opendcai
- 47 skills
- 0 followers
- 6 hours ago last updated
- ▌ Solve Physics Problem · opendcaiSolve a physics problem (text + optional figure) using the SciReasoner caption→reason→critic pipeline. Trigger when the user asks to solve, verify, or critique a physics problem that has a diagram, plot, circuit, or other figure.
- ▌ Text2qa · opendcai bundlebuild concept, process, and case-application supervision datasets from markdown books or long markdown documents. use when generating training data from many .md files or precomputed chunk files and when full chunk coverage, resumable batch processing, status tracking, validation, and coverage auditing are required. use for book-to-sft pipelines where every chunk must end in exactly one final status and where answer-only qa is not sufficient because the dataset should also teach grounded reasoning patterns and rule application.
- ▌ Data Construction Skill · opendcai bundlebuild concept, process, and case-application supervision datasets from markdown books or long markdown documents. use when generating training data from many .md files or precomputed chunk files and when full chunk coverage, resumable batch processing, status tracking, validation, and coverage auditing are required. use for book-to-sft pipelines where every chunk must end in exactly one final status and where answer-only qa is not sufficient because the dataset should also teach grounded reasoning patterns and rule application.
- ▌ Expgen · opendcai bundleUse when turning a research idea, paper claim, hypothesis, or experiment request into a generated runnable experiment repository and validated handoff for exprun.
- ▌ Exprun · opendcai bundleUse when expgen has produced an experiment handoff and the task is to execute, monitor, repair operational failures, and report full experiments.
- ▌ DB Ops Sop · opendcaiDatabase operations runbook — backup, recovery, performance tuning, troubleshooting. Use for SQLite / MySQL / PostgreSQL ops questions, slow-query diagnosis, or outage response.
- ▌ Code Review · opendcaiComprehensive code review guidance — process, checklist, feedback conventions. Use when the user asks about code review flow, review criteria, best practices, or how to give/receive review feedback.
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌
- ▌ Create Vfx Effects · opendcaiCreate and control reusable game VFX in Unreal Engine 5 or Unity, including smoke, fire, explosions, dust, stylized ink/frost/cyber effects, and action-attached effects. Use for environmental or combat VFX, Niagara, Unity ParticleSystem, VFX lifecycle, animation/socket binding, or engine code that must reuse an existing effect template.
- ▌ Game Cg Director · opendcai bundleConvert game CG intent and optional media references into validated, model-specific directing envelopes for opening, cutscene, ultimate, or promo clips. Use for standalone CG prompt direction or as a child capability of a game-generation Harness. Supports T2VA/text-only, I2VA/first-frame image, FL2VA/first-and-last-frame images, and Ref2VA/image, video, or audio references; never calls a video-generation API itself.
- ▌
- ▌ Trainer · opendcaiUse this skill when the user wants LoopAI to validate training data, generate and approve training YAML, run LLaMA-Factory SFT or Verl GRPO, configure rewards, monitor or reconnect to Trainer workers, compare checkpoints, export a trained model, or inspect Trainer failures from starter.yaml or runtime state.
- ▌
- ▌
- ▌ Obtainer · opendcaiUse this skill when LoopAI needs dataset discovery, acquisition, web-page collection, DataMixer lakehouse operations, data processing, indexing, recipe planning, or production training-data export. In long-running Codex SDK loops, when Analyzer produces an analysis report, failure taxonomy, or user request that implies new training data is needed, Codex must activate this Obtainer skill, parse the data need into an intent, and delegate the whole workflow to the Obtainer Orchestrator agent (`dm obtainer-orchestrator start`), then poll its structured status. The orchestrator owns lake bootstrap and the dispatch/gating of the managed sub-agents (dataset-acquisition-agent, dataflow agent-run, sft-export-agent); the outer Codex context must not run lake init, acquisition bridges, download manifest, ingest, or export itself for a normal obtain task. The rest of this skill is the domain policy the orchestrator worker follows.
- ▌ Reviewing Dataflow Pipeline · opendcai bundleIndependently audit a generated DataFlow training-data pipeline and its trial artifacts with delegated rubric reviewers before it can be delivered or scaled.
- ▌ Curating Dataflow Pipeline Skills · opendcaiPromote exceptionally reviewed DataFlow pipelines into reusable task-pattern skills, or update an inferior same-task skill while preserving comparison evidence and history.
- ▌ Dataflow Pattern Math Reasoning Sft · opendcai bundleReuse a reviewed pipeline pattern for constructing benchmark-aligned math reasoning SFT records from question-and-answer datasets with verifiable final answers.
- ▌ Core Text · opendcai bundleReference documentation for DataFlow's core_text operators — 8 generators, 3 filters, 2 refiners and 5 evaluators. Read by generating-dataflow-pipeline when a task needs an operator beyond the six core primitives. This is a reference package, not a workflow: it is consulted, never invoked directly.
- ▌ Dataflow Dev · opendcai bundleDataFlow 开发专家上下文加载器。当用户在 DataFlow 仓库中进行开发时触发， 涵盖：新建算子/Pipeline/Prompt、诊断报错、规范审查、 以及感知仓库变更并建议更新知识库。 Trigger: user is developing in DataFlow repo, asks to create operator/pipeline/prompt, encounters errors, wants code review, or asks about operators.
- ▌ Prompt Template Builder · opendcai bundleBuild reusable DataFlow prompt_template classes for existing operators with two-round structured intake, two-stage auditable outputs, and static acceptance checks. Trigger when users ask to generate/rewrite/optimize prompt_template or reuse operator logic with new prompt requirements.
- ▌ Dataflow Operator Builder · opendcai bundleBuild production-grade DataFlow operator scaffolds (generate/filter/refine/eval) for Codex and coding agents. Trigger when users ask to create/new/scaffold operators, add OPERATOR_REGISTRY registration, generate DataFlowStorage-based CLI wrappers, or generate operator unit/registry/smoke tests.
