AgentScope-AI
Official@agentscope-ai · China
Offers a modular framework for multi-worker orchestration, control system engineering, and enterprise data processing across distributed HiClaw environments.
Agent Skills by AgentScope-AI
Showing 155 vetted skills indexed across 7 GitHub repositories.
agentscope-aimmx-cli
Generate text, images, video, speech, and music via the MiniMax AI CLI.
agentscope-airl-reward
Build RL reward signals for GRPO, DPO, and Best-of-N using OpenJudge graders.
agentscope-aieval-design
Design stratified evaluation datasets and adversarial test cases for OpenJudge grading pipelines.
agentscope-airag-eval
Diagnose RAG pipelines by separating retrieval quality from generation faithfulness.
agentscope-airedteam
Generates policy-derived attack prompts and computes ASR paired with over-refusal rates for LLM safety audits.
agentscope-aimetric-design
Select, configure, and combine evaluation graders into executable OpenJudge pipeline code.
agentscope-aimeta-eval
Diagnose evaluation scenarios and route users to the appropriate OpenJudge evaluation workflow.
agentscope-aiprompt-regression
Compare baseline and candidate prompts with position-debiased pairwise win rates and bootstrap confidence intervals.
agentscope-aibootstrap
Generates a v0 evaluation grader and calibration roadmap from scratch using OpenJudge.
agentscope-aialign-human
Measure judge-human agreement with TPR/TNR, Cohen's kappa, and bias detection from paired JSONL labels.
agentscope-aieval-report
Synthesizes evaluation run history into maturity assessments, weakness diagnoses, and prioritized recommendations.
agentscope-aidata-analysis
Computes basic statistics and writes summaries for user-provided datasets.
agentscope-aisummarizer
Condenses provided content into three to five bullet-point key takeaways.
agentscope-aitriage-ticket
Classifies inbound customer tickets and drafts the first-response reply.
agentscope-aisynthesize-report
Generates citation-grounded Markdown reports from workspace source files.
agentscope-aiexample-skill
Provides a starter template for authoring custom agent skill playbooks.
agentscope-aicode-search
Searches codebases with ripgrep regular expressions and git history queries.
agentscope-aigit-checkpoint
Creates git checkpoint commits before risky changes and rolls back failed edits.
agentscope-aiapply-patch
Applies multi-file code edits atomically using git apply with unified diffs.
agentscope-aiverify-changes
Runs project typecheck, build, lint, and targeted tests to verify code changes.
agentscope-aichart-rendering
Generate matplotlib charts from tidy analysis results as PNG images with reproducible scripts.
agentscope-aisql-analysis
Write, validate, and report SQL queries answering quantitative business questions.
agentscope-aimcporter
Discover and call MCP Server tools via the mcporter CLI.
agentscope-aiteamharness-organization
Defines team roles and responsibility boundaries for multi-agent collaboration workflows.
Frequently Asked Questions About AgentScope-AI
FAQPage SchemaWhat specific tasks can I perform using these skills?▼
You can orchestrate multi-worker teams, manage industrial control system parameters like PID gains, perform DC power flow analysis, and process complex document formats including PDF, XLSX, and DOCX within a unified HiClaw environment.
Which personas benefit most from this registry?▼
This registry is designed for systems engineers, control theorists, and distributed infrastructure architects who require granular management of worker nodes, state-space linearization, and cross-agent communication protocols.
How are these skills deployed and executed?▼
Skills are installed via the registry and invoked through the HiClaw manager-worker architecture, utilizing MinIO for file synchronization and Matrix for inter-agent messaging and notification routing.