HumanTask API Examples
See practical examples of AI agents creating human tasks, tracking completion and retrieving evidence.
Examples
Examples is the developer surface for one specific problem: show end-to-end patterns for common agent workflows instead of documenting isolated endpoints. The design has to combine ordinary software concerns with physical latency, evidence and money movement.
Photo verification agent
Photo verification agent should be documented as an explicit part of this interface. The caller needs to know which fields affect routing, which values are authoritative and which states can change after a human has accepted work.
Store stock agent
Store stock agent should be documented as an explicit part of this interface. The caller needs to know which fields affect routing, which values are authoritative and which states can change after a human has accepted work.
detect missing physical fact
→ create task
→ wait for task.completed
→ fetch evidence
→ validate required fields
→ continue the agent planProperty check workflow
Property check workflow is where the implementation should expose physical-world constraints rather than abstracting them away. If access, timing, identity or evidence changes the result, the schema should say so.
detect missing physical fact
→ create task
→ wait for task.completed
→ fetch evidence
→ validate required fields
→ continue the agent planRemote-hands recovery
For Remote-hands recovery, prefer a small number of well-defined fields over free-form conventions. The physical worker may read natural-language instructions, but software should get structured values for anything that affects control flow.
Local research workflow
Local research workflow is where the implementation should expose physical-world constraints rather than abstracting them away. If access, timing, identity or evidence changes the result, the schema should say so.
Design lessons across examples
A production treatment of Design lessons across examples needs both the normal path and the exception path. Human execution is reliable only when the client knows what happens when the environment does not match the request.
Example: Stock Check
A shopping agent identifies a product whose online availability is unreliable. It creates a stock-check task with store, SKU and evidence requirements. When the task completes, it reads the availability field and supporting photo before deciding whether to recommend collection.
Example: Remote Hands
An infrastructure agent detects that a device has stopped responding after software recovery attempts. It creates a tightly scoped reset task with approved physical steps and before-and-after indicator evidence. A failure state triggers escalation to a technician.
Example: Property Evidence
A property workflow needs current exterior conditions for a remote reviewer. It creates a standardized shot list, waits for evidence and stores the media references alongside the property record. The worker does not make an investment recommendation.
Error Model
The HTTP request can succeed while the real-world task cannot. Keep transport errors separate from execution outcomes so applications know whether to retry the API call or change the field task itself.
Production Readiness
Test the unhappy path deliberately: invalid scope, no available supply, delayed completion, missing proof and client retry. Real-world APIs are defined as much by those states as by success.
Common Integration Mistake
Examples should make failure and revision paths visible, not present every field task as a guaranteed success.
Next Step
Implement one examples flow against the canonical task lifecycle, inspect the real payloads and only then generalize the client for more capabilities or locations.
Implementation Notes for Examples
Photo verification agent
When implementing photo verification agent for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Store stock agent
When implementing store stock agent for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Property check workflow
When implementing property check workflow for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Remote-hands recovery
When implementing remote-hands recovery for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Local research workflow
When implementing local research workflow for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Design lessons across examples
When implementing design lessons across examples for Examples, define the contract in a way that another service can validate without reading hidden UI state. Document required fields, optional fields, state-dependent behavior and at least one exception. Because HumanTask API can create real-world work, every ambiguous write operation should also have a traceable client reference or audit path.
Test Matrix
A Examples integration should be tested against more than the happy path. Include a valid request, invalid input, duplicate retry, no matching supply, worker exception, incomplete evidence and cancellation where the interface supports it. The result of each test should be observable in the canonical task object so developers can reconcile UI, API and agent behavior.
Frequently asked questions
What is the purpose of the Examples documentation?
It explains how to show end-to-end patterns for common agent workflows instead of documenting isolated endpoints.
Should task creation be idempotent?
Yes when the interface can create paid work. Network retries must not silently dispatch duplicate humans.
How should physical-world failures be represented?
Use domain-specific states or structured exceptions for conditions such as denied access, unavailable items, unsafe conditions or an invalid target.
Can REST and MCP use different task models?
They should not. Multiple interfaces should resolve to one canonical task, worker and evidence model.
What is the main implementation mistake to avoid?
Examples should make failure and revision paths visible, not present every field task as a guaranteed success.
Related pages
- Human Task API for DevelopersIntegrate AI agents and software with verified human task execution through MCP and REST API.Explore
- QuickstartCreate your first real-world human task and retrieve verified results with HumanTask API.Explore
- MCPConnect AI agents to real-world human workers through the HumanTask API MCP server.Explore
- REST APIUse the HumanTask REST API to create, manage and verify real-world human tasks programmatically.Explore
Broader production endpoints are not public yet.
The public MCP demand-sensor and REST task-request endpoint are live for manual review. The broader REST/OpenAPI surface, automatic matching, payments and evidence-return remain in development.
View Live MCP