Snapshot
Continue to About ↓- What it does
- AgentKit is a hosted MCP gateway and credentials layer that connects AI agents to 600+ apps and 30,000+ tools behind one MCP endpoint.
More detail
- It stores and refreshes each end user's OAuth tokens at the gateway, enforces per-agent tool scoping, and logs every tool call with agent, user and result attribution.
- Built for teams hosting their own agents who need their agents to act on behalf of their own users.
- Best for
- Teams that build and host their own agents (LangChain, Google ADK, OpenAI/Anthropic SDKs, Vercel AI SDK, Mastra) in multi-tenant products, where each customer's agent must act inside that customer's own Gmail, Slack, Salesforce or project tool accounts with per-user tokens, scoped tool access and call-level audit logs.
- Pricing
- Free: $0/month with 5,000 tool calls per month on a hard stop, 10,000 trigger events, unlimited connected accounts and bring-your-own credentials.
More detail
- Growth: $99/month including 100,000 tool calls, then $0.50 per additional 1,000 calls
- 50,000 trigger events, then $3 per additional 1,000 events
- custom
- Main limitation
- Prototypes or early agents that will exceed 5,000 tool calls a month, since the Free tier hard-stops at that volume.
More detail
- Also poor fit if you need air-gapped or in-VPC deployment, which is Enterprise-only, or if you want to own the OAuth layer yourself rather than have it managed.
- Record
- Living and monitored
- Reviewed
- Evidence
- Public web research
- Machine access
- Structured profile ↗
About
- AgentKit is a hosted MCP gateway and credentials layer that connects AI agents to 600+ apps and 30,000+ tools behind one MCP endpoint.
- It stores and refreshes each end user's OAuth tokens at the gateway, enforces per-agent tool scoping, and logs every tool call with agent, user and result attribution.
- Built for teams hosting their own agents who need their agents to act on behalf of their own users.
What it helps with
- Each end user's OAuth tokens have to be collected, encrypted, refreshed and attached to the right call without exposing
- Tool catalogs are large enough to overload model context, so agents need a reduced tool set matched to their job rather
- Agents need to act as a specific user, with one user's run unable to reach another user's data.
- Tool call failures are hard to diagnose when it is unclear whether the connector, the upstream API or the local configur
- Event-driven agents otherwise require polling, cron jobs or webhook infrastructure to maintain.
- Enterprise security reviews demand audit trails, SIEM streaming, BYOK and data residency that are costly to build in-hou
Capabilities
A practical view of what is available and where limitations apply.
| Offering | Available | Detail |
|---|---|---|
| Connector and tool catalog | Yes | Vendor claims 600+ connectors and 30,000+ tools behind one MCP endpoint. Internal release notes cite lower figures at earlier dates, so treat the headline as a current marketing claim. |
| Managed per-user auth | Yes | OAuth 2.0, API keys, bearer and basic auth; tokens stored encrypted per tenant, refreshed, and attached at the gateway so code and prompts never hold them. Consent screen can be hosted on your own domain. |
| Virtual MCP Servers | Yes | Per-agent tool subsets from any connected app served as one endpoint. The vendor claims around 80% less tool overhead, which is unverified. |
| Triggers | Yes | Start an agent run on events in connected apps without polling, cron jobs or webhook endpoints. Metered separately from tool calls. |
| Per-call logs and SIEM streaming | Yes | Every call attributed to agent, user, tool and result, with failures tagged as connector error or API error. Streaming to Datadog, Splunk or any SIEM is an Enterprise feature. |
| Custom tools and API proxy | Yes | Add internal services, partner APIs or your own MCP server; custom connectors and API proxy are listed as paid Growth features. |
| Search Tools API | Yes | Natural-language tool discovery across the catalog, shipped September 2026. |
| Org-wide shared credentials | Yes | An admin can connect an app once for the whole organization; shipped September 2026, so earlier versions required per-user connections. |
| EU data residency | Limited | Available, but only as a paid add-on at $99/month on Growth; included in Enterprise. |
| Self-hosted or air-gapped deployment | Limited | Enterprise tier only; not available on Free or Growth. |
Bot trust
How confidently AI systems can verify this profile from public evidence.
62of 100
ChatGPT62/100
Claude—/100
What bots could verify
- • Identifies corporate entities, founders, and funding with third-party corroboration from Z47, Inc42, and Dealroom.
- • Breaks down public pricing tiers and transparently flags an internal discrepancy between 5,000 and 10,000 tool call limits.
- • Critically distinguishes vendor-claimed compliance (SOC 2, ISO 27001) from independently verified audit reports.
What weakens trust
- • Provides no independent customer evidence, case studies, or user reviews verifying successful production scale.
- • Security certifications (SOC 2 Type II, ISO 27001) and annual penetration test results remain uninspected by third parties.
- • No independent availability data, historical incident logs, or verified uptime track record are provided.
Evidence
Where this record came from, when it was reviewed, and how confidently it can be used.
- Verification
- Scribble researched these facts from the linked public sources.
- The owner has not manually confirmed every field.
- Last reviewed
- Primary source
- scalekit.com
Technical profile details
Record ID
2850ca4a-c2b4-49fb-8695-ff1197096cb6Canonical URLhttps://profiles.scribble.network/listing/scalekit-2850ca