ado and Agents¶
ado's typed resources, expressive CLI, and bundled agent skills make it a natural fit for agentic research workflows. Once prompted with a research problem, an agent can design the Discovery Space, write new experiments or reuse existing ones, and run the full exploration loop.
Why ado works well with agents¶
| Feature | How it helps agents |
|---|---|
| 🤖 Bundled agent skills | Ready-made skills guide agents through end-to-end discovery workflows — from formulating a problem to analysing results. |
| 🔍 Self-describing resources | Experiments and operators declare their required properties, so an agent can discover what's available and what's needed without parsing code. |
| 🧱 Validated schemas | Research intent is expressed as structured, validated configurations — constraining the agent to well-defined inputs rather than free-form code generation, reducing hallucinations and keeping experiments repeatable. |
| ✅ Safe execution loop | ado template and --dry-run support a tight generate → validate → fix → run cycle before any work is committed. |
| 📦 Structured & queryable results | All measurements and metadata are stored in a structured database, giving agents clean access to data for analysis and refinement. |
| 🔗 Full provenance | Every result is annotated with resource relationships and plugin versions, so an agent always knows where data came from and how to reproduce it. |
What you can ask your agent to do¶
With the skills we provide, you can ask your agent to handle complex tasks in plain language:
| Ask your agent to… | Example prompt | Skill used |
|---|---|---|
| Run a full study | "Design, run, and analyse an experiment to find the best vLLM config for throughput." | conduct-empirical-study |
| Create YAML files | "Formulate a discovery space for my new component." | formulate-discovery-problem |
| Summarise results | "Examine the operation I just ran and tell me what it found." | examining-ado-operations |
| Inspect a project | "Give me an overview of all experiments run in this project so far." | examining-ado-project |
| Query data | "Find all entities where lora_rank was 8 and export their validation_loss." | query-ado-data |
Getting set up for agent-assisted workflows¶
If you haven't already followed Path B in Getting Started, clone the repository and set up the full environment:
git clone https://github.com/IBM/ado.git
cd ado
uv sync --group test
source .venv/bin/activate
Open the cloned ado folder as your workspace root in an agent-enabled IDE (Claude, Cursor, Bob, and others will automatically detect and load the built-in skills).
Next steps¶
-
See ado in action
Walk through end-to-end examples that cover common research workflows.
-
Extend ado
Add custom experiments, benchmarks, or search strategies via the plugin model.