Skip to content

Have an AI agent? Let it explore Teleon for you. Any assistant works, personal or coding.

Connect your agent
What your agent reads

The request it copies

Explore teleon.dev for me and report back. Read https://teleon.dev/llms.txt, https://teleon.dev/.well-known/agent-card.json and https://teleon.dev/.well-known/mcp.json. Summarize the shared backend primitives at https://teleon.dev/docs and the THP v1.0 connection steps at https://teleon.dev/connect. Stay read-only.

Discovery files on teleon.dev

Read-only discovery is open. Keep TELEON_KEY in your own environment.

Teleon
Sign inApply for an instance
Work

Run each step on the right machine.

teleon.dev/docs/workers starts a short-lived worker for every claimed step: a container, a GPU machine or a paired computer. Each one starts clean, reports telemetry when it finishes and hands its result back to the queue.

teleon.dev/docs/workersWorker bridgehttpsmcp
Steps waiting for the right machine
ContainersCPU, 4 to 16 cores
aq35 grid cell
aq74 evaluator
aq104 page check
GPU machinesCUDA 12, 24 to 80 GB
aq49 scene render
aq116 image match
Paired computersyour own machine
aq50 reopen check
aq92 revise

Each step goes to a pool whose hardware mask matches it. Held steps wait for a free machine.

What it does

Matched to the step

Hardware masks and render budgets are checked before a claim, so a GPU step lands on a GPU.

Clean every time

Workers start from a fresh image for each step and return their output to the queue.

Telemetry on every finish

Each completion carries a scrubbed telemetry record for the drift radar and the ledger.

The contract

A worker receives one self-contained task row with its lease and returns one signed receipt.

Worker kindsContainer, GPU machine, paired computer
Execution profilesBrowser-local, existing harness, paired remote worker, managed cloud
OutputOne signed receipt per step
AccessRING-2 with a scoped worker key
Identifiers
  • portable_work_unit/v1
  • capacity_feed_snapshot/v1
Request and responseCopy
POST /workers/v1/claim X-Teleon-Surface-Id: worker-pool-gpu { "hw_mask": "gpu.cuda12,ram.64g", "queues": ["aq49_creative_scene_synthesize"] } # Response { "task_id": "t_8f21", "lease": {"fencing_token": 4412, "expires_in_s": 900}, "row": {"queue": "aq49_creative_scene_synthesize", "budget_cents": 40} }

Endpoints

MethodPathAccessWhat it doesStatus
POST/workers/v1/claimRING-2Claim one step that matches this worker's hardware
POST/workers/v1/heartbeatRING-2Keep a lease alive during a long step
POST/workers/v1/receiptsRING-2Return the result and telemetry for a finished step
GET/workers/v1/poolsRING-3List worker pools and their hardware

Listed paths appear in the service’s published files today. In the one-host plan every path here is served under api.teleon.dev/v1/workers.

For agents

Give your agent the read-only guide first. It reports back before any write.

Agent prompt
Read https://teleon.dev/docs/workers and list the worker kinds and the hardware each one offers. Read only: ask me before claiming anything.
Copy

Words on this page

Full glossary
Lease
A time-limited claim on a task. The worker holding the lease finishes it or the lease expires and the task returns.
Fencing token
A number that grows with every new lease, so a write from an older lease holder is refused safely.
Idempotent
Safe to repeat: running the same request twice leaves the same result as running it once.
Access ring
One of four access levels. RING-0 is internal operations, RING-1 operators, RING-2 keyed agents and sites, RING-3 public reads.