GitHub Copilot CLI Plugin

A single AI agent will miss things.
66 specialized agents will not.

AgentSpec brings Spec-Driven Development to GitHub Copilot CLI — 30 knowledge domains, 41 skills, and a 5-phase workflow that turns vague requests into production-grade data pipelines.

Install now Contribute ↗
$ /plugin marketplace add arthur1511/agentspec-copilot

The 5-Phase SDD Workflow

Every feature follows the same structured path — from vague idea to shipped code. No hallucinated SQL. No wrong incremental strategies. Just decisions backed by 30 knowledge domains.

0

Brainstorm

optional

Explore the idea space. Compare approaches, answer discovery questions, and apply YAGNI before a single requirement is written.

/agentspec:brainstorm → BRAINSTORM_{FEATURE}.md
1

Define

Capture requirements in one pass. Structured problem statement, user personas, success criteria, and a clarity score that must reach 12/15 to proceed.

/agentspec:define → DEFINE_{FEATURE}.md
2

Design

Architecture, file manifest, inline ADRs, and copy-paste code patterns. Every design decision is documented with rationale and alternatives rejected.

/agentspec:design → DESIGN_{FEATURE}.md
3

Build

Execute the file manifest. Build auto-delegates each file to the matched specialist agent — spark-engineer for PySpark, dbt-specialist for SQL models, and so on.

/agentspec:build → Code + BUILD_REPORT_{FEATURE}.md
4

Ship

Archive completed work with lessons learned and KB updates. Every shipped feature makes the next one faster.

/agentspec:ship → archive/{FEATURE}/SHIPPED_{DATE}.md

66 Specialized Agents

Across 9 categories — each agent has deep domain knowledge, confidence scoring, and KB-first resolution against 30 knowledge domains.

6

Other

  • agentspec:brainstorm-agent
  • agentspec:build-agent
  • agentspec:define-agent
  • agentspec:design-agent
  • +2 more
8

Architecture

  • architect-data-platform-engineer
  • architect-genai
  • architect-kb
  • architect-lakehouse
  • +4 more
10

Cloud

  • cloud-ai-data-engineer-cloud
  • cloud-ai-data-engineer-gcp
  • cloud-ai-prompt-specialist-gcp
  • cloud-aws-data-architect
  • +6 more
15

Data Engineering

  • de-ai-data-engineer
  • de-airflow-specialist
  • de-dbt-specialist
  • de-lakeflow-architect
  • +11 more
4

Developer Tools

  • dev-codebase-explorer
  • dev-meeting-analyst
  • dev-prompt-crafter
  • dev-shell-script-specialist
8

Data Science & ML

  • ds-eda-analyst
  • ds-experiment-tracker
  • ds-feature-engineer
  • ds-ml-deployer
  • +4 more
6

Microsoft Fabric

  • fabric-ai-specialist
  • fabric-architect
  • fabric-cicd-specialist
  • fabric-logging-specialist
  • +2 more
6

Python & Code Quality

  • python-ai-prompt-specialist
  • python-code-cleaner
  • python-code-documenter
  • python-code-reviewer
  • +2 more
3

Quality & Contracts

  • test-data-contracts-engineer
  • test-data-quality-analyst
  • test-generator

30 Knowledge Domains

Every agent resolves against this KB before answering. No hallucinated patterns — just validated, up-to-date documentation on the tools you actually use.

pandas

pandas data manipulation — DataFrame wrangling, indexing, groupby, merging, missing data, performance

scikit-learn

scikit-learn ML — Estimator API, Pipelines, preprocessing, cross-validation, classification, regression, model selection

statistical-analysis

Statistical analysis — probability distributions, hypothesis testing, A/B test design, correlation, effect sizes, confidence intervals

data-visualization

Data visualization in Python — matplotlib foundations, seaborn statistical plots, plotly interactive charts, publication-quality figures

time-series

Time series analysis and forecasting in Python — stationarity, decomposition, ARIMA, Prophet, ML-based forecasting, and evaluation metrics

