build // Lightning · Infrastructure · Open Source

Mission-Control Style Router

Score graph edges with fees, CLTV, and historical failures. Benchmark success rate against a naive Dijkstra baseline.

Advanced2 weeks

Overview

Pathfinding is a product differentiator. Build a service that proposes routes using your own scoring function.

Without a benchmark harness, "smarter routing" is just storytelling. Log attempt outcomes and measure success rate, fees, and retries.

Heuristics to try

Fee cost + risk penalty
Channel age / capacity priors
Historical failure decay (mission-control style)
Avoid recently failed edges for a cooldown window

Learning objectives

Ingest a channel graph and invoice constraints
Rank paths with custom fee and reliability heuristics
Benchmark success rate against a baseline router

Prerequisites

Pathfinding
Failure Handling
Onion Routing

Required skills

PathfindingGraph algorithmsBenchmark harnesses

Part of the lightning roadmap.

Architecture diagram

Graph snapshot → scorer → path candidates → payment attempt feedback

Requirements

`POST /routes` accepts destination, amount, max fee, max CLTV
Returns ranked candidate paths with expected fees
At least two scoring strategies behind a flag
Deterministic routes for a fixed graph fixture
Comparison report: success rate, average fees, attempts

Stretch goals

Feedback endpoint for success/failure learning
Multipart / split payment path sets under a fee budget
Cooldown window for recently failed edges
Visualization of chosen vs rejected paths

Recommended resources

Example repositories

lnd routing package
Production pathfinding and mission-control ideas.
CLN askrene / plugins
Plugin-oriented routing experimentation surface.

Submission instructions

Service repo with fixture graph and benchmark harness
Written report comparing strategies on the same dataset
API examples for route requests/responses
Clear limitations vs full node private-state routers

Discussion

Loading page