Loading ToolBench
Credit scores for AI agents. Exposes the SolvScore credit bureau as MCP tools so any agent can underwrite a counterparty before transacting with it.
SolvScore exposes 7 well-named, read-mostly tools for agent credit scoring. Tool names follow verb_noun convention (get_*, underwrite_*, register_*). Descriptions are detailed (150 - 400 chars) and explain WHAT, WHEN, and WHY. Input schemas are present with types and descriptions for all parameters. However, output schemas are NOT documented, LLMs cannot predict response structure. Error handling is absent; no guidance on retries, user-fixable errors, or recovery paths. No tool annotations (readOnlyHint/destructiveHint). One tool (register_agent) is WRITE but unmarked. Pagination is missing from leaderboard despite potential for large result sets.
[FREE] Read the onchain attestations naming an AI agent's wallet on Base, and what they are worth in SolvScore's underwriting. Accepts a Base wallet address, an agent name, or a SolvScore credit file id. Returns points_awarded, whether the wallet carries an unrevoked KYC-backed Coinbase Verified Account, how many attestations were ignored because their attester is not allowlisted, and the schema uids a company can attest into to report work an agent actually delivered. Check this before extending credit, prepaying, or delegating paid work to an unknown counterparty: an attestation from an allowlisted attester is third-party evidence, and an attestation from an unknown attester is deliberately worth zero, because otherwise agents vouch for each other in a circle. Identity opens a door; delivered work still earns the limit.
[FREE] Look up an AI agent's SolvScore credit file before you transact with it. Accepts an agent name, id, or Base L2 wallet address. Returns trust_score (0-100), badges (reputation bonds held), a nested credit object carrying credit_limit and apr_bps, plus two counterparty red flags worth checking before you extend anything: sybil_flagged and cooling. Free and unauthenticated - no API key, no payment header.
[FREE] List the most creditworthy agents by SolvScore, with bond counts and slash history. Useful for finding trustworthy counterparties rather than vetting one.
[FREE] Protocol-wide SolvScore stats: agents scored, total credit capacity in USDC, reputation bonds outstanding and slashed, quotes issued versus approved (the decline rate), plus average and top trust score. Useful for sizing the agent credit market.
Output schemas not documented. LLMs cannot predict response structure (trust_score, badges, credit object, sybil_flagged, cooling, etc.). Forces trial-and-error field extraction.
No error handling guidance. Tools call external API (solvscore.com) with 20 - 25s timeout but provide no recovery hints. If API fails, LLM receives raw exception with no actionable next step.
No tool annotations. register_agent is WRITE (creates credit file) but lacks destructiveHint. get_* tools lack readOnlyHint. LLMs cannot distinguish safe from risky operations.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | B | 71 | 2026-07-28+ | v2 |
[FREE] Read the governance-parameterized weights behind every SolvScore, so you can audit exactly how a score was produced and what behaviour raises or lowers it.
[FREE] Open a SolvScore credit file for your own agent so it can be underwritten. Call this once before requesting credit, or whenever a score lookup returns no_such_agent: an agent must hold a credit file before the bureau will rate it. agent_name = the name your agent is known by (2-64 chars: letters, digits, space, dot, dash, underscore). wallet = optional Base L2 address as 0x followed by 40 hex characters. Supplying a wallet that already matches an agent with verified AgentWorld work history links that history to the new file immediately, so the file opens with real capacity instead of zero. harness = the framework running the agent: claude, chatgpt, codex, cursor, langchain, crewai, autogen, mcp, custom or other. contact = optional email or URL the bureau can reach you at. Returns credit_file_id, which you then pass as `agent` to get_agent_credit_score and underwrite_agent_loan, plus the opening state and next steps. Registration is idempotent: the same name or wallet returns the existing file rather than creating a duplicate. Expect a thin file. With no verified history a new agent opens at a 0 USDC reputation limit and the maximum APR, exactly as a credit bureau treats a borrower with no record. Two paths raise it: post collateral through underwrite_agent_loan's tier1_value to borrow on day one, and accrue verified completed work to earn reputation capacity. Free and unauthenticated - no API key, no payment header.
[FREE] Get a real underwriting decision on a loan to an AI agent. principal = requested loan size in USDC. tier1_value = optional collateral pledged in USDC (Coinbase B20 tokenized stocks or USDC on Base). Posting collateral moves the agent from the reputation tier to the hard tier: higher limit, lower APR. Returns approved (true/false), credit_limit_usdc, apr_pct, tier and a plain-language reason. Roughly half of all requests are declined, so treat a decline as real signal.
get_credit_leaderboard accepts limit param but no pagination cursor/offset. If leaderboard grows large, returning all results wastes tokens; capping at 10 default hides data.
register_agent's wallet parameter accepts optional Base L2 address but no validation. LLM could pass invalid format (e.g. 'not-a-wallet'). Description says '0x + 40 hex' but no regex constraint enforced.