YYLO · 0.1.0-rc.19
Bound coding-agent execution.
Run providers through equivalent yy and yylo launchers.
Developer tools / JUNO AI INC.
Put coding agents to work without losing the thread.
YYLO (why-lo) turns a defined outcome into bounded, inspectable iterations—and connects each run to durable task truth and evaluation evidence.
$ yy init --task
"Add keyboard navigation" --subagent pi
task explicit outcome
loop max iterations: 5
proof tests · commit · response
✓ yylo is an equivalent launcherWorkflow / inspectable by default
A loop with a beginning, evidence, and a handoff.
YYLO does not replace your repository workflow. It makes agent execution participate in it.
- 01Task truthA prompt or kanban task defines the bounded outcome.
- 02Agent runChoose a provider; YYLO carries context through iterations.
- 03EvidenceCommits, task responses, logs, and session IDs remain inspectable.
- 04ContinueResume the right session or hand the result to the next step.
Three products / one contract
Orchestrate, record, evaluate.
YYLO Ledger · 0.1.0rc2
Preserve task truth.
Track dependencies, receipts, responses, archives, and commit evidence.
Explore Ledger →YYLO Benchmark · 0.1.0-rc.5
Compare runs with evidence.
Validate cases, isolate snapshots, reconcile artifacts, and report.
Explore Benchmark →Quick start / YYLO 0.1.0-rc.19
Give the first run a verifiable finish.
Install the RC after publication, choose a provider, and keep the first scope small enough to review.
Begin with one task