mlflow

MLflow experiment tracking and model registry — run logging, autologging, artifact management, model versioning, production serving

dbt

dbt development patterns — Fusion Engine, Mesh, Semantic Layer, models, macros, tests

spark

PySpark and Spark SQL patterns — DataFrames, performance, Real-Time Mode, Delta integration

airflow

Orchestration patterns — Airflow 3.x TaskFlow, Dagster, Prefect comparison, DAG design

sql-patterns

Cross-dialect SQL — window functions, CTEs, deduplication, DuckDB/Snowflake/BigQuery/Spark

streaming

Stream processing — Flink, Kafka, Spark Streaming, RisingWave, Materialize, CDC

data-modeling

Schema design — dimensional modeling, Data Vault, SCD types, schema evolution

data-quality

Data quality, contracts, and observability — Soda, GE, dbt tests, ODCS, Monte Carlo

lakehouse

Open table formats and catalogs — Iceberg v3, Delta Lake 4.1, DuckLake, Unity, Gravitino

cloud-platforms

Cloud data platforms — Snowflake Cortex, Databricks LakeFlow, BigQuery AI, cross-platform

medallion

Medallion Architecture — Bronze/Silver/Gold layer design, data quality progression, schema evolution

aws

AWS data engineering — Lambda, S3, Glue, Redshift, MWAA, SAM deployment

gcp

Google Cloud Platform — Cloud Run, Pub/Sub, GCS, BigQuery, IAM, Secret Manager

microsoft-fabric

Microsoft Fabric — Lakehouse, Data Warehouse, Pipelines, Real-Time Analytics, KQL, CI/CD, AI

lakeflow

Databricks Lakeflow — DLT pipelines, materialized views, streaming tables, expectations, DABs

ai-data-engineering

AI data engineering — RAG pipelines, vector DBs, feature stores, LLMOps, embeddings

modern-stack

Modern data tools — DuckDB, Polars, SQLMesh, Malloy, Evidence.dev, local-first analytics

prompt-engineering

Prompt engineering — chain-of-thought, structured extraction, few-shot, system prompts

genai

GenAI architecture — multi-agent systems, RAG, state machines, tool calling, guardrails

pydantic

Pydantic patterns — BaseModel, validators, LLM output validation, extraction schemas

python

Python patterns — dataclasses, type hints, generators, async, project structure

testing

Testing patterns — pytest, fixtures, mocking, parametrize, integration tests, Spark testing

terraform

Terraform IaC — resources, modules, providers, state, workspaces, GCP/AWS patterns

xgboost

XGBoost gradient boosted trees — training pipelines, hyperparameter tuning, early stopping, feature importance

supabase

Supabase development -- pgvector, RLS policies, Edge Functions, Auth, Realtime, migrations

Get started in 30 seconds

Three ways to install — pick the one that fits your workflow.

Recommended

Marketplace (Recommended)

One command, always up to date

/plugin marketplace add arthur1511/agentspec-copilot
/plugin install agentspec@agentspec

Clone & install locally

Install from disk for offline use

git clone https://github.com/Arthur1511/agentspec-copilot.git
copilot plugin install ./agentspec-copilot/plugin-copilot

Build from source

For contributors and customization

git clone https://github.com/Arthur1511/agentspec-copilot.git
cd agentspec-copilot
./build-copilot.sh          # Linux / macOS
.\build-copilot.ps1        # Windows (PowerShell)
copilot plugin install ./plugin-copilot

After installing, every Copilot CLI session has 66 agents, 41 skills, and 30 KB domains available instantly.

Documentation

Everything you need to go from zero to shipping production data pipelines with AgentSpec.

Getting Started ↗

Install AgentSpec, run your first skill, and understand the SDD workflow in 5 minutes.

Concepts ↗

Understand agents, skills, KB domains, confidence scoring, and the SDD phase contracts.

Tutorials ↗

Step-by-step walkthroughs: build a Spark pipeline, design a star schema, run Airflow DAGs.

Reference ↗

Full agent list, skill catalogue, KB domain index, and SDD template reference.