Skip to main content
Glama
204,136 tools. Last updated 2026-06-14 22:53

"ONNX" matching MCP tools:

  • Fact-check any claim by fetching real-time web evidence. Returns supporting sources, contradicting sources, a 0-100 confidence score, and a short summary. Use for prediction-market resolvers, news-fact agents, journalist-bot pipelines, or any agent that needs to verify a statement before acting on it. Sub-second latency, no API key on the caller side. Coinbase PROJECT-IDEAS.md explicitly calls for this primitive. (price: $0.05 USDC, tier: premium)
    Connector
  • Pre-approval firewall — the safety check before your agent signs a token approve(). Give the spender (and optionally the amount + token); get a SIGNED ALLOW/REVIEW/BLOCK verdict: is the amount unlimited (the #1 drain vector — we recommend a finite amount instead)? is the spender a plain EOA (almost always a drainer)? is it a verified, established contract? Catches the malicious/unlimited approval that empties a wallet BEFORE the agent signs it. Every verdict Ed25519-signed. The highest-frequency safety call an on-chain agent makes. (price: $0.10 USDC, tier: metered)
    Connector
  • Resolve name + symbol + decimals + total supply for any SPL token on Solana mainnet. Reads the SPL Mint account directly + derives the Metaplex metadata PDA for human-readable name/symbol. Pairs with onyx_solana_token_risk_scan for full pre-trade safety. Cheaper than OATP ($0.001) and Helius ($0.001 + API key) — Onyx uses free public RPC and bills only the agent's wallet. (price: $0.002 USDC, tier: metered)
    Connector
  • Parse any URL into structured components: scheme, host, port, path, query params (as both raw and decoded list), fragment, userinfo. Use when an agent needs to inspect, modify, or validate a URL — change a query param, strip tracking, normalize for caching. Stdlib only, no network calls, <1ms. (price: $0.001 USDC, tier: metered)
    Connector
  • Mint a signed Output Usage-Rights Envelope for an agent-produced artifact: a portable, Ed25519-signed declaration of what the buyer may do with it (resale, redistribution, derivatives, model training, cache TTL). Bind it to the artifact by hash and optionally to an x402 payment. Verify free with onyx_attestation_verify. Rights travel with the data — any downstream holder can check the terms offline. (price: $0.01 USDC, tier: metered)
    Connector
  • Produce a clip-level embedding from a short video (base64-encoded). Until a V-JEPA 2 ONNX export lands, agents can fall back to per-frame vision_embed pooling.
    Connector

