Best of · Agent runtime
Best human approval and handoff for AI agents
All 10 ranked human approval and handoff on the Anchor benchmark, with a pick for each need and where each one falls short. Scores come from public evidence, re-checked as vendors change.
- 10 ranked
- 3 agent-ready
- 8 hosted endpoints
- Updated 9 October 2026
Top three
Picks by need
Worked out from the scores, prices and facts, so they change when the research does.
Maintenance & community
Restate BB
92/100 on maintenance & community, against 90 for the overall leader.
Transparency & trust
Trigger.dev BB
73/100 on transparency & trust, against 70 for the overall leader.
The shortlist
| # | Tool | Grade | Best for | Price | Where |
|---|---|---|---|---|---|
| 1 | Temporal Temporal Technologies |
BB 77.2 | A team that already runs, or wants to run, agents as durable workflows with strict guarantees and an audit trail per run. | $0.05 / 1k calls | local |
| 2 | Trigger.dev Trigger.dev |
BB 74.7 | A TypeScript team that wants an approval pause inside background jobs or AI chat agents, with the reviewer UI built in their own app. | Freemium | hosted and local |
| 3 | Restate Restate GmbH |
BB 73.3 | A team that already writes agent steps as durable handlers and wants a long, cheap wait for a human answer in TypeScript, Python, Java, Go or Rust, with the option to self-host. | $75 / mo | local |
| 4 | Hatchet Hatchet Technologies, Inc. |
B 68.3 | It suits a team already running background jobs or agent loops on Hatchet that wants an approval pause inside the same durable task, in Python, TypeScript, Go or Ruby. | $0.03 / 1k calls | hosted and local |
| 5 | Inngest Inngest |
B 66 | A TypeScript, Python or Go team that already runs agent steps as durable functions and wants a cheap, long wait for a human answer. | $99 / mo | hosted |
| 6 | Permit MCP Gateway Permit.io |
C 54.1 | A security team that wants approvals and per-user limits on MCP tools across many clients without touching agent code. | Paid | hosted |
| 7 | Orkes Conductor Human tasks Orkes |
C 54 | Teams that already orchestrate agents as Conductor workflows and need formal routing, per-assignee deadlines and escalation without writing them. | Paid | hosted and local |
| 8 | withHuman Metoro Inc |
D 51.8 | Teams that want tool calls from coding agents or their own agents held for approval by rules, with escalation and an audit log, and no review UI to build. | $20 / seat-mo | hosted |
| 9 | Pushary Pushary |
D 51.4 | One developer running coding agents unattended who wants questions and approvals on a phone. | $9.99 / mo | hosted |
| 10 | gotoHuman gotoHuman |
E 43.6 | A team that wants a ready reviewer inbox where people edit AI output before it ships, without building a UI. | $99 / mo | hosted and local |
How to choose
- Channels the approver can reachCheck which channels, such as chat, email or a web inbox, the approver can answer from, since an approval the person cannot reach in time stalls the whole run.
- Timeout behaviour on no answerCheck what happens when nobody answers, whether the request times out, escalates or is denied by default, because a silent default can approve an action nobody saw.
- How the answer reaches the agentCheck how a decision returns to the agent and whether it resumes with the same state, since an answer arriving after a restart can be lost or applied twice.
- What each decision logsCheck what each decision logs, including who was asked, what they saw and any timeout, and whether the log can be exported.
How the benchmark tests this category. An agent asks for approval of a risky action over two channels and gets one approval, one rejection and one timeout. We check the routing, what the approver sees, how the answer reaches the agent and what is logged.
Each one in detail
Temporal
BB 77.2/100Open-source durable execution platform with SDKs in Go, Java, Python, TypeScript, .NET, PHP, Ruby and Rust, run yourself or on Temporal Cloud.
Verdict Waits of any length with a timeout survive worker restarts and deploys. No reviewer inbox, notifications or routing, so the human side is all your code.
