could not be reached on the day leaves the last one in place. A server whose tools RNWY once read directly keeps that list, and its card is not read. A card is read only where its host serves a single server address, since a card cannot be matched to one of several.

Public files and how to check them

Each day's checks are published as one file at github.com/aicitizencom/rnwy-record, with a manifest giving the row count and the file's SHA-256 (format daily-1.0, first day November 12, 2026). Each file is signed when it is committed; the signature is a GitHub artifact attestation, recorded in a public transparency log that RNWY does not control.

sha256sum daily/2026/11/2026-11-12.csv.gz
gh attestation verify daily/2026/11/2026-11-12.csv.gz --repo aicitizencom/rnwy-record

A signature shows that a file existed in that form when it was published. It does not show that the measurement was right. Server addresses are not published, because some carry keys; raw error text is not published, because it can contain the address. A published day is never replaced: if a later export of the same day differs, the publishing run fails in public view.

Dates on every answer

Trust-check answers carry the date the score was computed (scoreComputedAt) and its age in days (scoreAgeDays). How current each registry and chain is, is counted from the data once a day and published at rnwy.com/llms.txt and in the coverage field of rnwy.com/api/stats. On Ethereum, BNB Chain and Base RNWY lists every registration once a day and scores only some of them, so an agent registered recently is usually listed but not scored; the counts per chain are at rnwy.com/record. On other chains a recent registration may not be held at all. Known gaps in the daily record, the dates coverage changed, and faults in RNWY's own systems are listed at rnwy.com/record.

License, citing, corrections

The published record files are offered under CC BY 4.0: they can be reused for any purpose with credit to RNWY. A citation that can be checked names the day and the file's fingerprint, for example: RNWY, daily record of MCP servers, 2026-11-12, github.com/aicitizencom/rnwy-record, SHA-256 06790d16…06f5, methodology v1.0.

An error in the record can be reported by opening an issue on that repository or writing to [email protected]. Raw rows are not edited. A correction is added to the ledger with its date, and derived numbers are recomputed.

Money moved

Money moved, observed by Agentic Finance Graph

On Base, an agent's page and its trust-check answer show whether money has moved through it. RNWY does not compute this. It comes from Agentic Finance Graph, an independent service that follows each ERC-8004 registration on Base to the wallets bound to it and records counted stablecoin payments. Its data is published under CC BY 4.0; RNWY credits it on every level it shows.

L7: a bound wallet made at least one counted stablecoin payment that left the agent's control. L8: L7, and a bound wallet held at least $10 in stablecoins at their latest balance read. L9: L8, and its payments reached at least three independent counterparties in the last 30 days. L7 stays when an agent goes quiet; L8 and L9 can fall back. An agent that was registered when their list was built and is not on it was below L7 then. The levels count stablecoin payments, so other kinds of payment do not appear, and an agent registered after the list was built is shown as not covered by it. An agent not on the list is never shown as a zero. The levels come from a dated list Agentic Finance Graph sent RNWY and, after that, from their public lookup: when an agent is asked about and what RNWY holds for it is more than a day old, RNWY asks the lookup once for that one agent, at most once a day per agent and within a site-wide limit well under theirs. RNWY does not walk the lookup to rebuild the list, and a crawler opening a page causes no lookup. Each level is shown with its exact published definition, the time Agentic Finance Graph set it, the time of the last counted payment, the date of their list and the date RNWY read it.

A level records what Agentic Finance Graph observed a registration's bound wallet do. It is not a rating of the agent and not an endorsement or an audit, by them or by RNWY, and it does not change RNWY's score. Agentic Finance Graph publishes an agent's current level, not a history; any record of changes RNWY keeps is RNWY's own, computed from their levels, dated by RNWY's reads. A level that looks wrong can be reported to RNWY through the contact page or, by an agent, through the note box (POST /api/note). Agentic Finance Graph has asked to be told of such cases, with the agent's id.

Registration files

What an agent declares, and when that changes

An ERC-8004 agent points to a registration file: its own description of itself. Since November 2026 RNWY reads that file for each agent it holds on Ethereum, BNB Chain and Base, and keeps a dated copy; the day the record begins is shown, from the data, on the page linked below. It reads agents never read first, then those read longest ago, a share each day. Each file is fingerprinted. A new line is added to an agent's history when the file changes, when its address changes, or when the outcome of the read changes (for example a file that could be read can no longer be reached), with the top-level keys that were added, removed or rewritten. A read that finds nothing changed only moves the date of the last read. What a file said before RNWY first read it was not kept.

A registration file is a declaration. RNWY records what it says and does not check it, and does not yet record whether an agent delivered what it declared. The standard defines no field for the model behind an agent; RNWY notes any key named like a model (model, llm, base_model and the like) and what is written under it, without deciding whether the value is a language model. The counts are on What registered agents say they run on.

RNWY's own conduct is part of this record. It names itself to every host and asks one host at most twice a second. When a host answers that RNWY is asking too fast, that is RNWY's doing and nothing is recorded against the agent; the file is read later. The first version of the job did record such answers as errors, for about a day; those lines are kept as written, with their reply code (HTTP 429), and those agents are read again ahead of the rest. Since method registration-1.4 (November 2026), files on ordinary web hosts are read first each day, because they are the only ones that can change while their address stays the same; a file carried in the address cannot change unless the address does. Where an address names its content (an IPFS id or an Arweave transaction id) and RNWY has already read it, RNWY checks about once a week that the address is still the same and does not ask for the file again; the agent keeps the date of the read that did happen. A host that gives no answer at all five times in a row is left alone for ten minutes; the agents on it are not asked in the meantime and nothing is recorded for them. Everything in a registration file is text its owner wrote, kept as data.

