io.github.nirholas/3d-agent-mcp

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

• Injection surface

  unscannable25% of grade

  Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

  No tool descriptions, server instructions, prompts, or resources were fetched; nothing to scan for injection.

• Supply chain

  warn25% of grade

  Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

  dependency "@x402/evm" is edit-distance 2 from popular package "ejs" (typosquat candidate); dependency "@x402/mcp" is edit-distance 2 from popular package "yup" (typosquat candidate); dependency "@x402/svm" is edit-distance 2 from popular package "@anthropic-ai/sdk" (typosquat candidate); dependency "@x402/core" is edit-distance 1 from popular package "cors" (typosquat candidate); dependency "@metaplex-foundation/js" is edit-distance 1 from popular package "ws" (typosquat candidate); dependency "@ethereum-attestation-service/eas-sdk" is edit-distance 2 from popular package "aws-sdk" (typosquat candidate)

• Credential hygiene

  warn15% of grade

  How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

  src/tools/pump-snapshot.js: assigns a literal to token (HZ1J…t3 (len 44)) — non-provider secret literal; prefer environment intake. src/tools/sentiment-pulse.js: assigns a literal to token (HZ1J…t3 (len 44)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. src/tools/agenc-client.js) — the recommended intake shape.

• Permission scope

  pass15% of grade

  Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

  Network egress present (src/tools/mesh-forge.js: outbound HTTP/socket call) — expected for most servers and not a finding on its own. No shell-exec, unbounded/sensitive filesystem, or download-and-run primitives found in scanned source. PASS = no static red flag. Static scanning is blind to code fetched or executed at runtime.

• Version behavior

  fail10% of grade

  Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

  vs prior 1.2.1 (patch bump): new capability primitive(s) in source: http-client; new hardcoded outbound host/BCC destination(s): ..., api.avax.network, api.dexscreener.com, api.mainnet-beta.solana.com, api.replicate.com, arb1.arbitrum.io, arbitrum-one-rpc.publicnode.com, avalanche-c-chain-rpc.publicnode.com, base-rpc.publicnode.com, base.llamarpc.com, bsc-dataseed1.binance.org, bsc-rpc.publicnode.com, cloudflare-eth.com, docs.agenc.tech, eth.llamarpc.com, ethereum-rpc.publicnode.com, facilitator.payai.network, forno.celo.org, frontend-api-v3.pump.fun, github.com, lite-api.jup.ag, mainnet.base.org, mainnet.helius-rpc.com, mainnet.optimism.io, optimism-rpc.publicnode.com, polygon-bor-rpc.publicnode.com, polygon-rpc.com, replicate.delivery, rpc.ankr.com, rpc.linea.build, rpc.scroll.io, sns-api.bonfida.com, solana-rpc.publicnode.com, three.ws.

• Transport config

  unscannable10% of grade

  Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

  local stdio server: no remote transport or auth surface, and no phone-home/telemetry signal found in source. Absence of transport findings is NOT a safety guarantee — phone-home hosts can be constructed at runtime and defeat static analysis.