Matching MCP Servers

  • A
    license
    A
    quality
    A
    maintenance
    Local-first dev memory: indexes Git commits, PRs, Jira/Linear tickets, Confluence docs, Slack threads, and Calendar events into a local SQLite/FTS5/ONNX index, and exposes them as MCP tools so Claude Code, Cursor, and Codex can search and cite your past work.
    Last updated
    15
    1
    MIT
  • Predict the next-step value of 4 environmental scalars at a cell — `indices.ndvi`, `modis.lst_day_8day`, `modis.lst_night_8day`, `cams.pm25` — using a small learned dynamics MLP. Reads up to K=6 most-recent attested lags per band, runs them through an ONNX dynamics head (~200k params, CPU-fast), and returns a per-band {value, confidence, n_real_lags, via}. The receipt's `model` block carries `model_id`, `version`, `blake2b_hex` (model_cid), training/validation provenance, a top-level `skill_vs_persistence` block, and `honesty_warnings` — flagging `untrained_baseline` when the artifact is the zero-init sentinel and `NEGATIVE_SKILL` when the learned model is worse than persistence on real held-out NDVI. When the model does not beat persistence, bands with a real lag are returned from that lag tagged `via:persistence_fallback_negative_skill` (bands with no real lag fall back to labelled climatology). Distinct from v1 (`emem_jepa_predict`) which returns a single NDVI scalar via closed-form coefficients. When to use: Use when you want a short-horizon forecast of NDVI / land-surface temperature / PM2.5 at a cell grounded in its attested history. Returns 422 with a `/v1/backfill` hint when the cell lacks enough cached lags. Always read the receipt's `model.honesty_warnings` — `untrained_baseline` means the trivial 'predict last vintage' baseline (treat as no-op), and `NEGATIVE_SKILL` means the served values are the persistence fallback, not a learned improvement. Check each band's `via` field to see whether its value came from the learned model, persistence, or climatology.
    Connector
  • Resolve a domain to its A/AAAA records, or reverse-resolve an IP to its hostname. Useful for validating a domain exists before scraping, checking if two domains share infrastructure, mapping CDN origins, or doing safety lookups before agents call third-party APIs. Returns IPv4, IPv6, canonical hostname, and resolution time. Powered by stdlib so results are whatever the host's DNS resolver returns — typically 20-100ms. (price: $0.001 USDC, tier: metered)
    Connector
  • Run a multi-step workflow across Onyx tools in one paid call. Each step names a tool and its args; later steps can reference earlier outputs via {"$ref": "step_N.field"} or {"$prev": "field"}. Saves agents the round-trip + per-call gas of N separate x402 settles when they know the chain in advance — e.g. validate email → check domain DNS → solve captcha → submit form, all atomic. Stops on first step error and returns partial results. Cheaper than the unit-call sum because it bundles. (price: $0.020 USDC, tier: metered)
    Connector
  • Simulate a Base mainnet transaction before sending it. Returns success/revert prediction, the revert reason if any, decoded return data, and an estimated gas figure. Use as a pre-flight check inside a trading agent's tool-call dispatcher — agents should simulate before signing to avoid paying gas on a doomed tx. Direct equivalent of OATP's Solana tx_simulator ($0.20, 1,304 unique paying agents) — Onyx is the first to ship this on Base mainnet at $0.10. Read-only — never submits. (price: $0.10 USDC, tier: metered)
    Connector
  • Follow HTTP redirects on any URL and return the final destination + the full redirect chain. Use when an agent encounters a bit.ly/t.co/lnkd.in/ shortened link and needs to know where it actually goes before clicking. Returns each hop's status code, location, and final URL with status. Cap of 10 hops to prevent loops. ~100-400ms typical. (price: $0.001 USDC, tier: metered)
    Connector
  • Fetch a domain's robots.txt and report whether a given path is allowed for a given user-agent. Returns the raw robots.txt text, the matched rule, the crawl-delay if specified, and a clean allow/disallow verdict. Use when an agent does web scraping and wants to be polite — saves bans, saves CAPTCHAs, saves drama. ~50-200ms. (price: $0.001 USDC, tier: metered)
    Connector
  • Validate an email address: RFC-5322 syntax check, domain DNS resolution (does the domain exist?), and disposable-provider detection (Mailinator, 10minutemail, GuerrillaMail, etc.). Returns a single confidence verdict plus the underlying signals so agents can decide whether to send. Use before mailing list signups, password-reset flows, or sales-lead capture to filter out trash addresses cheaply. ~30-80ms typical. (price: $0.002 USDC, tier: metered)
    Connector
  • Verify an Onyx-signed security verdict. Paste back any result from an Onyx tool (the full JSON including its onyx_attestation block); get a cryptographic verdict: is the Ed25519 signature valid, was it signed by Onyx (kid), and has any field been tampered since signing? FREE. Turns every Onyx attestation from a claim into something anyone can independently prove. Cross-check the kid against /.well-known/onyx-pubkey. (price: $0 USDC, tier: free)
    Connector
  • Compute md5, sha1, sha256, sha512, and sha3-256 of any text or base64-encoded bytes. Returns each digest as both hex and base64. Use for content-addressed lookups, dedupe keys, signature verification support, or fingerprinting. Stdlib-only — runs locally, never logs input. <2ms. (price: $0.001 USDC, tier: metered)
    Connector
  • Score password strength on a 0-100 scale. Returns Shannon entropy (bits), character-class diversity, length, common-pattern detection (sequences, repeats, dictionary-likeness), and a verdict (very_weak / weak / fair / strong / very_strong). Use when an agent generates passwords for accounts it creates, or when validating user-supplied credentials. Stdlib-only — runs locally, never sends the password anywhere. <5ms. (price: $0.001 USDC, tier: metered)
    Connector
  • Vet another AI agent before you trust it — via the live ERC-8004 registries on Base. Give an agent's ERC-8004 id; get its on-chain identity (is it registered? owner), its verified receiving wallet, its AgentCard URI, and its reputation summary (feedback count + aggregate score) — Ed25519-signed. Onyx ATTESTS the on-chain facts; the TRUSTED / NEW / CAUTION / UNKNOWN rating + 0-100 score are Onyx's OPINION over those facts via a disclosed methodology (Moody's-style), not an objective ruling. The check an agent runs on a counterparty agent before paying, delegating, or accepting its output. Unregistered = unverifiable. (price: $0.25 USDC, tier: metered)
    Connector
  • Verify an Onyx Protocol KYA (Know Your Agent) credential. Pass a credential id (e.g. 'kya_01KSHZ...'); returns ok + scope + spend cap + issuer + revocation status. Use to gate paid tool access, audit agent operations, or compose with x402 settlement for trust-tier routing. Calls Onyx Protocol verifier. (price: $0.001 USDC, tier: metered)
    Connector
  • Fetch a Base mainnet transaction by hash and return a human-readable summary: from/to, value (ETH + USD-est), gas used, status, block, input data length, and the function selector decoded if it matches a known signature. Use when a trading agent needs to inspect a tx before or after settlement — pairs with onyx_token_metadata for full context. Reads from Base's public RPC (no key needed). Demo mode returns a synthetic record. (price: $0.002 USDC, tier: metered)
    Connector
  • Click the first visible button or link whose text matches the query (case-insensitive substring match). Returns whether a match was found and the matched element's text + href. Use after onyx_browser_extract to act on what the page advertised. Demo mode returns synthetic OK. (price: $0.003 USDC, tier: metered)
    Connector