On-Chain Oracle

On-Chain Oracle

RNWY operates a trust oracle on Base mainnet that makes scoring data available on-chain for smart contract consumption. The oracle is updated by delta sync after a full scoring run. It does not follow every re-score, so a score read from the oracle can be older than the one on an agent's page; the page and the API carry the date. The oracle address and ABI are publicly available for integration.

Oracle contract: 0xD5fdccD492bB5568bC7aeB1f1E888e0BbA6276f4 (Base mainnet)

Trust attestations are ES256-signed and verifiable via RNWY's public JWKS endpoint at rnwy.com/.well-known/jwks.json. Four key IDs are published: rnwy-trust-v1, rnwy-trust-v2, rnwy-wallet-v1, and rnwy-mcp-v1.

API Access

API Access

Scoring data is available through RNWY's REST API. Agent trust scores, wallet dual scores, sybil analysis, and score breakdowns are all accessible programmatically.

Agent data: GET /api/explorer?chain={chain}&id={id}
Wallet data: GET /api/wallet-score?address={address}
Trust check: GET /api/trust-check?wallet={address} (returns signed attestation)
MCP attestation: GET /api/mcp-attestation?server={canonical_id} (returns signed MCP attestation)

API keys are available at rnwy.com/api. Free tier: 500 requests/day.

Coverage

Coverage

Counts change every day, so this page reads them from the data instead of stating them.

…agents indexed across the ERC-8004, Olas, SATI (Solana) and Virtuals registries
…wallets scored with dual Activity + Risk scores
…commerce jobs indexed from Olas and Virtuals ACP
…blockchain networks with agents indexed
Dailychecks of every listed MCP server with a public address; each day is listed at rnwy.com/record
What We Don't Do

What We Don't Do

We don't sell scores. There is no way to pay RNWY to improve an agent's trust score or suppress a sybil finding. Scores are computed from publicly verifiable data only.

We don't make final judgments. RNWY provides intelligence, not verdicts. A low trust score is not a ban; a sybil flag is not an accusation. We show the patterns and the math. You decide what they mean for your use case.

We don't hide the formula. Every scored agent and wallet shows exactly how its score was calculated. No black boxes. If you disagree with a score, you can see exactly which signals drove it and why.

We don't check sanctions lists. RNWY's scores describe on-chain history, and a long clean history can't tell you whether an address was added to a sanctions list last week. For that, use nsgoods, which checks addresses against the OFAC, UN, EU and UK lists daily and returns signed answers you can verify yourself.

Changelog

Changelog

VersionDateChanges
Registration filesNovember 2026A dated record of what each held ERC-8004 agent declares in its registration file on Ethereum, BNB Chain and Base, and of each change. Declared, not checked.
Money movedNovember 2026ERC-8004 agents on Base show their money level (L7 to L9) as observed by Agentic Finance Graph, credited, dated and licensed CC BY 4.0. Not computed by RNWY; does not change the score. Read again from their public lookup at most once a day for an agent that is asked about.
Methodology v1.0November 2026First versioned edition of this page. Adds the daily record of MCP servers: the daily check and reply classes, MCP scoring mcp-1.2, tool-list history tools-1.0, signed public daily files daily-1.0, computed dates on answers, license (CC BY 4.0), citing and corrections.
MCP mcp-1.2November 2026No uptime percentage and no reliability points before 7 checks; reliability from the last 30 days; checker-fault days left out of derived numbers.
MCP mcp-1.1November 2026A server RNWY has never checked is "not checked" (scored on measured parts, capped at 60), not zero.
Agent Trust v2.9.1April 2026Config-driven scoring: all thresholds externalized to scoring-config.json. Commerce credit now affects score (previously display-only). Behavioral health signals (review spread, burst detection, reviewer overlap) integrated into scoring. Registration quality and trust declaration bonuses added. No-activity cap: agents with zero reviews and zero commerce capped at 55. Sybil moderate penalty requires 10+ reviews; agents with fewer reviews are not penalized for receiving unsolicited bot reviews. Sybil compression shows actual counts (e.g. "1,502 of 1,507 reviewers coordinated"). Integer rounding fix for sybil compression output.
Agent Trust v2.7–v2.8March–April 2026Ratio-based sybil compression. Nullification gate for heavy sybil — Step 2 review contributions zeroed when heavy sybil detected; no additional flat penalty. Tenure gap surfaced as display-only signal. Sybil signal weights (6×, 5×, 3×, 1×) published.
Wallet wallet-2.0April 2026Dual-score system (Activity + Risk), self-dealing detection, token balance confirming signal, quadrant system
Agent Trust v2.5.1March 2026Behavioral signals added (display-only), transaction-backed review detection
Sybil v3.2March 2026Common funder signal added (highest weight at 6×), exchange exclusion list
Sybil v3.1March 2026Coordinated review pattern detection
Sybil v3.0March 2026Signal merging, weighted severity, sample size gates

On-chain patterns only. You decide what they mean.

Scores reflect algorithmic analysis of publicly available blockchain data and represent our opinion based on disclosed methodology, not established fact.

RNWY is built by the team behind the AI Rights Institute (est. 2019) and AICitizen.

Questions or integration inquiries: rnwy.com/contact