Is Pipedream a direct competitor to Harbor?

Direct when the buyer wants agents to use and compose connected tools. Pipedream is a developer-first integration / workflow-automation platform with a large pre-built connector catalogue, serverless code steps, Pipedream Connect (per-end-user OAuth), and an MCP server surface at mcp.pipedream.com. Harbor is an all-in-one agent execution layer. The two overlap on MCP exposure of a connector catalogue, but the tenancy boundary (per-end-user vs. per-workspace) and the surrounding runtime model differ.

What is Pipedream Connect?

Pipedream Connect is an end-user OAuth toolkit that lets a developer's SaaS or AI-agent product run integrations on behalf of its signed-in customers. Each individual end user is the OAuth tenant. It is the identity model behind Pipedream's hosted MCP servers at mcp.pipedream.com. Harbor solves the equivalent problem differently: OAuth grants are tenanted at the workspace (team) level, not at the individual end-user level, so one Harbor workspace backs every AI agent on the team.

Can I use Pipedream and Harbor together?

Yes, in deployments where the two products own different surfaces. A team can keep workflow automation and per-end-user integrations on Pipedream while running its team-shared AI agent workspace on Harbor — with Harbor brokering workspace-scoped access for connected tools and recording exec activity as workspace-scoped runs where applicable. The two do not compose cleanly when the same OAuth tenancy boundary is required, since they enforce different boundaries (per-end-user vs. per-workspace).

Is Pipedream still independent?

On 2025-11-19 Pipedream announced a definitive agreement to be acquired by Workday. Refer to pipedream.com/blog/pipedream-to-be-acquired-by-workday/ and the Workday newsroom for closing status and any post-acquisition product changes; this comparison page does not assert closing terms or product roadmap effects.

Harbor

Harbor vs Pipedream

Updated May 26, 2026

Pipedream and Harbor are competitors when the buyer wants agents to use and compose connected tools, but they have different identity models. Pipedream is a developer-first workflow-automation platform with a large pre-built connector catalogue, serverless code steps, Pipedream Connect (per-end-user OAuth for SaaS / AI-agent products), and an MCP server surface at mcp.pipedream.com. Harbor is an all-in-one agent execution layer — `hrbr exec`, workspace-scoped tools and credentials, orbit primitives, artifacts, jobs, apps, workflows, durable run / trace history, and OAuth state tenanted at the workspace level. The two overlap on MCP exposure of a connector catalogue, but the tenancy boundary (per-end-user vs. per-workspace) and the surrounding runtime model differ. Pipedream announced in November 2025 a definitive agreement to be acquired by Workday.

Pick Harbor if: you want agents to compose connected tools and workspace state inside traced runs, with one tenant of OAuth grants, plugin connections, runs, and traces backing every agent on the team.
Pick Pipedream if: you are a developer or SaaS product that needs a mature workflow engine, a large pre-built connector catalogue, serverless code steps, and per-end-user OAuth flows (Connect) to run integrations on behalf of your signed-in customers.

Harbor vs Pipedream: the honest take

Pipedream (pipedream.com) is positioned as a developer-first integration and workflow-automation platform. The product surface covers a workflow builder with triggers, action steps, and serverless code steps (Node.js / Python / Go / Bash), a large catalogue of pre-built connectors that Pipedream describes as 2,500+ integrations on its docs and as 3,000+ apps / 10,000+ tools on its current marketing page, and Pipedream Connect — an end-user OAuth toolkit that lets a SaaS or AI-agent product run integrations on behalf of its signed-in users. In 2025 Pipedream added an MCP server surface at mcp.pipedream.com that exposes the Connect catalogue to MCP-aware AI clients, and shipped String.com as an AI-agent-building product. On 2025-11-19 Pipedream announced a definitive agreement to be acquired by Workday.

