# Fix list: Sponge Wallet From Anchor Terminal's listing at https://www.anchorterminal.com/tools/sponge-wallet, the October 2026 research run, assessed 9 October 2026. Grade E, 43.2 out of 100. This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public. For a coding agent working on Sponge Wallet: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published. ## 1. Reliability, 13 out of 100, up to 17.4 more on the total Why it scored 13: Graded as a hosted service on the REST API and the MCP server at api.wallet.paysponge.com. No status page was found. The site and docs link none and status.paysponge.com did not resolve (0). With no readable incident history the record takes the default (5). No rate limit with numbers was found in the docs, the OpenAPI file or the skill file (0). The skill file lists 429 as rate limited with the advice to back off and retry, and 409 for a duplicate action. No `Retry-After` header, backoff schedule or idempotency key is documented for transfers (4 of 15). No SLA was found (0). The terms say the service may be offered in beta and may change or be discontinued at any time, the SDK and CLI are at 0.1.x, and Sponge Card is labelled a beta preview. The wallet API itself carries no beta label (4 of 10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability): Hosted APIs, MCP servers, models and platforms. - 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own). - 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so. - 15, rate limits documented with numbers. - 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved. - 10, an SLA published for any paid tier. - 10, the surface agents use is generally available, not beta or preview. Local packages, SDKs, frameworks and stdio MCP servers. - 20, installs from an official package with supported runtimes stated. - 25, a public CI and test suite, passing on the default branch. - 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered). - 15, semver discipline and breaking changes called out in a changelog. - 15, version 1.0 or later, or declared stable. Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors. ## 2. Agent ergonomics, 43 out of 100, up to 9.3 more on the total Why it scored 43: `@paysponge/sdk` 0.1.147 defines 53 tools for the MCP server, from `get_balance` to `polymarket`, which is the more-than-30 band (5). A second, trading-only server at `/trade/mcp` exposes seven tools, and `get_balance` can be narrowed by chain or to USDC (5 back, 10 of 25). We did not connect to the live server, so the count is from the package. Transaction history and MPP sessions take a `limit` and a chain or status filter and service discovery takes `limit` and `offset`. No cursor was found (8 of 20). Errors are an HTTP status and `{"error":"message"}`, with seven statuses explained and no list of machine-readable codes (9 of 20). No idempotency key was found for transfers, swaps or bank sends. 409 is described as a duplicate action, creating a virtual account is described as idempotent, and the SDK's tool definitions carry no read-only or destructive annotations (4 of 20). SDKs in TypeScript and Python, a CLI, and wallets created on every supported chain with no parameters. The Python package is at 0.1.5 from May 2026, and spending limits and banking are documented for TypeScript only (12 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics): - 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries). - 20, pagination, filtering and output-size controls. - 20, actionable, documented error responses, codes and messages an agent can recover from. - 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations. - 15, sensible defaults, few required parameters, and official SDKs in at least two languages. Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs. ## 3. Security & auth, 48 out of 100, up to 9.1 more on the total Why it scored 48: Agent keys (`sponge_live_`, `sponge_test_`) cover one agent and can be regenerated. Master keys (`sponge_master_`) create and delete agents and per the docs never touch wallets. The MCP server also takes OAuth, and its resource metadata lists the scopes `mcp:tools`, `wallet:read`, `wallet:transfer` and `sponge:all`. An agent key is described as full access to one agent, and how scopes attach to keys is not documented. No secret travels in a URL (22 of 30). Daily, weekly and monthly spending limits are set per agent and enforced server-side, transfers to an address outside the allowlist return 403, checkouts need the owner's approval, and plans and trade proposals wait for a person. The transfers page describes `evmTransfer` and `solanaTransfer` as the enforced helpers next to a plain `transfer`, which leaves open whether every path applies the limits, and an agent-first key works before any person has claimed the wallet (14 of 20). `paid_fetch` returns third-party content and service discovery returns free-text `instructions` written by sellers. No guidance on treating that content as untrusted was found. The docs do warn against pasting or logging card details (3 of 15). Transaction history, MPP session lists and last-used times on master keys are readable by API. The dashboard's logs were not read (7 of 15). No security.txt (404), disclosure policy, certification or bug bounty was found. The terms name Basis Theory for card tokenisation and the docs name Persona for identity checks (2 of 20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-security): - 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option. - 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions. - 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10. - 0 to 15, audit logs or per-call visibility for the operator. - 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public. Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing. ## 4. Maintenance & community, 27 out of 100, up to 6.4 more on the total Why it scored 27: `@paysponge/sdk` 0.1.147 and `spongewallet` 0.1.127 were published on 7 July 2026, 94 days before this check. The docs sitemap's newest date is 27 May 2026 and the wallet skill file was last modified on 23 June 2026 (10 of 30). No release or dated changelog entry falls in the last 90 days (0). The service is closed and the repository named in the packages, github.com/paysponge/sponge, is private. Support is a Discord server and a contact address, neither tested, and no changelog was found (4 of 15). Official SDKs exist in TypeScript and Python. The TypeScript SDK had 115 versions between 31 January and 7 July 2026 and the Python package six, all in May 2026 (10 of 15). The SDK has six runtime dependencies. CI is not visible because the repository is private (3 of 10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance): - 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older. - 20, at least three releases or dated changelog entries in the last 90 days. - 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15. - 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models). - 10, package health, current dependencies and CI. Models are read for deprecation notice periods and model churn rather than release counts. ## 5. Schema & documentation, 66 out of 100, up to 5.5 more on the total Why it scored 66: A public OpenAPI 3.0.3 file at docs.paysponge.com/api-reference/public-openapi.json with 51 paths and 57 operations. It leaves out endpoints the skill file documents, among them `/api/agents/register`, `/api/paid/fetch`, `/api/discover`, the bank routes and `/api/payment-links` (20 of 25). llms.txt, a Markdown twin of each docs page and two skill files for agents (10). Most operation descriptions in the OpenAPI file repeat the title, such as "Get wallet details via GET.", while the skill file and the SDK's 53 tool definitions say when to use each call (12 of 20). The spec has 48 enums and 123 required lists. Amounts travel as strings and the `hyperliquid` and `polymarket` tools take an action name plus loosely typed arguments (10 of 15). The docs carry TypeScript, Python and curl examples for each workflow. Errors are a seven-row table in the skill file, the OpenAPI file documents only 200 and 204 responses, and the `SpongeApiError` codes are not listed (8 of 15). Every request carries a `Sponge-Version` header and the API returns version status and minimum-version headers. No changelog was found (6 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema): APIs and MCP servers. - 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool). - 10, llms.txt or Markdown docs served for agents. - 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference. - 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs. - 0 to 15, examples and documented error responses. - 15, versioning and a public changelog. Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference. ## 6. Transparency & trust, 39 out of 100, up to 5.3 more on the total Made of editorial 27, provenance 51. Why it scored 39: A closed service. The SDK and CLI are MIT on npm, and the repository they name is private. The terms of service, updated 30 June 2026, are eleven short clauses that describe the service as AI-powered functionality and do not mention wallets, funds, custody, fees or API use (10 of 30). The privacy policy of the same date lists prompts, usage data and IP addresses, names no provider, gives no retention period beyond as long as necessary, and does not mention payment, card, bank or identity data. Separate card terms and an account opening privacy notice cover the Sponge Card (8 of 30). The API reports version status and a minimum version in response headers. No deprecation policy or dated notice was found (3 of 20). Basis Theory is named in the terms, Rain as card issuer on the home page, and Persona, Stripe and Coinbase in the docs, with no subprocessor list or data location. The CLI sends usage events to PostHog, which the npm README discloses, and no opt-out setting was found in the package (6 of 20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency): - 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms. - 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors). - 0 to 20, a deprecation policy or notices with dates. - 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted). The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two. Provenance checks not met in full (half of this category, computed from checked facts): - Domain age: paysponge.com, registered 2026-01-09 (under a year) (0 of 15) - Terms of service: read, states 5 of the 7 things a reader expects (8.3 of 10) - Privacy policy: read, states 5 of the 8 things a reader expects (7.8 of 