Active goal
Ship a reviewable result
On track
Goal-level control plane
Your agents keep
the night shift.
You keep the judgment.
A durable control plane for long-horizon work across sessions, runtimes, and hosts—keeping authority, gates, todos, quota, evidence, recovery, and handoffs in one loop.
One message to your current agent
01Detects your host
02Preserves existing state
03Activates the loop
Paste it into Codex, Claude Code, OpenCode, Cursor, or another shell-capable agent. It will follow the official guide and select the matching host path.
2 lanes moving
ResearcherGathering & analyzing
Running
WriterDrafting sections
Running
Waiting for you
Human gateReview & approve
Waiting
Evidence written
128 artifacts capturedLogs, docs, outputs
Completed
Explicit gatesHuman judgment at the right moments.
Safe side lanesIndependent work continues without bypassing authority.
Reviewable outcomesEvery transition leaves evidence for the next run.
One durable state kernel
Long-running work stays governed.
LoopX keeps the goal lifecycle, authority, recovery, intervention, validation, and outcomes coherent across every agent run.
Lifetime goal
One stable directionAuthority gate
A concrete decisionAgent lanes
Scoped continuationEvidence
A validated outcomeNext run
Recoverable progressAcross sessions, runtimes, and hosts
Governed continuity for long-horizon agents.
Codex App, Codex CLI, Claude Code, OpenCode, and custom runners can share one goal lifecycle while each agent stays inside an explicit scope.
Durable state
Goals, user decisions, todos, claims, quota, run history, and handoffs survive session boundaries.
Cross-runtime
Change the executor or host without reconstructing the objective from chat history.
Gate aware
A blocked P0 stays visible while independent P1 and P2 lanes continue safely.
Outcome driven
Every run writes validation, ownership, review, and handoff evidence into the next loop.
Evidence from real loops
200+ hours, still legible.
Two public-safe trajectories preserve delivery, decisions, evidence branches, and recovery across many bounded agent turns.
LoopX AgentPublic trajectory 01 · Open-source issue fix
$ loopx status --goal-id issue-fix
GOAL
Ship the focused fix and preserve reusable knowledge.
ACTIVE
P0
Focused PR delivery · bounded diff and validation
RUNNING
GATE
Reviewer authority · approve the focused change
WAITING
P1
Repository knowledge and terminal resume remain available
READY
EVIDENCE
Merge/open result linked to the next reusable discovery
WRITTEN
NEXT
Review the focused diff, then continue the capability lane.
SAFELoopX AgentPublic trajectory 02 · Auto ML experiment
$ loopx status --goal-id auto-ml
BASELINE
Matched candidate and evaluation contract locked.
MATCHED
BRANCHES
PositiveNegativeInvalid lineageRunning replicate
VISIBLE
P1
Replicate capacity stays bounded while evidence accumulates
RUNNING
EVIDENCE
Negative results and invalid lineages remain traceable
WRITTEN
P0 GATE
Promote or stop after the running replicate completes
WAITING
NEXT
Compare replicate evidence, then record the decision.
SAFEElapsed lifetime is wall-clock project time. It does not claim continuous model execution or unattended production autonomy.
Manual fallback
Prefer the shell? Install LoopX manually.
Use this path when your current agent cannot run shell commands. The same official installer registers lightweight command entries for supported hosts.
$ curl -fsSL https://raw.githubusercontent.com/huangruiteng/loopx/main/scripts/install-from-github.sh | bash
$ export PATH="$HOME/.local/bin:$PATH"
$ loopx doctor
$ cd /path/to/your-project
$ loopx connect
$ loopx status
✓ CLI health checked
✓ project state readable
→ finish host activation from the onboarding packet