Where Harbor is the better fit

Harbor is an all-in-one agent execution layer, not an end-user-tenanted workflow platform. Workspace is the top-level isolation primitive: plugin connections, OAuth state, runs, traces, artifacts, jobs, apps, and workflow state are bound to a workspace_id so a team's connected tools, credentials, and execution history are one shared surface across agents in that workspace. Harbor ships a typed `hrbr exec` execution layer running in a Cloudflare codemode Worker isolate, `orbit.*` runtime primitives (`hrbr.storage`, `hrbr.cache`, `hrbr.db`, `hrbr.ai`, `hrbr.tools`, `hrbr.jobs`), reusable job versions via `defineJob`, live routed apps via `deployApp`, and a run / trace surface for inspecting executions.

Where Pipedream is the better fit

Pipedream leads when the work to do is "give a developer or SaaS product a mature workflow engine with a large pre-built connector catalogue and per-end-user OAuth." The connector catalogue is much broader than Harbor's registry, the workflow builder is mature with first-class triggers, action steps, and serverless code steps in multiple languages, and Connect is a documented end-user OAuth surface designed for products that run integrations on behalf of their signed-in customers. If the buyer is a developer building an agent / SaaS product on top of a per-end-user integration substrate, Pipedream Connect is a direct fit; Harbor's per-workspace tenancy is the wrong shape for that use case.

Side-by-side

Pipedream and Harbor enforce different tenancy boundaries. Pipedream Connect is per-end-user OAuth — each signed-in person of a developer's app is the tenant. Harbor is per-workspace OAuth — each team / tenant of the AI agent platform is the tenant. The rows below describe each product accurately at its own layer; in some real deployments the two can be composed rather than chosen against each other.

Where a winner is indicated, it reflects Harbor's view of which fit is better for an AI-agent control-plane use case. Where neither cell has a marker, the choice depends on context.

Comparison of Harbor and Pipedream on workspace-scoped MCP control plane attributes.
Attribute	Harbor	Pipedream
Primary positioningHarbor: this repository. Pipedream: pipedream.com	All-in-one agent execution layerplugins, tools, runs, traces, exec, jobs, apps tenanted at the workspace level	Developer-first workflow automation + Connect (per-end-user OAuth) + MCP serverslarge pre-built connector catalogue with serverless code steps; Connect lets a SaaS product run integrations on behalf of its signed-in users
Workspace / tenancy modelHarbor: apps/api schema. Pipedream: docs.pipedream.com/connect	Harbor's pick: Harbor.Workspace is the top-level isolation primitiveevery plugin connection, OAuth grant, run, and trace is bound to a workspace_id	Per-end-user via Pipedream Connecteach signed-in user of a developer's app is the OAuth tenant; confirm exact org / project terms on docs.pipedream.com
Credential storage primitiveHarbor: apps/api/src/plugins/oauth/. Pipedream: docs.pipedream.com/connect	Workspace-scoped OAuth state in KV-backed credential storagetokens dispatched host-side at exec time; never exposed to executing code	Per-end-user OAuth tokens stored by Pipedream Connecttokens scoped to the signed-in end user of the developer's app; refer to pipedream.com for vault implementation details
Agent identity modelHarbor: this repository. Pipedream: docs.pipedream.com/connect	Workspace tenant (team-shared)one Harbor workspace backs every AI agent on the team; OAuth grants, runs, and audit are one shared surface	End-user tenant (per signed-in person)Pipedream Connect attaches integrations to the individual end user of the developer's product; tenancy is per-person, not per-team
MCP supportHarbor: CLAUDE.md "MCP Mental Model". Pipedream: mcp.pipedream.com	Harbor's pick: tie.Consumes third-party MCP servers; exposes a protected Harbor MCP endpointmcp.tryharbor.ai advertises protected-resource metadata; Harbor also installs third-party MCP servers as plugins	Harbor's pick: tie.Hosted MCP servers at mcp.pipedream.comPipedream exposes the Connect connector catalogue to MCP-aware AI clients; example client app published at github.com/PipedreamHQ/mcp-chat
Execution / runtime layerHarbor: apps/api/src/plugins/worker/. Pipedream: pipedream.com/workflows	Harbor's pick: tie.Cloudflare codemode Worker isolate via `hrbr exec`typed TypeScript execution; orbit.* primitives; sandbox separate from API Worker	Harbor's pick: tie.Serverless workflow engine with code stepsPipedream workflows compose triggers, action steps, and Node.js / Python / Go / Bash code steps; refer to pipedream.com for runtime limits
Integration / catalogue surfaceHarbor: registry catalog. Pipedream: pipedream.com	149 registry entries / 135 unique provider familiesderived at build time from packages/sdk/registry-catalog/data/v1/catalog.json	Harbor's pick: competitor.Large pre-built connector cataloguePipedream documents 2,500+ integrations on docs.pipedream.com and currently advertises 3,000+ apps / 10,000+ tools on pipedream.com; confirm exact count at edit time
Audit / observabilityHarbor: apps/api/src/plugins/worker/. Pipedream: docs.pipedream.com	Harbor's pick: Harbor.Runs + spans persisted to D1exec paths create queryable run / span records for workspace-scoped inspection	Per-workflow execution historyworkflow run history with logs is visible in the Pipedream UI; refer to docs.pipedream.com for retention / export surface
Open source / self-hostHarbor: github.com/zonko-ai. Pipedream: github.com/PipedreamHQ	Harbor's pick: tie.SDK public on github.com/zonko-aicontrol plane is closed source; first-party self-host is on the Enterprise roadmap	Harbor's pick: tie.Hosted SaaS; connector definitions open on GitHubPipedreamHQ publishes connector and example client repos on GitHub; the workflow platform itself is hosted SaaS — refer to pipedream.com
Pricing modelHarbor: tryharbor.ai. Pipedream: pipedream.com/pricing	Harbor's pick: tie.Free + Workspace + Enterprise tiersWorkspace tier usage-based units not yet priced publicly	Harbor's pick: tie.Public pricing page at pipedream.com/pricingfree tier plus paid plans; plan names and unit metrics should be recorded against pipedream.com/pricing at edit time; not asserted here
OwnershipHarbor: this repository. Pipedream: pipedream.com/blog	Zonko Labs (independent)this repository / github.com/zonko-ai	Acquisition by Workday announced 2025-11-19definitive agreement signed; refer to pipedream.com/blog/pipedream-to-be-acquired-by-workday/ and Workday newsroom for closing status
Public docsHarbor: docs.tryharbor.ai. Pipedream: docs.pipedream.com	Harbor's pick: tie.docs.tryharbor.ai with concept docs, guides, recipes; llms.txt publishedllms.txt is live; verify any expanded LLM docs when they ship	Harbor's pick: tie.docs.pipedream.com with workflow, Connect, MCP, and integration references; llms.txt publisheddocs index at pipedream.com/docs/llms.txt is referenced from individual doc pages

