Skip to main content
Agents are best when the task needs code, tools, or per-input adaptation instead of a fixed workflow graph. The source lives in Git, so you work with normal files and commits.

​
1. Clone the agent source

eigenpal agents clone --out tmp/org-repos/acme-local
cd tmp/org-repos/acme-local
Use a directory name that makes the organization and environment obvious.

​
2. Edit the agent

Open the agent files, change the prompt, tools, or code, then run any local checks your project defines. 
git status

​
3. Commit, push, and release

eigenpal agents save
eigenpal agents release 1.0.1
eigenpal agents sync agents.extract-invoice
save validates, commits, and pushes local changes. release creates an immutable source tag, and sync points the automation at the latest release.

​
4. Run the agent

eigenpal run agents.extract-invoice --input-file document=invoice.pdf
The command prints a run id. Poll or watch it like any other run: 
eigenpal runs watch <run-id>
eigenpal runs trace <run-id>

​
5. Improve with feedback

When a run is wrong, add feedback and promote the corrected case into the dataset: 
eigenpal runs feedback update <run-id> --rating fail --message "Missed the VAT id"
eigenpal runs promote <run-id> --name missed-vat-id
The next experiment can now catch the same miss before it ships again.