Choose it for A team that already runs, or wants to run, agents as durable workflows with strict guarantees and an audit trail per run.
Strengths
- Waits of any length with a timeout survive worker restarts and deploys
- Each workflow's event history is an audit trail of every signal, without extra logging
- Contractual SLA of 99.9 per cent per namespace, 99.99 with High Availability
Weaknesses
- No reviewer inbox, notifications or routing, so the human side is all your code
- Needs running workers and a Temporal service before the first approval
- Signals and timers are billed Actions on Cloud, and the $150 trial credit needs a card
Price $0.05 / 1k callsAuth OAuth or keyx402 nolocal
Trigger.dev
BB 74.7/100Open-source background jobs and durable tasks in TypeScript.
Verdict Tokens complete from a backend, a pre-signed callback URL or the browser with a token scoped to one waitpoint. 10-minute default timeout on tokens.
Choose it for A TypeScript team that wants an approval pause inside background jobs or AI chat agents, with the reviewer UI built in their own app.
Strengths
- Tokens complete from a backend, a pre-signed callback URL or the browser with a token scoped to one waitpoint
- No compute billed for waits over 5 seconds
- Idempotency keys and tags on tokens, and a list endpoint filtered by status
Weaknesses
- 10-minute default timeout on tokens
- No built-in reviewer UI, notifications, routing or record of who completed a token
- Tasks are written in TypeScript, though any language can complete a token over HTTP
Price FreemiumAuth OAuth or keyx402 nohosted and local
Restate
BB 73.3/100Durable execution runtime from Restate GmbH in Berlin, run as Restate Cloud, in the customer's own cloud account or self-hosted. Handlers pause on durable promises (awakeables and signals) and resume when a person answers over HTTP.
Verdict An awakeable pauses a handler at no compute cost and resumes it from one HTTP request, with a durable timeout, on a free plan of 100,000 actions a month. Restate sends nothing to the approver, so the Slack message, email or inbox is the builder's code, and the published terms and privacy statement predate the paid plans.
Choose it for A team that already writes agent steps as durable handlers and wants a long, cheap wait for a human answer in TypeScript, Python, Java, Go or Rust, with the option to self-host.
Strengths
- One HTTP request resolves or rejects an awakeable, and the waiting handler resumes after restarts or redeployment
orTimeoutputs a durable timer on the wait, so the remaining time survives a crash- Free plan of 100,000 actions a month with no card, and $25 per extra million actions on Starter and Business
Weaknesses
- No reviewer inbox, Slack app or email. The approval request and the reply endpoint are the builder's code
- Signals can't be resolved over HTTP yet, only from another handler (open issue of 5 October 2026)
- Self-hosted ingress and admin ports have no authentication, so the awakeable ID alone resolves a wait there
Price $75 / moAuth API keyx402 nolocal
Hatchet
B 68.3/100Hatchet is an open-source (MIT) platform for durable tasks, queues and workflows, sold as Hatchet Cloud or self-hosted. A durable task can pause until an event arrives, which is its building block for human approval steps.
Verdict A durable task waits for a keyed event without holding a worker slot, and resumes from its event log after a restart, in Python, TypeScript, Go or Ruby. Hatchet supplies only the pause and resume. It has no approver inbox, notification channel or record of who answered, and its MCP server cannot push the event.
Choose it for It suits a team already running background jobs or agent loops on Hatchet that wants an approval pause inside the same durable task, in Python, TypeScript, Go or Ruby.
Strengths
- A durable task waiting on an event is evicted from its worker slot and resumes from the durable event log after a restart or crash
- Events can be filtered with a CEL expression, and a scope with a lookback window catches an answer pushed before the wait began
- MIT licence, self-hostable, with SDKs for Python (1.41.1), TypeScript (1.36.0), Go and Ruby (0.9.0)
Weaknesses
- No approver inbox, Slack or email channel, routing or reminder. The owner builds the request and the way the answer is pushed
- An event wait has no timeout of its own. A deadline needs a sleep condition in the same or group
- API tokens are scoped to a tenant with an expiry, with no narrower scopes, and the role and payload restrictions do not apply to them
Price $0.03 / 1k callsAuth API keyx402 nohosted and local
Inngest
B 66/100Durable execution platform for TypeScript, Python and Go, with event-driven workflows and support for human approval.
