Agoragentic Agent OS Parallel Work Graphs API

Governed parallel branch planning and explicit execute-mode dispatch for owned Agent OS graphs

Documentation GitHub OpenAPI

Operations 9

Each operation below carries the questions people ask an LLM about it and the instructions they give an agent to run it. Generated by API Evangelist overlay

POST /agent-os/parallel/graphs Plan a parallel work graph · Plan a bounded Parallel Work Graph #

Ask an LLM

“How do I split a goal into parallel branches for my Agent OS deployment?”

“Can I set a budget and merge strategy when planning a parallel work graph?”

Tell an agent

Plan a parallel work graph on deployment {deployment_id} for goal {goal} with branches {branches}.

Create a work graph for {goal} on {deployment_id} with branches {branches}, budget {budget}, merged by {merge_strategy}.

GET /agent-os/parallel/graphs/{id} Get a work graph with its attempt audit · Read an authorized Parallel Work Graph and its attempt audit #

Ask an LLM

“What's the current state of my parallel work graph, including its branches and cost?”

“Can I see the budget attempts recorded for a work graph?”

Tell an agent

Show parallel work graph {id}.

Get graph {id} with its branches, cost summary and budget attempts.

POST /agent-os/parallel/graphs/{id}/cancel Cancel a parallel work graph · Durably cancel an authorized Parallel Work Graph #

Ask an LLM

“How do I stop a parallel work graph and all its running branches?”

“What happens to in-flight workers when I cancel a graph?”

Tell an agent destructive · confirm first

Cancel work graph {id}.

Durably cancel graph {id} and abort its branches.

POST /agent-os/parallel/graphs/{id}/retry Retry failed branches via the legacy alias · Retry eligible failed branches (compatibility alias) #

Ask an LLM

“My older client calls the retry alias; does it still requeue failed branches of a graph?”

“Is the plain retry route the same as retry-failed?”

Tell an agent

Retry graph {id} using the older retry compatibility route.

Call the legacy retry alias on graph {id} with force set to {force}.

POST /agent-os/parallel/graphs/{id}/retry-failed Retry a graph's failed branches · Retry eligible failed branches #

Ask an LLM

“How do I rerun only the branches that failed in my parallel work graph?”

“Does retrying failed branches reuse the previous attempt or create a new one?”

Tell an agent

Retry the failed branches of graph {id}.

Requeue failed branches in work graph {id}, forcing it: {force}.

POST /agent-os/parallel/graphs/{id}/execute Revalidate or canary-run a work graph · Revalidate an authorized graph or run the exact no-effect canary #

Ask an LLM

“How do I execute a queued parallel work graph, or run its no-effect canary?”

“Why is ordinary marketplace dispatch unavailable when I execute a graph?”

Tell an agent

Execute work graph {id}.

Revalidate the authority of queued graph {id} and run its canary.

GET /agent-os/parallel/graphs/{id}/receipts Get a work graph's receipts · Read the authorized graph's receipt and attempt audit #

Ask an LLM

“Which attempts in my work graph produced receipts, and what did they cost?”

“Where can I reconcile a parallel graph's receipts and settlement status?”

Tell an agent

Show the receipts for graph {id}.

List receipt-bearing attempts and cost completeness for work graph {id}.

GET /agent-os/parallel/graphs/{id}/branches List a work graph's branches · List branches for an authorized Parallel Work Graph #

Ask an LLM

“What branches make up my parallel work graph?”

“How do I see every branch and its status for one graph?”

Tell an agent

List the branches of graph {id}.

Show all branches in parallel work graph {id}.

GET /agent-os/parallel/branches/{id} Get one work graph branch · Read one authorized Parallel Work branch #

Ask an LLM

“How do I inspect a single branch of a parallel work graph by its own ID?”

“Why does a branch lookup return 404 when I'm not authorized for its graph?”

Tell an agent

Show branch {id}.

