DSL knowledge base · preview
Write NoETL playbooks against the grammar that will run them.
Everything here is generated from noetl-orchestrate-core, the types
the server parses playbooks with. Nothing is hand-written, so an AI grounded on it cannot be
taught a field or tool kind the server does not have.
Version
- loading
- …
Tool kinds
Accepted by the parser under default gates. Accepted is not the same as executable: which kinds a live worker can run is reported by workers (RFC Phase 3).
Flag-armed features
Catalog references
Worker runtime
Resource and performance model
Every figure is labelled MEASURED, DEFAULT, DEPLOYED,
GUIDANCE, TARGET or NOT_YET_MEASURED. Hover an item for its source.
Full model: /dsl/resource-model.json.
Step kinds
Pools and scaling
Measured overhead
Planning heuristics
Not yet measured
Generate a playbook
Prompt or spec to a validated playbook happens in the NoETL console, which calls this knowledge base and its validator, so the console and every AI share one source of truth. The console generation workspace is on the roadmap.
For AIs and tools
GET /schema/v2/playbook.schema.json: release artifact (static)GET /dsl/manifest.json,/dsl/source.json,/llms.txt: staticGET /dsl/resource-model.json: slots, limits, measured latency, planning heuristics (static)GET /api/dsl/{version,schema,tools}: live, from the deployed serverPOST /api/dsl/validate: live dry-run validation, rule-tagged findingsPOST /mcp: MCP toolsnoetl.dsl.describe,noetl.dsl.schema,noetl.dsl.validate