NewEvery agent tool call now routes through the governed MCP gateway — policy check, audit log, then dispatch.Read the changelogarrow_forward
XOOZLYAI Orchestrator
routeHow it workssmart_toyAgent fleetshieldGovernance & trustsellPricingmenu_bookDocsarticleBloghistoryChangelog
Log inStart free
grid_viewProductsrouteHow it workssmart_toyAgent fleetshieldGovernance & trustsellPricingmenu_bookDocsarticleBloghistoryChangelogHelp centreLog in
Start free
  1. XOOZLY
  2. chevron_rightDocumentation
Loading the knowledge base…
XOOZLYAI Orchestrator

Ready to orchestrate your AI fleet with total confidence?

Start a project, set a cost cap, and watch the loop run with every tool call policy-checked and logged before it leaves the platform.

Start freearrow_forwardSee what shipped
  • check_circleNo credit card required
  • check_circleYour keys, your model spend
  • check_circleEvery tool call audited

Platform

  • All productsarrow_forward
  • Orchestration looparrow_forward
  • Agent fleetarrow_forward
  • Governed control planearrow_forward
  • Pricingarrow_forward

Trust & legal

  • Trust centrearrow_forward
  • Security & compliancearrow_forward
  • AI use policyarrow_forward
  • Report a vulnerabilityarrow_forward
  • Data subject requestsarrow_forward

Resources

  • Documentation & guidesarrow_forward
  • Engineering & research blogarrow_forward
  • Changelogarrow_forward
  • Uptime & reliabilityarrow_forward

Browse by domain

  • Strategy & Researcharrow_forward
  • Developmentarrow_forward
  • QAarrow_forward
  • Governance & Orchestrationarrow_forward
  • All 10 domainsarrow_forward
XOOZLY
© 2026 XOOZLY · 61 agents · 10 domains · 8 SDLC stationsAI use policyarrow_forwardLog inarrow_forward99.9% uptime SLAarrow_forward
  1. XOOZLY
  2. chevron_rightDocumentation
  3. chevron_rightGuide
Loading this guide…
  1. XOOZLY
  2. chevron_rightDocumentation
  3. chevron_rightUnderstanding run states and lifecycle transitions
arrow_backBack to the knowledge base
schedule2 min read·verifiedChecked against the code on 25 October 2026
Core concepts & getting startedrun-states-and-lifecyclecodeapi-server/orchestrator/run_lifecycle.pycodeapi-server/orchestrator/graph.py

Orchestration

Understanding run states and lifecycle transitions

A run is one pass through stations A to H, and its status is a small closed set. Knowing which state means "still working", which means "waiting for a person" and which means "finished" is most of what you need to read the run log.

open_in_newOpen Run log in the consolearrow_forward
01

The states a run can be in

A run's status is a small closed set, and each value answers one question: is this still working, is it waiting for a person, or is it over?

  • queued — accepted and waiting for a worker. Nothing has been spent yet.
  • planning — a station is producing its plan. This is where a run stops today in any environment with no AI provider key configured.
  • executing — a station is acting on that plan, including any tool calls it is granted.
  • completed — station H finished with no approval outstanding.
  • failed — a station could not proceed, and the run carries the reason.
  • paused — stopped deliberately, most often by a cost-cap breach or an approval gate, and resumable.
02

One run is one pass through A–H

A run is one pass through stations A to H. It completes when H finishes — the loop back to A is the next run, not this one starting over. Reading it any other way makes a healthy fleet look like it is repeating work.

hubStations, hub and support are three different thingsThe eight stations are the work. Category I, Governance & Orchestration, is the hub that routes between them and gates on approvals; it is not a ninth station and holds no run state of its own. Category J is cross-cutting support — it can read a run's state but never advances the loop.
03

Why a run pauses

A pause is deliberate and resumable — it is not a failure. Two things cause one, and both hand the next decision to a person.

  1. savingsA cost cap breachThe project has spent its allowance. The run stops at its next step boundary, and the project is marked paused — not just this run.
  2. how_to_regAn approval gateThe hub is waiting for someone to approve what the run proposes. Nothing proceeds until they do, and declining is a valid outcome.
04

What completes a run

A run completes when station H finishes with no approval outstanding. It does not complete because every station reported success — a station can finish having recorded that it skipped its work, and the run still reaches H.

skip_nextOne agent genuinely skips most stationsThe research agent declares that it only works at station A and writes a skipped step at B through H. A trace full of skipped stations for that agent is correct, not broken.
05

The failure most first runs hit

keyNo AI provider key is configuredIn an environment with no provider API key set, every run stops immediately after planning begins with a specific “provider key not configured” error. If that is what you are seeing, the fleet is wired correctly and is missing exactly one credential.

boltPut it into practice

Run a fleet with these guardrails on from day one.

Hard cost caps, one audited gateway in front of every tool call, and versioned rule files — the defaults, not an enterprise add-on.

Start freearrow_forward

On this page

5 sections
  1. 1. The states a run can be inchevron_right
  2. 2. One run is one pass through A–Hchevron_right
  3. 3. Why a run pauseschevron_right
  4. 4. What completes a runchevron_right
  5. 5. The failure most first runs hitchevron_right
feedbackSomething missing?

Related guides

  • articleSetting a per-project cost capSet a cost cap
  • articleConnecting your first MCP serverConnect an MCP server
  • articleHow rule files govern an agent, and how a version is approvedAgent fleet