| name | landing-orchestrator |
| description | Inspect and manage an AI Landing Factory run, refresh run-manifest.json, and move the run through intake, build, audit, and release. Use when coordinating multiple landing skills or checking readiness between stages. |
Landing Orchestrator
This skill is the control plane for landing runs. It should prefer the helper script over manual manifest editing.
Primary Commands
python scripts/landing_factory/run_pipeline.py status <run-dir>
python scripts/landing_factory/run_pipeline.py sync <run-dir> --auto-advance
python scripts/landing_factory/run_pipeline.py sync <run-dir> --aggregate-scorecard
python scripts/landing_factory/run_pipeline.py advance <run-dir> --to build
Stage Model
intake: owned by landing-intake
build: owned by landing-strategy, landing-copy, landing-design-system, landing-ui-build
audit: owned by the audit swarm
release: owned by landing-release with scorecard evidence
Workflow
- Run
status first to inspect the current manifest.
- If audits changed, run
sync --aggregate-scorecard.
- If the current stage is ready, use
sync --auto-advance or advance --to <stage>.
- Keep the run grounded in artifacts, not chat memory.
What It Adds
- per-stage
skill_checks
- one
recommended_skill with a concrete reason
- release-time
repair_queue when scorecard gates fail or remain pending
Guardrails
- do not advance stages while required artifacts are still missing unless the user explicitly wants an override
- prefer
advance --force only for deliberate rollback or manual operator intervention
- treat
run-manifest.json as the operational truth for the current landing run