Verdict Waits are durable and cost nothing while suspended, with runs up to 30 days on Free and 366 on Business. No reviewer inbox, Slack app or email, so the human side is your code.
Choose it for A TypeScript, Python or Go team that already runs agent steps as durable functions and wants a cheap, long wait for a human answer.
Strengths
- Waits are durable and cost nothing while suspended, with runs up to 30 days on Free and 366 on Business
- Two resume paths, events matched by ID for many runs and transactional signals for one
- Free plan of 50,000 executions a month with no card, and per-execution prices published
Weaknesses
- No reviewer inbox, Slack app or email, so the human side is your code
- An answer sent before
waitForEventstarts listening isn't matched - 16 incidents on the status page between 7 July and 26 September, with function execution at 99.14 per cent
Price $99 / moAuth OAuth or keyx402 nohosted
Permit MCP Gateway
C 54.1/100Hosted proxy between MCP clients and MCP servers that signs in the human behind the agent, checks each tool call against Permit.io policy and logs it.
Verdict No SDK or client change, since the client points at the gateway URL and keeps its tool list. Approvals are Enterprise only, through a demo, with no published price.
Choose it for A security team that wants approvals and per-user limits on MCP tools across many clients without touching agent code.
Strengths
- No SDK or client change, since the client points at the gateway URL and keeps its tool list
- Fails closed, with timeouts that reject and disconnects that cancel
- OAuth 2.1 with consent, a trust ceiling per user and admin revocation
Weaknesses
- Approvals are Enterprise only, through a demo, with no published price
- Only gateway admins approve, so routing to the right person needs admin seats
- 5-minute default window, extendable 5 minutes at a time, suits live sessions more than overnight review
Price PaidAuth OAuthx402 nohosted
Orkes Conductor Human tasks
C 54/100Workflow orchestration platform, built on the open-source Conductor, with a Human task that pauses a workflow, assigns a form to a user or group and resumes with the submitted answer.
Verdict Escalation chains with a time limit per assignee and a choice of leaving the task open or failing the workflow. No built-in Slack or email prompt, so alerts need a trigger policy and a second workflow.
Choose it for Teams that already orchestrate agents as Conductor workflows and need formal routing, per-assignee deadlines and escalation without writing them.
Strengths
- Escalation chains with a time limit per assignee and a choice of leaving the task open or failing the workflow
- Reviewers can be Conductor users or people in your own identity system, by email or group
- Human task search by state, assignee, claimant, full text and input or output fields
Weaknesses
- No built-in Slack or email prompt, so alerts need a trigger policy and a second workflow
- Several objects to set up (form, task, workflow, application key) before the first approval
- Paid editions are contact sales only, and the free Developer Edition isn't for production
Price PaidAuth API keyx402 nohosted and local
withHuman
D 51.8/100withHuman is a hosted approval service for AI agents from Metoro Inc. It holds an agent's tool calls for a person to approve or deny through the Agent Approval Protocol API, coding-agent hooks, an MCP gateway and an MCP server.
Verdict The approval API is small and carefully specified, with a public OpenAPI 3.1 document, mandatory idempotency keys and agent credentials that can never approve. The service is in early access under its own terms, with no status page, published rate limits or changelog, and the open-source repository the site links to returned 404 on 8 October 2026.
Choose it for Teams that want tool calls from coding agents or their own agents held for approval by rules, with escalation and an audit log, and no review UI to build.
