Prerequisites
- You are a member of the workspace.
- The target reports as connected.
- Your role permits reading the target and creating read-only runs.
- At least one model provider is ready for the workspace.
1. Review the target
Open the target and confirm that current inventory or findings are available. An offline target may remain visible, but AcornOps cannot gather fresh evidence or invoke its tools.2. Start in read-only mode
Open the target’s chat surface and select read-only access. Ask a bounded question that names the symptom and desired evidence, for example:3. Review effective capabilities
Before launch, inspect the tools that the target and your role make available. Target chat uses the target’s generic assistant; it does not select one of the workspace’s specialist Agents. Type/ in the composer to reference an exact enabled tool or skill. A reference helps the assistant choose a capability, but it does not grant access or bypass the selected run mode.
4. Observe the run
Follow run events as AcornOps gathers evidence. The run retains its exact target and tool scope. If a required dependency becomes unavailable or authorization changes, the run fails closed rather than substituting another capability.5. Review the result
Check that the conclusion is supported by the gathered evidence. Keep the session for follow-up questions so the conversation and run history remain together. Open the run trace to review compact tool evidence. When an eligible tool produced a retained complete result, choose View full redacted result to load it separately. A completed assistant answer can also be exported with Generate PDF.When to use a Workflow
Use a Workflow when the procedure should be reusable, scheduled, triggered by an event, or coordinated across specialist Agents. Use target chat for direct investigation of one connected target. See Target chat and evidence for references, evidence retention, PDF export, and collaboration behavior.A write-capable follow-up uses a new effective scope and may pause for approval. Approval applies only to the exact requested tool and arguments.