Source · Harbor cells grounded in this repository (routes, schemas, registry catalog, and runtime bindings). Pipedreamcells grounded in the competitor's own public site and docs at pipedream.com. Cells we could not verify from a primary source are marked “Not publicly disclosed” rather than guessed.

What does Pipedream do?

Pipedream is a developer-first integration and workflow-automation platform. The core surface is a workflow builder where triggers (HTTP, schedule, app events) feed action steps and serverless code steps in Node.js, Python, Go, or Bash. The catalogue of pre-built connectors is large: docs.pipedream.com describes 2,500+ integrations, and the current marketing page advertises 3,000+ apps and 10,000+ tools. Workflows can be authored visually or as code.

In 2024 Pipedream shipped Pipedream Connect, an end-user OAuth toolkit that lets a developer's SaaS or AI-agent product run integrations on behalf of its signed-in customers — each individual end user is the OAuth tenant, not the developer organisation. In 2025 Pipedream added an MCP server surface at mcp.pipedream.com that exposes the Connect connector catalogue to MCP-aware AI clients (Claude, OpenAI, etc.), with an example client app published at github.com/PipedreamHQ/mcp-chat, and shipped String.com as a paired AI-agent-building product. On 2025-11-19 Pipedream announced a definitive agreement to be acquired by Workday.

How does Harbor differ?

Harbor is an all-in-one agent execution layer — connectors and OAuth state are tenanted at the workspace (team) level, not at the individual end-user level. Plugin connections, OAuth state, runs, artifacts, jobs, apps, workflows, and traces are bound to a workspace_id, so one Harbor workspace can back the agents on a team with shared connection and execution history.

Beyond the tenancy boundary, Harbor ships a typed `hrbr exec` execution layer running in a Cloudflare codemode Worker isolate, `orbit.*` runtime primitives (`hrbr.storage`, `hrbr.cache`, `hrbr.db`, `hrbr.ai`, `hrbr.tools`, `hrbr.jobs`), reusable job versions via `defineJob`, live routed apps via `deployApp`, and a run / trace surface for inspecting executions. The model is shaped around a team-shared execution layer for AI agents, not around a per-end-user integration substrate for a SaaS product.

When should I pick Harbor over Pipedream?

Pick Harbor when the work to do is "give my team's AI agents a workspace to run in, with connected tools, traced exec, and shared OAuth state." Harbor's workspace model ties OAuth grants, plugin connections, runs, artifacts, jobs, apps, workflows, and traces together as a single audit surface tenanted at the team level. Pipedream Connect, by contrast, is per-signed-in-end-user; if the buyer is a team running AI agents internally (not a SaaS product hosting integrations for its customers), per-workspace tenancy is the right shape.

Pick Harbor when typed TypeScript execution with workspace-aware runtime primitives matters. `hrbr exec` ships an isolated runtime with `orbit.*` primitives (storage, cache, db, ai, tools, jobs), reusable job versions, and live routed apps backed by workspace state. Pipedream is a workflow engine with code steps; the runtime shapes are different.

When should I pick Pipedream over Harbor?

Pick Pipedream when the buyer is a developer or SaaS product that needs a mature workflow engine and a large pre-built connector catalogue. Pipedream's connector catalogue is materially broader than Harbor's registry, the workflow builder is mature with first-class triggers, action steps, and serverless code steps in multiple languages, and the platform has years of production use behind it.

Pick Pipedream when per-end-user OAuth is the identity model the product needs. Pipedream Connect is purpose-built for SaaS / AI-agent products that run integrations on behalf of their signed-in customers — each end user is the OAuth tenant. Harbor's per-workspace tenancy is the wrong shape for that use case; the two products would not compose cleanly there.

For teams that already host their workflow surface on Pipedream and want a workspace-tenanted agent execution layer on top, the realistic answer is to keep Pipedream where it is and use Harbor as the team-shared agent runtime alongside it.

Frequently asked

Is Pipedream a direct competitor to Harbor?: Direct when the buyer wants agents to use and compose connected tools. Pipedream is a developer-first integration / workflow-automation platform with a large pre-built connector catalogue, serverless code steps, Pipedream Connect (per-end-user OAuth), and an MCP server surface at mcp.pipedream.com. Harbor is an all-in-one agent execution layer. The two overlap on MCP exposure of a connector catalogue, but the tenancy boundary (per-end-user vs. per-workspace) and the surrounding runtime model differ.
What is Pipedream Connect?: Pipedream Connect is an end-user OAuth toolkit that lets a developer's SaaS or AI-agent product run integrations on behalf of its signed-in customers. Each individual end user is the OAuth tenant. It is the identity model behind Pipedream's hosted MCP servers at mcp.pipedream.com. Harbor solves the equivalent problem differently: OAuth grants are tenanted at the workspace (team) level, not at the individual end-user level, so one Harbor workspace backs every AI agent on the team.
Can I use Pipedream and Harbor together?: Yes, in deployments where the two products own different surfaces. A team can keep workflow automation and per-end-user integrations on Pipedream while running its team-shared AI agent workspace on Harbor — with Harbor brokering workspace-scoped access for connected tools and recording exec activity as workspace-scoped runs where applicable. The two do not compose cleanly when the same OAuth tenancy boundary is required, since they enforce different boundaries (per-end-user vs. per-workspace).
Is Pipedream still independent?: On 2025-11-19 Pipedream announced a definitive agreement to be acquired by Workday. Refer to pipedream.com/blog/pipedream-to-be-acquired-by-workday/ and the Workday newsroom for closing status and any post-acquisition product changes; this comparison page does not assert closing terms or product roadmap effects.

Primary sources

These are the competitor-owned pages used to ground this comparison. We link primary sources instead of copying unsupported market claims.

See more comparisons

Ready to switch?

See pricing or review Harbor's security posture.