- ▌ Generating Dataflow Pipeline · opendcai bundlePlan and write a standard DataFlow pipeline from a target and representative JSONL data. Use when a user asks to select DataFlow operators, trace field dependencies, generate runnable pipeline code, or repair a pipeline with schema or field-flow errors.
- ▌ General Filter · opendcai bundleReference documentation for the GeneralFilter operator. Covers the constructor, rule-based filtering logic, and pipeline usage notes. Use when: filtering rows based on column value conditions that can be expressed as lambda functions without LLM calls.
- ▌ Prompted Filter · opendcai bundleReference documentation for the PromptedFilter operator. Covers the constructor, actual scoring and filtering behavior, and pipeline usage notes. Use when: filtering rows based on LLM semantic quality judgment rather than simple rule-based conditions.
- ▌ Pandas Operator · opendcai bundleReference documentation for the PandasOperator operator. Use when: applying custom DataFrame transformations without LLM.
- ▌ Prompted Evaluator · opendcai bundleReference documentation for the PromptedEvaluator operator. Use when: scoring text quality with LLM without filtering rows.
- ▌ Prompted Refiner · opendcai bundleReference documentation for the PromptedRefiner operator. Use when: refining text with LLM, overwriting original column.
- ▌ Kcentergreedy Filter · opendcai bundleReference documentation for the KCenterGreedyFilter operator. Covers the constructor, K-Center Greedy algorithm behavior, embedding serving requirements, and pipeline usage notes. Use when: downsampling a large dataset by semantic diversity using embedding vectors.
- ▌ Prompted Generator · opendcai bundleReference documentation for the PromptedGenerator operator. Covers constructor parameters, run() signature, actual row-processing behavior, and pipeline usage notes. Use when: integrating PromptedGenerator into a DataFlow pipeline for single-field LLM generation.
- ▌ Bench Dataset Evaluator · opendcai bundleReference documentation for the BenchDatasetEvaluator operator. Covers the constructor, two comparison modes (match/semantic), and pipeline usage. Use when: comparing predicted answers against ground truth answers in benchmark evaluation.
- ▌ Embedding Generator · opendcai bundleReference documentation for the EmbeddingGenerator operator. Covers the constructor, embedding serving requirements, actual dataframe flow, and runnable pipeline usage. Use when: converting one text column in a dataframe into embedding vectors for retrieval, clustering, similarity search, or downstream vector-based operators.
- ▌ Retrieval Generator · opendcai bundleReference documentation for the RetrievalGenerator operator. [Purpose] Reads one text column from storage, forwards every non-empty row to `llm_serving.generate_from_input(...)`, and writes the returned list into a new output column. [Default backend] Use `LightRAGServing` by default. [Important] `run()` is async. The operator itself does not initialize the serving object; it only awaits `llm_serving.generate_from_input(llm_inputs, system_prompt)`.
- ▌ Text2qa Sample Evaluator · opendcai bundleReference documentation for the Text2QASampleEvaluator operator. Use when: evaluating QA pair quality across multiple dimensions.
- ▌ Bench Answer Generator · opendcai bundleReference documentation for the BenchAnswerGenerator operator. Covers the constructor, full run() signature, actual generation behavior, and integration notes for unified bench evaluation pipelines. Use when: generating model answers from benchmark question rows before passing the dataframe into UnifiedBenchDatasetEvaluator.
- ▌ Text2multihopqa Generator · opendcai bundleReference documentation for the Text2MultiHopQAGenerator operator. [Purpose] Generates multi-hop QA pairs from one text column and writes two output columns: one for `qa_pairs` and one for metadata. [When to use] Use it when you want reasoning-style QA pairs derived from longer text chunks. If only simple single-hop QA is needed, use `Text2QAGenerator` instead.
- ▌ Chunked Prompted Generator · opendcai bundleReference documentation for the ChunkedPromptedGenerator operator. Covers the constructor, file-path based chunking flow, actual prompt construction, and output file writing behavior. Use when: the dataframe stores file paths, the file content may exceed a single LLM context window, and you want the generated results written into new text files.
- ▌ Unified Bench Dataset Evaluator · opendcai bundleReference documentation for the UnifiedBenchDatasetEvaluator operator. Use when: evaluating model answers on benchmark datasets.
- ▌ Bench Dataset Evaluator Question · opendcai bundleReference documentation for the BenchDatasetEvaluatorQuestion operator. Extended version of BenchDatasetEvaluator with question and subquestion support. Use when: evaluating answers with question context or multiple subquestions.
- ▌ Format Str Prompted Generator · opendcai bundleReference documentation for the FormatStrPromptedGenerator operator. Covers the constructor, prompt template restrictions, placeholder-to-column mapping, actual prompt-building logic, and runnable example usage. Use when: one generation task needs multiple dataframe columns combined into a single prompt through a template.
- ▌ Random Domain Knowledge Row Generator · opendcai bundleReference documentation for the RandomDomainKnowledgeRowGenerator operator. [Purpose] Calls an LLM repeatedly with the same domain-generation prompt and writes the generated results into one column of an existing DataFrame. [When to use] Use it when you already have a seed DataFrame with the exact target row count and want to fill one output column with domain-specific generated content. If you need to build prompts from existing row fields, use PromptedGenerator or FormatStrPromptedGenerator instead.
- ▌ One Eval · opendcai bundle驱动 One-Eval 对 API 或本地模型做端到端评测，覆盖纯文本、多模态、代码生成、函数调用和 Agent benchmark。当用户想评测模型在一个或多个 benchmark 上的表现、比较分数、补充 metric，或生成图文评测报告时使用本 skill。