Strengths
- Public OpenAPI 3.1 document with 116 operations and 148 schemas, plus llms.txt and a Markdown copy of every docs page
Idempotency-Keyis required on request creation, instance creation and decisions, and a reused key with changed input answers 409- An agent credential can only create requests and wait for decisions. Approving or denying needs a signed-in person or a personal key
Weaknesses
- The terms say the service is in early access, and the API document is version 0.1.0
- No status page, published request rate limits, SLA outside an enterprise contract, or changelog found
- The GitHub repository the site links as its open-source edition returned 404 on 8 October 2026, and the organisation lists no public repositories
Price $20 / seat-moAuth OAuth or keyx402 nohosted
Pushary
D 51.4/100Hosted MCP server that lets a coding agent notify you and ask you a yes or no, multiple-choice or free-text question on your phone, Mac, Slack or browser, then wait for the answer.
Verdict Six small tools with a detailed skill that says when to ask, when to notify and when to stay quiet. No free plan, and the 3-day trial takes a card up front.
Choose it for One developer running coding agents unattended who wants questions and approvals on a phone.
Strengths
- Six small tools with a detailed skill that says when to ask, when to notify and when to stay quiet
- Every result says whether it was answered and what to do next (
answered,status,handoffAction) - Lock-screen approve and deny, plus a Mac app, Slack and browser
Weaknesses
- No free plan, and the 3-day trial takes a card up front
- Plain MCP clients get cooperative questions only, with no enforcement
- Wait times are set by the user's delivery mode, so an agent can't count on a long block
Price $9.99 / moAuth API keyx402 nohosted
gotoHuman
E 43.6/100Hosted review inbox for AI agents.
Verdict Built for agent review, with TypeScript and Python SDKs, a 3-tool MCP server, an n8n node and a Make app. No status page, published rate limits or documented error responses.
Choose it for A team that wants a ready reviewer inbox where people edit AI output before it ships, without building a UI.
Strengths
- Built for agent review, with TypeScript and Python SDKs, a 3-tool MCP server, an n8n node and a Make app
- Reviewers can edit the output before approving, and the webhook returns the edited data
- Web inbox, email and Slack on every plan, including the $0 one
Weaknesses
- No status page, published rate limits or documented error responses
- No review expiry or timeout that we could find in the docs
- Review routing and the Reviews API start at $99 a month, audit logs at $950
Price $99 / moAuth API keyx402 nohosted and local
Head to head
- Temporal vs Trigger.dev BB 77.2 vs BB 74.7
- Restate vs Temporal BB 73.3 vs BB 77.2
- Hatchet vs Temporal B 68.3 vs BB 77.2
- Inngest vs Temporal B 66 vs BB 77.2
- Restate vs Trigger.dev BB 73.3 vs BB 74.7
- Hatchet vs Trigger.dev B 68.3 vs BB 74.7
- Inngest vs Trigger.dev B 66 vs BB 74.7
- Hatchet vs Restate B 68.3 vs BB 73.3
- Inngest vs Restate B 66 vs BB 73.3
- Hatchet vs Inngest B 68.3 vs B 66
Questions
What are the highest-rated human approval and handoff for AI agents?
Temporal has the highest benchmark score of the 10 ranked human approval and handoff, 77.2 (BB). Trigger.dev is second with 74.7 (BB).
How many human approval and handoff are agent-ready?
3 of the 10 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark.
Which human approval and handoff accept x402 payments?
None of the ranked listings here accepts x402 for its main call yet.
Which of these human approval and handoff is cheapest?
By published paid prices, Hatchet, at $0.005 per 1,000 calls, the lowest of the 4 listings here with a paid price in this unit (free allowances aside). Plans, volume tiers and free allowances change the sum, so check the listing's price table.
How is this list ranked?
By the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 9 October 2026.
How this list is made
The order is the Anchor benchmark score, the same number as on each listing and in the top list. Each listing is graded from public evidence against the benchmark checklist, and the picks above are worked out from those grades, prices and facts. No listing pays for its place, and paid audits or listing help never change a score.
Full ranked table · 45 head-to-head comparisons · Best tools in every category