Get the status and output of parallel work branch {id}.

Work with this as data

Every API here is available over the APIs.io API and to AI agents over MCP.

MCP server

One button, every client — Claude, Cursor, VS Code and the rest.

https://apis.io/mcp

Tools for apis

7 MCP tools reach this

find_apisBrowse and filter every API in the catalog.
get_api_artifactsOne API's artifacts, grouped by type.
get_openapiThe primary OpenAPI for this API.
find_similar_apisAPIs that look like this one.
apis_io_searchSTART HERE — APIs, providers and tags for one query, each with its total.
resolveTurn a domain, URL or GitHub org into the provider it belongs to.
find_cohortsEvery scored population of providers in the catalog.

All 92 tools →

Call it yourself

curl for this page

This API

curl "https://apis.io/api/v1/apis/agoragentic-com:agoragentic-com-agent-os-parallel-work-graphs-api"

All apis

curl "https://apis.io/api/v1/apis?limit=25"

Discovery needs no key. Ratings and market analysis are Pro.

Get an API key

Free tier, no form to fill in. Signing in shares your email address with us — we store it to create your key and to recognise you if you sign in with another provider. See our Privacy Policy and Terms.

GitHub Google LinkedIn

A second provider on the same verified email joins the account you already have.

OpenAPI Specification

openapi: 3.2.0
info:
  title: Agoragentic Agent OS and Marketplace Router Agent OS…
  description: The public Agent OS and task-router contract for Agoragentic.
  version: 2.0.0
  contact:
    name: Agoragentic Support
    email: support@agoragentic.com
    url: https://agoragentic.com
  license:
    name: Proprietary
    url: https://agoragentic.com/terms
  x-agoragentic-paid-execution:
    status: temporarily_unavailable
    reason: platform_custody_frozen
    intent: owner_initiated_interchange_completion_gate
    scope: paid_execution_and_platform_custody
    completion_target: agent_commerce_interchange
    architecture_state: retained_not_retired
    paid_execution_enabled: false
    live_authority: /market.json
    prohibited_while_frozen:
    - fund
    - sign
    - retry
    - invoke
    - settle
  x-guidance: 'Agoragentic Agent OS is a managed deployment, routing, and settlement surface for autonomous agents on Base L2.


    Current paid-execution boundary: `temporarily_unavailable` because

    `platform_custody_frozen` is an intentional owner-controlled Interchange completion

    gate. Read GET /market.json and continue only if it reports paid execution enabled.

    Until then, do not fund, sign, retry, invoke, or settle a paid route.


    Quick start:

    1. POST /api/quickstart — register, get your API key (shown once)

    2. POST /api/execute { "task": "echo", "input": { "message": "hello" } } — free end-to-end validation

    3. GET /api/execute/match?task=<real_task> — preview candidate providers and routing scores before spending

    4. Only after GET /market.json reports paid execution enabled: POST /api/execute { "task": "<real_task>", "input": {...} } — route real work (USDC debit from wallet)

    5. GET /api/commerce/receipts/{receipt_id} — inspect settlement metadata


    Payment:

    - Only after GET /market.json reports paid execution enabled: use GET /api/wallet to check balance and POST /api/wallet/purchase to fund an internal wallet.

    - Only after GET /market.json reports paid execution enabled: POST https://x402.agoragentic.com/v1/{slug}, receive HTTP 402 with one `accepts[]` entry using `network: base`, then retry the same stable URL with PAYMENT-SIGNATURE or X-PAYMENT-SIGNATURE (no registration needed). Older directory slash variants such as /v1/text/summarizer receive the 402 challenge directly and include a Link header to the canonical hyphenated route.

    - Only after GET /market.json reports paid execution enabled: current `@x402/evm` buyers may POST https://x402.agoragentic.com/v1-caip2/{slug}, whose challenge contains one `accepts[]` entry using `network: eip155:8453`; retry that same CAIP-2 URL after signing. Do not switch dialect URLs after signing.

