Reference for the people running spacecraft.

Install & auth

The LTL.ai SDK ships in Python, Go, TypeScript, and C++. All authenticated via short-lived OIDC tokens, optionally backed by hardware keys for production-impacting actions.

# Python
pip install ltlai-sdk

# Go
go get github.com/ltl-ai/ltlai-go

# TypeScript
npm install @ltlai/sdk

from ltlai import Client

client = Client(
    tenant="helio-prod",
    region="us-west-2",
    auth="oidc",                 # or "hardware-key"
)

Your first plan

Plans are immutable artifacts. Once submitted, you can simulate, validate, approve, or reject — but never mutate. This is what gives the audit log its meaning.

plan = client.plans.create(
    asset="TUG-07",
    objective={
        "type": "transfer",
        "target_orbit": {"a": 42164e3, "e": 0.0001, "i": 0.05},
        "deadline": "2026-08-14T00:00:00Z",
    },
    constraints=["power-envelope", "eclipse-aware", "itar-default"],
    optimize_for="propellant",        # or "time", "composite"
)

print(plan.cost.delta_v, plan.duration, plan.confidence)
# 156.4 m/s · 14h 26m · 0.982

POST /v1/plan

Submit a mission objective. Returns a plan handle that can be simulated, validated, and (after approval) commanded.

POST /v1/plan
Authorization: Bearer <oidc-token>
Content-Type: application/json

{
  "asset": "TUG-07",
  "objective": { "type": "transfer", "...": "..." },
  "constraints": ["power-envelope", "eclipse-aware"],
  "optimize_for": "propellant"
}

200 OK

{
  "plan_id": "pln_8KJ2N9F",
  "status": "draft",
  "cost": { "delta_v_mps": 156.4, "propellant_kg": 3.42 },
  "duration_s": 51960,
  "confidence": 0.982,
  "validation": { "status": "pending" }
}

POST /v1/approve

Sign a plan for execution. Dual-control plans require two distinct approver identities. Hardware-key signatures available where the tenant policy requires it.

Sample audit log

Every action — plan created, simulation run, gate evaluated, approval signed, command uplinked, telemetry received — is written to the immutable audit log. Exportable as NDJSON, signed per record.

{"ts":"2026-05-26T14:02:11.482Z","ev":"plan.created","plan":"pln_8KJ2N9F","actor":"u_anya","asset":"TUG-07","sig":"ed25519:..."}
{"ts":"2026-05-26T14:02:13.901Z","ev":"solver.complete","plan":"pln_8KJ2N9F","cost":{"dv":156.4},"sig":"ed25519:..."}
{"ts":"2026-05-26T14:02:14.118Z","ev":"gate.power.pass","plan":"pln_8KJ2N9F","sig":"ed25519:..."}
{"ts":"2026-05-26T14:02:14.122Z","ev":"gate.eclipse.pass","plan":"pln_8KJ2N9F","sig":"ed25519:..."}
{"ts":"2026-05-26T14:05:42.073Z","ev":"approve.signed","plan":"pln_8KJ2N9F","actor":"u_anya","key":"yk-7","sig":"ed25519:..."}
{"ts":"2026-05-26T14:08:11.402Z","ev":"approve.signed","plan":"pln_8KJ2N9F","actor":"u_jonas","key":"yk-3","sig":"ed25519:..."}
{"ts":"2026-05-26T14:08:11.518Z","ev":"command.uplink","plan":"pln_8KJ2N9F","network":"ksat-svalbard","sig":"ed25519:..."}

Sample policy file

Policy is declarative. Stored in your tenant's git repo, reviewed via pull request, evaluated on every plan. Below is a minimal example.

name: transfer-default
applies_to: { mission_type: transfer }

gates:
  - name: power-envelope
    rule: "plan.power.peak_kw <= asset.power.envelope_kw"
    severity: block

  - name: eclipse-budget
    rule: "plan.eclipse.total_minutes <= 45"
    severity: block

  - name: propellant-reserve
    rule: "asset.propellant.remaining_kg - plan.propellant_kg >= 8.0"
    severity: block

  - name: confidence-floor
    rule: "plan.confidence >= 0.95"
    severity: warn

approvals:
  required: 2
  distinct_actors: true
  hardware_key: true

Architecture in one diagram

LTL.ai is a five-layer stack with clean contracts between layers. Every layer is independently scalable and replaceable.

Experience → Intelligence (TACK) → Data (twin + telemetry) → Connectivity (ground network) → Foundation (security & tenancy). Full detail on the Architecture page.

Console basics

The operator console is built around three primary surfaces: Fleet (every asset at a glance), Plan (compose, simulate, validate, approve), and Audit (search, export, replay). Key shortcuts:

• g f — go to fleet
• g p — go to plan composer
• g a — go to audit search
• cmd/ctrl + k — open command palette
• ? — show all shortcuts

Runbook templates

Each tenant ships with a library of runbook templates: anomaly response, conjunction avoidance, end-of-life, safe-mode recovery, and post-burn telemetry checks. Customize per asset class, version-control alongside policy.

Want the full handbook? Customer-only content includes the complete API reference, all 14 SDK quickstarts, the operator certification course, and the integration cookbook. Talk to us about access.