Empower your coding agent with the decision intelligence capabilities of RelationalAI.
Skills are markdown files encoding expert knowledge – heuristics, workflows, and patterns – distributed as folders and installed into a location the agent can discover (e.g. ~/.claude/skills/). At runtime, the agent reads relevant skills to inform its reasoning, and calls tools and APIs to take action.
+---------+
| Agent |
+---------+
/ \
reads calls
/ \
+-------------+ +-------------+
| Skills | | Tools/APIs |
| <knowledge> | | <actions> |
+-------------+ +-------------+
The skills in this repo instruct your agent how to use the relationalai Python package to leverage RelationalAI ontologies and advanced reasoners by generating and executing PyRel code. They roughly map to workflow steps.
- Generating PyRel enables the agent to create the RelationalAI ontology, extend it with reasoners, and use its outputs.
- By executing PyRel, the agent can then query the ontology to answer questions, resolve issues, or help with setup.
Invoke the skills using the /rai- command.
| # | Skill | Area | Description |
|---|---|---|---|
| 1 | rai-onboarding | Setup | First-time setup — install, connect to Snowflake, validate |
| 2 | rai-configuration | Setup | Config files, connections, authentication, model and engine settings |
| 3 | rai-pyrel-coding | Development | Language syntax — imports, types, concepts, properties, data loading |
| 4 | rai-build-starter-ontology | Ontology | Build a first ontology from Snowflake tables or local data |
| 5 | rai-ontology-design | Ontology | Domain modeling — concepts, relationships, data mapping, enrichment |
| 6 | rai-rules-authoring | Ontology | Business rules as PyRel derived properties — validation, classification, alerting |
| 7 | rai-querying | Reasoning | Query construction — aggregation, filtering, joins, ordering, export |
| 8 | rai-discovery | Reasoning | Surface answerable questions, classify by reasoner type, route to workflow |
| 9 | rai-graph-analysis | Reasoning | Graph algorithms — centrality, community detection, reachability, similarity |
| 10 | rai-prescriptive-problem-formulation | Reasoning | Formulate optimization — decision variables, constraints, objectives |
| 11 | rai-prescriptive-solver-management | Reasoning | Solver lifecycle — selection, creation, execution, diagnostics |
| 12 | rai-prescriptive-results-interpretation | Reasoning | Post-solve — solution extraction, status codes, quality, sensitivity |
| 13 | rai-cortex-integration | Operations | Deploy RAI models as Snowflake Cortex Agents |
| 14 | rai-health-skill | Operations | Diagnose engine performance — memory, CPU, demand metrics, remediation |
Assumes relationalai (PyRel) v1.0.13+
The RelationalAI Native App for Snowflake must be installed in your account by an administrator.
- Request access here.
- See the RAI Native App docs for details.
The rai_developer role is needed to execute PyRel programs.
Note: For most coding agents, the installed skills will be available in your next session.
- Ask your agent to copy the contents of this repo's skills folder into its skills folder.
- Vercel's skills CLI (requires
npmv5.2.0+) helps you manage & update skills for most coding agents.
npx skills add RelationalAI/rai-agent-skills --skill '*'
# optionally specify an agent
npx skills add RelationalAI/rai-agent-skills --skill '*' --agent cortexFollow these instructions to point at this repo.
Also see this quick video for an overview.
Example:
/plugin marketplace add RelationalAI/rai-agent-skills
/plugin install rai@RelationalAI
# or use the wizard
/plugin
Restart your session after installing.
Follow these instructions.
In short, clone this repo to your file system then use the /skill dialog to add the skills folder.
Follow these instructions to point at this repo.
Example:
// settings.json
"chat.plugins.marketplaces": [
"RelationalAI/rai-agent-skills"
]
Follow these instructions.
In short, specify this repo's URL https://github.com/RelationalAI/rai-agent-skills.git as a Remote Rule (Github).
Follow these instructions.
In short, download this repo's contents and ask your agent to copy the skills repo contents to the .agents/skills in your working repo root directory
- Open the Claude Desktop app and go to Customize in the left sidebar.
- Under Plugins, browse the directory and find Rai by RelationalAI.
- Click to install, then toggle the plugin on.
Alternatively, you can download this repo's contents and copy the skills into the Claude app.