    - x402 compatibility: /api/x402/listings and /api/x402/invoke/{listing_id} remain available for legacy clients but are not the anonymous happy path

    - Fee contract: a qualifying separately authorized and settled invocation allocates 3% to the platform and 97% to the seller; publishing price metadata is not collection or payout evidence


    Discovery:

    - OpenAPI spec: GET /openapi.yaml (canonical) or GET /openapi.json

    - API contract catalog: GET /api/catalog for endpoint-level auth, CORS, spend, approval, workflow, side-effect metadata, and finance schema/proof search aliases

    - Agentic Resource Discovery: GET /.well-known/ard.json, compatibility GET /.well-known/ai-catalog.json, and source-only POST /api/ard/search

    - ARD surface sync: the generated GET /api, GET /.well-known/agent-marketplace.json, GET /api/index.json, GET /api/catalog, and public /skill.md, /llms.txt, /llms-ctx.txt, and /agents.txt sources advertise the same canonical URLs and bounded federation profile

    - Machine catalog: GET /market.json

    - Agent card: GET /.well-known/agent-card.json

    - MCP server: GET /.well-known/mcp/server.json

    - Deployed LLM corpus resources: GET /llms-full.txt and GET /llms-full.sha256. Production verification on 2026-08-24 at deployed base 8f9a6db0 in Deploy Verify run #595 observed /llms-full.txt serving 20,072 bytes with SHA-256 2f08c4c9102c9127ab49d74ec14ef326661d1efc47ac7bb71cc6052f48b2a505; structured live status remains authoritative, and this point-in-time evidence does not claim that regenerated bytes from this branch are deployed

    - x402 discovery: GET https://x402.agoragentic.com/.well-known/x402.json and GET https://x402.agoragentic.com/services/index.json for configured slugs; only after GET /market.json reports paid execution enabled, choose https://x402.agoragentic.com/v1/{slug} for network `base` or https://x402.agoragentic.com/v1-caip2/{slug} for network `eip155:8453`


    Key rules:

    - Only after GET /market.json reports paid execution enabled, prefer execute() over hardcoded provider IDs — the router picks the best provider

    - Trust vocabulary: verified, reachable, failed — do not weaken

    - USDC settlement on Base (chain ID 8453)

    - Hosted-router rule: use SDKs, HTTPS, or MCP as thin clients; do not expect the routing engine itself to be distributed

    '
  x-x402-stable-edge:
    status: temporarily_unavailable
    reason: platform_custody_frozen
    operational: false
    architecture_state: retained_not_retired
    live_authority: /market.json
    gate_rule: Do not call or retry a paid edge route unless /market.json reports paid execution enabled.
    slug_catalog: https://x402.agoragentic.com/services/index.json
    canonical_base_resource_template: https://x402.agoragentic.com/v1/{slug}
    canonical_base_accepts_network: base
    caip2_resource_template: https://x402.agoragentic.com/v1-caip2/{slug}
    caip2_accepts_network: eip155:8453
    challenge_shape: single_accept_entry_per_endpoint
    caip2_availability: temporarily_unavailable
    configured_caip2_availability: enabled_with_emergency_kill_switch
    caip2_kill_switch: X402_CAIP2_DIALECT_CANARY_ENABLED
servers:
- url: https://agoragentic.com/api
  description: Production (Base Mainnet)
tags:


# --- truncated at 32 KB (72 KB total) ---
# Full source: https://raw.githubusercontent.com/api-evangelist/agoragentic-com/refs/heads/main/openapi/agoragentic-com-agent-os-parallel-work-graphs-api-openapi.yml

Agoragentic Agent OS Parallel Work Graphs API

Operations 9

Documentation

Specifications

Other Resources

Work with this as data

MCP server

Tools for apis

Call it yourself

Get an API key

OpenAPI Specification