Atlas / MCP servers / superglue-ai / Superglue

SuperglueBLOCK

Name: Superglue
Author: superglue-ai

mcp/superglue-ai/superglue

MCP server by superglue-ai · agentic workflows · agents · ai · Audited 2026-09-23

superglue (YC W25) builds integrations and tools from natural language. Get production-grade tools for long tail and enterprise systems.

Verdict

BLOCK

Grade

Trust score

40 /100

Exposed tools

20 15r · 5w · 0d

Transport

streamable-http

License

NOASSERTION

Stars

2,062

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

superglue's AI agents connect, migrate and implement enterprise systems. Cloud-hosted or on your own infrastructure.

[](https://www.ycombinator.com/companies/superglue) [](https://www.npmjs.com/package/@superglue/client) [](https://hub.docker.com/r/superglueai/superglue) [](https://app.workweave.ai/reports/repository/org_0S2o9PLamHvNsTjHbszc38vC/914997268)

What is superglue?

superglue learns how your systems work from your company's knowledge. It performs the implementation that normally requires human coordination and engineering work.

Implement NetSuite, Sage Intacct, SAP, Business Central or Acumatica in days. Agents map and migrate legacy data, configure the system, and keep data in sync post go-live.
Connect ERP, CRM, databases, and internal systems to AI platforms like Claude. Create governed data access for AI agents and track data usage across your org.
Connect customer systems, import historical data and get new integrations live in hours. Let agents manage the entire implementation process end-to-end.

Some example usecases

Read from source at commit 862b45b37adcOBSERVED · 2026-09-23

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.

claude-code

claude mcp add web --env AI_GATEWAY_API_KEY=${AI_GATEWAY_API_KEY} --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env AUTH_TOKEN=${AUTH_TOKEN} --env AWS_ACCESS_KEY_ID=${AWS_ACCESS_KEY_ID} -- npx -y @superglue/[email protected]

claude-desktop

{
  "mcpServers": {
    "web": {
      "command": "npx",
      "args": [
        "-y",
        "@superglue/[email protected]"
      ],
      "env": {
        "AI_GATEWAY_API_KEY": "${AI_GATEWAY_API_KEY}",
        "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
        "AUTH_TOKEN": "${AUTH_TOKEN}",
        "AWS_ACCESS_KEY_ID": "${AWS_ACCESS_KEY_ID}"
      }
    }
  }
}

Exposed tools (20)

15 read · 5 write · 0 destructive.

Tool	Risk	Description
`Admin`	read	Full access to all tools and systems in OSS.
`Member`	read	Standard role placeholder kept for RBAC compatibility in OSS.
`authenticate_oauth`	read	Initiates OAuth flow for a system. Credentials (client_id/secret) must already be stored on the system. On success, tokens are auto-saved.
`build_tool`	read	Builds a new superglue tool by accepting the full tool configuration JSON. Successful builds are auto-saved — use the returned toolId for all subsequent operations. In the tool playground, builds remain draft-only until explicitly saved.
`create_system`	write	Creates and saves a new system. If credentials are provided, a confirmation UI appears that lets users review credentials you have provided and enter missing values. Leave empty for auth-free systems.
`custom_tool`	read	A custom tool
`edit_system`	write	Edits an existing system. Provide only the fields to change — omitted fields are preserved. Cannot remove existing documentation, only append via files field.
`find_role`	read	Look up a saved role by ID, or list all roles. Returns the persisted role configuration (not the current UI draft). Use inspect_role for the current draft state.
`find_system`	read	Look up an existing system by ID or search by query. Also returns system knowledge (OAuth config, documentation URL) for systems not yet created. Use
`find_tool`	read	Look up an existing tool by ID or search for tools by query. Use
`get_runs`	write	Fetches recent run history for saved tools. Draft executions (build_tool, run_tool with draftId) do NOT create runs — errors are in the tool result directly. Set fetchResults=true only when investigating, not when listing runs.
`inspect_system`	read	Inspect specific parts of the current system editor state in detail. Unlike find_system (which looks up the saved server-side system), this inspects the current unsaved system editor state in the sidebar.
`limited_tool`	read	A tool with limited uses
`load_skill`	read	Loads superglue skills into context. Available skills: ${skillIndexDescription} Some skills include additional tools that become available after loading.
`run_tool`	write	Executes a tool — either a draft (by draftId) or a saved tool (by toolId), not both. Keep JSON payload data in payload and bind uploaded files separately in files. Set includeStepResults: true only when debugging wrong/empty output. Set returnFullConfig: true only when you need the full config.
`save_tool`	write	Persists a draft tool to the database. Requires a draftId from build_tool. After saving, the tool can be executed by ID using run_tool with toolId.
`search_documentation`	read	Search documentation for specific information about API structure, endpoints, authentication patterns, etc. Use this when you need to understand how an API works, what endpoints are available, or how to authenticate. Returns relevant documentation excerpts matching your search query.
`tool_a`	read	Tool A
`tool_b`	read	Tool B
`unlimited_tool`	read	No limit

Trust audit

BLOCKgrade F · trust 40/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.

Layer	What it checks	Result
L0	Provenance & inventory	FAIL
L1	Static analysis of the code	FAIL
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	PASS
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: declared (6 observation(s))
Network: declared (18 observation(s))
Shell: declared (9 observation(s))
Dependencies: not all pinned
Secrets in source: found

Findings (25)

CRITICALHard-coded secrets · secret.private_key · CWE-798, CWE-321

.env.example:110

#   VERTEX_PRIVATE_KEY="-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n"

HIGHInventory / provenance · inv.suspicious_name · CWE-1104

packages/web/src/components/tools/hooks/use-payload-validation.ts

use-payload-validation.ts

Why it matters. member named after an attack tool

Fix. remove or justify