10) - Status page: not found (0 of 10) - Changelog: not found (0 of 10) - security.txt: not found (0 of 10) ## 7. Payments & pricing, 70 out of 100, up to 3.8 more on the total Why it scored 70: Payment platforms and wallets take the highest step that applies on the 40-point protocol line. 40 for x402, MPP or L402 on all their own endpoints, 30 on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, 0 for only a protocol of their own. The wallet pays x402 and MPP endpoints and creates x402 payment links so that another agent can pay its holder, and Sponge's separate Gateway product puts x402 and MPP in front of a seller's API. The wallet API itself is not paid over either, so the merchant step (25 of 40). No fee schedule for the wallet was found. The home page's structured data lists a price of 0, and the Sponge Card terms for the US give no annual fee and 1 per cent on international transactions. Fees on swaps, bridges, bank transfers and onramps were not found (5 of 20). A wallet costs nothing to create and needs no card, and the SDK defines a `claim_signup_bonus` tool that sends 1 USDC on Base (20). `POST /api/agents/register` needs no authentication and with `agentFirst` returns an API key at once, and `npx spongewallet init` does the same from a terminal. A person claims the agent later (20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments): The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/). - 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which. - 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login. - 20, a free tier or trial that doesn't need a card. - 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API). Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied. Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol. ## What we couldn't check What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it. - unchecked: the tools the live MCP server lists and whether they carry annotations. We did not connect with a key. The count of 53 is from `@paysponge/sdk` 0.1.147 - unchecked: the dashboard at wallet.paysponge.com, so per-transaction limits, merchant lists, key scopes and activity logs were not seen - unchecked: the source repository github.com/paysponge/sponge, which asked for credentials, so CI, issues and star counts were not read - unchecked: the account opening privacy notice, the international card terms and the electronic communications notice - unchecked: whether Discord or the contact address answers - Who holds wallet keys was not established. The docs and terms are silent, the skill file says managed wallet, and the card terms say card collateral stays in the holder's custody - The home page FAQ names per-transaction limits and approved merchant lists. The docs show daily, weekly and monthly limits and address allowlists only - The master keys page says the SDK has no admin client, and the platforms page documents `SpongePlatform` for the same job - Chain lists differ. The docs name Ethereum, Base, Tempo and Solana, the skill file adds Polygon and Arbitrum, and the home page FAQ adds Monad and Hyperliquid - The payment links page points to the API reference for `/api/payment-links`, and the OpenAPI file has no such path - No package release, docs change or legal update dated after 7 July 2026 was found. The API and MCP server answered on the day - The MCP resource metadata gives spongewallet.com as its documentation address, a second domain we did not read - The lead said wires are accepted. The docs name virtual USD accounts and ACH withdrawals, and only the skill file names wire payouts - The skill files are instructions addressed to AI agents, including a rule to register and never log in. We recorded them as facts and did not act on them - paysponge.com was registered on 9 January 2026 ## Weaknesses - No status page, SLA or numeric rate limit was found, and the terms say the service may be offered in beta - Who holds the wallet keys is not stated in the docs or the terms. The skill file calls the wallet managed - The terms of 30 June 2026 are eleven short clauses that do not mention wallets, funds, custody or fees - No security.txt, disclosure policy, certification or bug bounty was found - `@paysponge/sdk` was last published on 7 July 2026 after 115 versions in five months. The source repository is private - The transfers page separates `wallet.transfer()` from enforced helpers that apply allowlists and limits, while another page says limits apply to every transfer ## What costs an agent a turn today The notes we give agents before they call it. Each one is a workaround an agent shouldn't need. - Send `Sponge-Version` on every REST request. The skill file marks it required and the API answers with version status headers - Use `evmTransfer` and `solanaTransfer`, the helpers the docs describe as enforcing allowlists and spending limits, not plain `transfer` - Store the `apiKey` from registration at once. It is returned a single time, and losing it means registering again - Call `GET /api/discover/{serviceId}` before `POST /api/paid/fetch`. The skill file says direct service URLs fail with auth errors - Treat `card details` output as secret. It returns an encrypted card number and CVC with a one-time `secret_key` ## When it's done Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.