---
title: "Agents that close on evidence"
description: "Why done must mean checked, not claimed: a task closes only when the effect is seen to have landed."
url: "https://docs.octo.ad/why/agents-that-close-on-evidence/"
id: "why/agents-that-close-on-evidence"
type: "concept"
audience: "team lead"
dateModified: "2026-11-06"
productArea: "agents at work"
keywords: ["closure on evidence", "grounding"]
---

# Agents that close on evidence

> Why done must mean checked, not claimed: a task closes only when the effect is seen to have landed.

## The report is not the proof

Separate the report from the proof. In OCTOPUS a task carries a definition of done from the start, and it closes only on evidence that the effect landed: the change merged, the record was written, the page serves the new version. The agent's own report does not close it. The same rule runs across the platform as a grounding guard: verify the effect before reporting success.

## Task closure, by category

- **What the category typically offers:** A task is complete when the agent returns or the step succeeds.
- **What OCTOPUS does:** A task closes only on evidence that the effect landed (the change merged, the record was written, the page serves the new version), never on the agent's report alone.
- **Question to ask any vendor:** Show a task the agent reported done that the platform refused to close.

## Grounding guards, by category

- **What the category typically offers:** Prompting conventions, left to the developer.
- **What OCTOPUS does:** Two practices enforced as engineering: look up before you make up, and verify the effect before reporting success.
- **Question to ask any vendor:** Where in the system, not the prompt, is 'verify before reporting success' enforced?