mit einem Klick
CargoDash
CargoDash enthält 3 gesammelte Skills von Berdyanskov, mit Repository-Berufsabdeckung und Skill-Detailseiten auf SkillsMP.
Skills in diesem Repository
Complete the empty processing methods in a CargoDash pipeline that a user scaffolded in the WebUI. Use when the user has a .cdgraph.json or an exported pipeline.py whose graph (nodes, edges, schemas) is built but the Processor / Judge / Vote function bodies are blank stubs, and asks you to fill them in. Handles both the .cdgraph.json source-of-truth and the generated pipeline.py.
Build a complete, runnable CargoDash pipeline script from a natural-language description of a data synthesis / augmentation task. Use when the user describes what they want a pipeline to do ("read this jsonl, augment each row with an LLM, filter low-quality samples, write out") and there is no existing graph or scaffold to complete. Produces a single pipeline.py wired with CargoDash primitives.
Authoritative API reference and semantic rules for the CargoDash LLM data-synthesis library (Processor / Judge / Vote / LLMCall / ChatClients / Schema / Pipeline). Read this before writing or completing any CargoDash pipeline script. Shared by cargodash-from-scratch and cargodash-fill-pipeline.