Local AIPROPRO REQUIREDFC-AI-009
Dataset Claw
dataset-claw
v0.2.0May 22, 2026Dataset engineering — deduplication, contamination detection, class balance, and HuggingFace dataset management
A dataset is not a file — it is an opinion about your model encoded as examples. I care about class balance ratios, annotation inter-rater agreement (Fleiss' kappa, not just a percentage), train/dev/test split contamination, and whether your "high quality" dataset is actually high-variance enough to avoid memorisation. I will audit your JSONL/Parquet files for schema drift, duplicate samples via embedding cosine similarity, and scoring rubrics that conflate helpfulness with verbosity. I generate cleaned, deduplicated, rebalanced datasets with datasheets-for-datasets provenance cards.
OpenClaw
Hermes
Claude Code
Codex
Cursor
Windsurf
Aider
OllamaPRIMARY ACTION
Unlock with ProCOMPATIBLE WITH
OpenClawHermesClaude CodeCodex+4
8 compatible tools4 output formats
When to Use
- Run local models with private workflows
- Tune inference for local hardware
- Choose effective GGUF variants
- Benchmark practical latency and quality
Compatible Frameworks
OpenClawHermesClaude CodeCodexCursorWindsurfAiderOllama8 TOOLS
Quality Gates
- Complete data pipeline documented
- Quality filters calibrated to domain
- Deduplication effective and verified
- Standard formats used and validated
- Licenses and consent verified
4 GATES DEFINED
Expected Outputs
Dataset Audit Report with Duplicate, Drift, and Balance DiagnosticsCleaned and Deduplicated Dataset with Provenance ManifestDatasheet-for-Datasets Card with Annotation Protocol and Bias NotesRebalanced Split Configuration (Train/Dev/Test) with Contamination-Free Guarantee
Native exports per tool
OpenClaw10 files
openclaw/AGENTS.mdopenclaw/SOUL.mdopenclaw/TOOLS.md+7 moreHermes5 files
hermes/skills/flickclaw/dataset-claw/SKILL.mdhermes/skills/flickclaw/dataset-claw/references/workflow.mdhermes/skills/flickclaw/dataset-claw/references/quality-gates.md+2 moreClaude Code6 files
claude-code/CLAUDE.mdclaude-code/.claude/skills/dataset-claw/SKILL.mdclaude-code/.claude/skills/dataset-claw/references/workflow.md+3 moreCodex5 files
codex/AGENTS.mdcodex/.flickclaw/agents/dataset-claw/codex.mdcodex/.flickclaw/agents/dataset-claw/workflow.md+2 moreCursor3 files
cursor/.cursor/rules/flickclaw-dataset-claw.mdccursor/.cursor/rules/flickclaw-dataset-claw-workflow.mdccursor/.cursor/rules/flickclaw-dataset-claw-quality-gates.mdcWindsurf3 files
windsurf/.windsurf/rules/flickclaw-dataset-claw.mdwindsurf/.windsurf/rules/flickclaw-dataset-claw-workflow.mdwindsurf/.windsurf/rules/flickclaw-dataset-claw-quality-gates.mdAider3 files
aider/CONVENTIONS.mdaider/aider.mdaider/.aider.conf.ymlOllama4 files
ollama/Modelfileollama/system-prompt.mdollama/template.md+1 moreInstall Commands
Install the FlickClaw CLI, then select your AI agent framework below to get the correct install command.
Step 1: Install CLI (one-time)
npm install -g @flickclaw/cli@latestStep 2: Select Framework
OpenClaw
npm exec --yes @flickclaw/cli@latest -- install dataset-claw --target openclawDownload as ZIP
Example Prompt
Try this prompt with Dataset Claw to see what it can do:
Optimize the local model setup for performance. Benchmark current config and suggest improvements for Dataset Audit Report with Duplicate, Drift, and Balance Diagnostics, Cleaned and Deduplicated Dataset with Provenance Manifest, Datasheet-for-Datasets Card with Annotation Protocol and Bias Notes.Example Output
IllustrativeWhat a typical Dataset Claw report looks like:
# Dataset Claw — Assessment Report **Project**: ollama-deployment **Context**: a local LLM deployment running Llama 3.1 8B on consumer GPU hardware **Generated**: 2026-07-10 ## Executive Summary The Dataset Claw completed its review of ollama-deployment (a local LLM deployment running Llama 3.1 8B on consumer GPU hardware). 3 findings were identified with concrete remediation steps. All quality gates were verified before delivery. ## Findings | # | Severity | Area | Finding | Recommended Action | |---|----------|------|---------|-------------------| | 1 | **P1** | Performance | Inference latency spikes to 8s under concurrency | Enable continuous batching and set max_batch=4 | | 2 | **P2** | Memory | Model consumes 18GB VRAM, headroom insufficient | Switch to Q4_K_M quantization, target <12GB | | 3 | **P2** | Setup | No Modelfile with system prompt defined | Create Modelfile with role, constraints, and templates | ## Quality Gates - [✓] duplicate_detection_via_embedding_cosine - [✓] class_balance_audit_with_rebalancing - [✓] inter_annotator_agreement_fleiss_kappa ## Outputs Generated - **Dataset Audit Report with Duplicate, Drift, and Balance Diagnostics**: Included in the report above. - **Cleaned and Deduplicated Dataset with Provenance Manifest**: Included in the report above. - **Datasheet-for-Datasets Card with Annotation Protocol and Bias Notes**: Included in the report above. - **Rebalanced Split Configuration (Train/Dev/Test) with Contamination-Free Guarantee**: Included in the report above. ## Validation - [x] All quality gates passed (3/3) - [x] 3 findings documented with severity and remediation - [x] 4 output sections generated - [x] Evidence collected and referenced --- *This is an illustrative example output from FlickClaw. Results vary based on project context.*
RELATED AGENTS