An AI agent can execute a command successfully and still fail the user’s task. A script may return zero while changing the wrong file. A test may pass while the requested behavior remains unimplemented. Nerva should make verification a separate step after tool execution.

Verification depends on the task

For a code edit, useful evidence may include:

  • The final diff matches the approved proposal.
  • The focused test passes.
  • Type checking completes without new errors.
  • The expected file exists at the intended path.

For a file-organizing task, verification might compare file counts and paths. For a browser task, it might confirm the expected page state without submitting another form. There is no single “AI confidence” number that proves every result.

Show incomplete evidence honestly

If a command was not run, Nerva should say “not verified.” If a test failed for an unrelated environment problem, it should show that failure rather than calling the task complete. If the model predicted a result but no tool checked it, that remains a prediction.

This is especially important in backend systems, where a green unit test may not prove queue behavior, transaction boundaries, or production configuration. Verification should match the claim being made.

The result can be compact:

Approved edit: matched
Focused test: passed
Full test suite: not run
Network behavior: not verified
Status: locally verified, broader behavior unknown

That kind of honesty makes the agent more useful because the developer knows what to inspect next.

The useful design question

The useful question here is not whether an AI agent can produce an impressive demo. It is whether the workflow stays understandable when the task is incomplete, the repository is messy, or the model is wrong. For this part of Nerva, the focus is proving outcomes with task-specific evidence. That means each state should have a visible owner, a small input, and a result that another part of the system can inspect.

A concrete scenario

Consider checking a test result and final diff after an automated code edit. A good Nerva run would first state the scope, identify the files or operation it needs, and show the next proposed step. It would keep the original request separate from model-generated instructions. If an action needs approval, the user would see what will happen before it happens. Afterward, the task history would retain the command, result, and verification status instead of reducing everything to a confident paragraph.

This approach also makes a failure useful. A denied permission, unavailable provider, missing package, or changed file is information about the task state. Nerva should preserve it and explain the smallest recovery option. It should not silently broaden the request, retry a side effect, or claim completion because a model supplied plausible-looking text.

What I would test first

For an early implementation, I would test three paths: a normal successful run, a denied action, and an interrupted run resumed after the workspace changes. The key test is what Nerva should say when it cannot verify the requested result. In each path, the interface should make clear what Nerva knows, what it proposed, what actually ran, and what remains unverified.

Useful acceptance checks include:

  • the user can inspect the selected context before model execution;
  • sensitive paths and credentials are excluded or redacted;
  • every side effect has a narrow permission decision;
  • failures remain attached to the step that caused them;
  • a final answer links to evidence instead of implying that unrun work succeeded.

Nerva is still an upcoming project, so these are design and implementation goals rather than claims that a finished product already provides them. That distinction matters: a trustworthy agent should earn confidence through observable behavior, not through a feature list.

The boundary that matters in production

Verification should be selected from the request rather than bolted on at the end. A formatting task may need a diff and formatter result; a bug fix may need a focused test and reproduction; a migration may need a database state check. When evidence is unavailable, “not verified” is a correct result. It gives the developer a chance to finish the check manually instead of hiding uncertainty.

For the first vertical slice, keep the boundary small enough to explain in a code review. Define the input, the decision, the side effect, and the evidence for completion. Add a test for the denied path as well as the successful path. This is where a local-first design earns trust: the user can inspect the context and history on the device, see when a provider is involved, and stop before an action crosses the permission boundary. The design should remain honest about its limits. Nerva is planned software, not a claim that every workflow can be made safe automatically. A clear “cannot verify” state is more useful than a polished but unsupported success message.

A review of this design should use a small fixture repository and record both allowed and rejected outcomes. That keeps the discussion concrete: reviewers can compare the requested scope with the selected context, action, and evidence. It also prevents Nerva from being judged only by how persuasive its response sounds.

The design is informed by the boundaries I have had to think through in CryptoEx and SentinelFi: backend work becomes easier to review when context, permissions, side effects, and verification are explicit.