auto-knowledge-base

skardi-skills

Agent Skills for working with Skardi — a backend engine to provide SQL pipelines as http endpoints. These skills give your AI coding agent (Claude Code, Cursor, or any Agent Skills-compatible tool) deep knowledge of Skardi's patterns so you don't have to re-explain them each session.

Check out our demo here.

Available skills

Directory	Skill name	What it covers
skardi_on_sealos/	skardi-deploy-and-patterns	Core Skardi concepts (auth, pipelines, CSRF, DataFusion SQL dialect) + deploying to Sealos via kubectl
auto_knowledge_base/	auto_knowledge_base	Agent-autonomous knowledge-base construction — turn any directory of text/markdown into a queryable KB using Skardi CLI + SQLite + sqlite-vec + FTS5 (Postgres/pgvector and Lance supported as overrides). Handles prereq detection, model download, chunking, ingest, and hybrid (vector + full-text + RRF) retrieval end-to-end. Supports candle, gguf, and remote_embed UDFs.
auto_rag/	auto_rag	Server-backed RAG over a user-supplied datastore (Postgres+pgvector, MongoDB, or Lance) — runs skardi-server in front of the user's DB so retrieval lives on a network endpoint instead of a local SQLite file. Renders ctx + ingest/search-vector/search-fulltext/search-hybrid pipelines, starts the server (local-process / Docker / Kubernetes), drives ingestion and querying over HTTP, and embeds client-side via the host CLI so the same templates run unchanged across all three runtimes. Never creates schema for the user — prints the SQL and waits.

Installation

Claude Code (plugin marketplace, recommended)

All three skills are published as plugins in this repo. From inside any Claude Code session:

/plugin marketplace add SkardiLabs/skardi-skills
/plugin install skardi-deploy-and-patterns@skardi-skills
/plugin install auto-knowledge-base@skardi-skills
/plugin install auto-rag@skardi-skills

That's it — the skills are now available across all your projects, and /plugin marketplace update skardi-skills pulls future versions.

Claude Code (manual copy)

If you'd rather not use the plugin marketplace, copy the skill(s) into your personal skills directory so they're available across all projects:

# skardi-deploy-and-patterns (deployment + core concepts)
cp -r skardi_on_sealos/skills/skardi-deploy-and-patterns ~/.claude/skills/skardi-deploy-and-patterns

# auto_knowledge_base (agent-autonomous KB construction)
cp -r auto_knowledge_base/skills/auto_knowledge_base ~/.claude/skills/auto_knowledge_base

# auto_rag (server-backed RAG over a user-supplied datastore)
cp -r auto_rag/skills/auto_rag ~/.claude/skills/auto_rag

Claude Code will automatically load the relevant skill when your request matches it — e.g. deployment/auth/pipelines for skardi-deploy-and-patterns, "index these docs" / "build a RAG" / "make this folder searchable" for auto_knowledge_base, or "expose hybrid search as HTTP" / "RAG service over our pgvector DB" / "skardi-server with MongoDB" for auto_rag. You can also invoke any of them directly:

/skardi-deploy-and-patterns
/auto_knowledge_base
/auto_rag

Cursor

Copy the skill(s) into the project-level skills directory:

# skardi-deploy-and-patterns
cp -r skardi_on_sealos/skills/skardi-deploy-and-patterns .cursor/skills/skardi-deploy-and-patterns

# auto_knowledge_base
cp -r auto_knowledge_base/skills/auto_knowledge_base .cursor/skills/auto_knowledge_base

# auto_rag
cp -r auto_rag/skills/auto_rag .cursor/skills/auto_rag

Other Agent Skills-compatible tools

The SKILL.md files follow the Agent Skills open standard and work with any compatible tool. Place the skill directory wherever your tool resolves personal or project skills.

Bundled resources per skill

skardi_on_sealos/skills/skardi-deploy-and-patterns/templates/

Ready-to-use files referenced by skardi-deploy-and-patterns:

File	Purpose
skardi-sealos.yaml	Kubernetes manifest for deploying Skardi on Sealos
nextjs-sealos.yaml	Kubernetes manifest for a Next.js frontend on Sealos
Dockerfile.nextjs	Dockerfile for a Next.js app
nextjs-proxy.ts	Next.js API route that proxies requests to Skardi
docker-compose.yml	Local development stack
init-db.py	Database initialisation script

auto_knowledge_base/

Executable scripts, YAML templates, and reference docs the skill invokes: