AI agents

What each Nelix Agent specialist does, how it is configured, what it may touch, and how its work is verified.

• Agent architecture and responsibilitiesFive jobs, each done by its own agent, and the one rule they all share.
• How agents are configuredThe seven layers that shape what each Nelix Agent specialist knows, may do, and must prove.
• How agents use docs/DESIGN.mdWhich agents read the specification, what they take from it, and why it is treated as data.
• Models and task levelsHow Nelix Agent matches the size of the model to the size of the task.
• Context, permissions and boundariesWhat uses AI credits, what Nelix Agent can see, and what happens to your design.
• Generate a testbenchHave Nelix Agent write a first test for your design, and see it run before you accept it.
• Diagnosis, proposed patches and verificationHow Nelix Agent finds the cause of a failure, and why a repair is tested before you can accept it.
• Build specialists and optimisationThe specialists that read a finished build, and how a proposed timing fix is measured before you accept it.
• Source provenance and limitationsWhy every claim in the lab can be traced to a run, and every result to a version of the source.