AI/agent tooling and MCP/CLI/dev workflow support

Comparison updated: 2026-11-03

Coverage: 10/15 providers have at least one evidenced comparison dimension; 2/15 have evidence in every dimension.

> Evidence semantics: “Publicly evidenced” means a direct vendor page captured on or before the comparison date explicitly supports the cell. Not verified in monitored public evidence means the monitored sources are insufficient; it does not mean the capability is absent. Product availability, limits, and implementation quality were not independently tested.

How to read this comparison

  • The same four buyer-relevant dimensions are applied to every provider.
  • MCP and agent-tooling cells may point to the same product when an MCP server is the vendor's agent integration; they remain separate because buyers may require protocol compatibility as well as packaged agent workflows.
  • Generic “AI-powered” positioning, Ask-AI search, or an ordinary API/SDK is not enough by itself to prove MCP support or agent skills.
  • Blast API is retained in the monitored universe but is deprecated; capabilities on the redirected Alchemy documentation are attributed to Alchemy, not Blast.
ProviderMCP supportCLI / SDK workflowAgent skills / toolsAI-ready docsDirect sourcesEvidence coverage
QuickNodeNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidencehome, chains0/4
Alchemy✅ Publicly evidenced — MCP server✅ Publicly evidenced — Alchemy CLI✅ Publicly evidenced — Agent Skills✅ Publicly evidenced — Build with AI docsdocs, home4/4
InfuraNot verified in monitored public evidence✅ Publicly evidenced — MetaMask SDK workflowNot verified in monitored public evidence✅ Publicly evidenced — AI-agent documentation indexhome, docs2/4
Chainstack✅ Publicly evidenced — Chainstack MCP✅ Publicly evidenced — open-source SDKs and tools✅ Publicly evidenced — MCP agent integration✅ Publicly evidenced — LLM documentation indexhome, docs4/4
Ankr✅ Publicly evidenced — Agentic RPC MCP serversNot verified in monitored public evidence✅ Publicly evidenced — Agentic RPC and TORPC✅ Publicly evidenced — LLM-friendly docsdocs3/4
MoralisNot verified in monitored public evidenceNot verified in monitored public evidence✅ Publicly evidenced — Open in ChatGPT workflow✅ Publicly evidenced — AI assistant / ChatGPT entry pointsdocs2/4
NOWNodes✅ Publicly evidenced — MCP Servers in plan matrixNot verified in monitored public evidence✅ Publicly evidenced — MCP ServersNot verified in monitored public evidencepricing2/4
dRPC✅ Publicly evidenced — MCP positioningNot verified in monitored public evidence✅ Publicly evidenced — onchain skills for AI agentsNot verified in monitored public evidencehome, blog2/4
Blast APINot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidencedeprecation notice0/4 — service deprecated
BlockdaemonNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidence✅ Publicly evidenced — AI-agent docs indexdocs1/4
TatumNot verified in monitored public evidenceNot verified in monitored public evidence✅ Publicly evidenced — AI Builder✅ Publicly evidenced — AI-agent docs indexhome, docs2/4
Pocket Network / GroveNot verified in monitored public evidenceNot verified in monitored public evidence✅ Publicly evidenced — Agentic Portal and agent-callable servicesNot verified in monitored public evidencePocket Network, Grove wind-down notice1/4
LlamaRPCNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNo reachable direct source in this pass0/4
AllnodesNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceMonitored pages were access-protected0/4
Lava NetworkNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidenceNot verified in monitored public evidencehome, docs0/4

Cross-provider readout

  • Breadth: 10 of 15 providers have at least one directly evidenced dimension.
  • Completeness: Alchemy and Chainstack are the only providers with monitored public evidence across all four dimensions.
  • Dimension coverage: agent skills/tools 8/15; AI-ready docs 7/15; MCP 5/15; CLI/SDK workflow 3/15.
  • No topic-specific evidence in this pass: QuickNode, Blast API, LlamaRPC, Allnodes, and Lava Network. Blast is separately confirmed deprecated; the other gaps must not be read as product-absence claims.
  • Attribution warning: Grove is winding down while Pocket Network continues. Alchemy capabilities surfaced through the old Blast documentation redirect belong to Alchemy.

Buyer guidance

  1. Ask for the exact MCP transport, authentication model, tool schema, supported chains/methods, write-operation safeguards, and rate limits.
  2. Distinguish an MCP connector from packaged agent skills, an AI documentation index, an Ask-AI widget, and general “AI-powered” infrastructure positioning.
  3. Require a live proof for key management, transaction confirmation, spend controls, prompt-injection boundaries, audit logs, and human approval before permitting write actions.
  4. Verify whether CLI/SDK tooling manages provider resources, only calls RPC methods, or supports repeatable agent deployment workflows.
  5. Re-check all volatile availability and plan claims before procurement or sales use.

Next research priorities

  1. Locate product-specific QuickNode agent/MCP/CLI documentation rather than inferring capabilities from generic platform pages.
  2. Obtain direct CLI/SDK workflow evidence for Ankr, NOWNodes, dRPC, Moralis, Blockdaemon, Tatum, Pocket Network, and Lava.
  3. Confirm NOWNodes MCP plan availability and implementation documentation beyond the pricing matrix.
  4. Resolve LlamaRPC reachability and Allnodes' access-protected public evidence through first-party documentation or a manual browser pass.
  5. Test advertised MCP endpoints and agent workflows separately; this matrix verifies public claims, not implementation behavior.

Source freshness

Positive claims above were re-checked against the monitored first-party surfaces on 2026-11-03. Links without a positive claim are supplied only to show the checked surface. Verify before sales use.

Research log (audit trail)
  • 2026-11-03 daily evidence report — normalized the matrix from current direct evidence, removed unsupported carry-forward classifications, preserved the full 15-provider universe, and recorded access/deprecation limitations.
  • Earlier dated reports remain available in the monitoring encyclopedia; they are provenance records, not the current comparison.