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
Operate and govern

Every task on its cheapest reliable mode.

teleon.dev/docs/modes defines 12 execution modes, MODE_1A to MODE_8, built from three foundations: code only, parameterized code first with a model fallback, and full agentic work. The Gravity Ratchet moves each workload down to the cheapest mode that stays reliable for it.

v38.0 to v40.0teleon.dev/docs/modesMode ratchethttpsjson
The Gravity RatchetApproximate cost per task
1MODE_3A: full agenticabout $0.045 a taskStarts here
2MODE_2A or 2B: code first, model fallbackabout $0.002 a taskRatchet step
3MODE_1A: code onlyabout $0.000001 a taskGoal

Work moves down a rung once the cheaper mode proves reliable. Deterministic Role 1 slots already carry about 75% of volume, at 0.15 to 1.9 ms.

What it does

Three foundations, 12 modes

Code only, code first with a model fallback, and full agentic work combine into MODE_1A through MODE_8.

The Gravity Ratchet

Work starts where it succeeds and moves down a mode each time a cheaper one proves reliable.

Context and reach, scoped

Five-tier context escalation adds detail as a task needs it, and a capability firewall scopes what each run can touch.

The contract

Each task stores its execution mode on the task row, MODE_2A by default; the ratchet lowers it as cheaper modes prove reliable.

ModesMODE_1A to MODE_8, store S159
RatchetMODE_3A to MODE_2A or 2B to MODE_1A, S160
ContextFive-tier escalation, S161; six-slot composer, S163
IsolationWASI 0.3 coroutines and a capability firewall, S168
Identifiers
  • teleon_tasks_cold_payload.execution_mode
  • MODE_2A
Request and responseCopy
GET /v1/modes/ratchet?queue=aq471 Host: api.teleon.dev # Response { "queue": "aq471", "current_mode": "MODE_2A", "next_candidate": "MODE_1A", "evidence": {"runs": 4200, "pass_rate": 0.996, "mutation_score": 0.84}, "cost_per_task_usd": {"MODE_2A": 0.002, "MODE_1A": 0.000001} }

Endpoints

MethodPathAccessWhat it doesStatus
GET/v1/modesRING-3List the 12 modes and what each runs
GET/v1/modes/ratchetRING-3Read a queue's mode and its ratchet evidence
POST/v1/modes/ratchet/{queue}/stepRING-1Move a queue down one mode

Listed paths appear in the service’s published files today. This surface comes from the Teleon Omnibus Codex v1.0 to v50.0, Oct 25, 2026. In the one-host plan every path here is served under api.teleon.dev/v1/modes.

Words on this page

Full glossary
Execution mode
One of 12 ways to run a task, MODE_1A to MODE_8, from code only to full agentic model work.
Gravity Ratchet
The rule that moves each workload to a cheaper execution mode once that mode proves reliable for it.
Capability firewall
Object-capability security for runs: code reaches only the handles it is given, such as one folder or one API.