Task planning & requirements

When you describe a change, the agent doesn’t just start coding. It plans — breaking your request into discrete tasks, sequencing them, and tracking each one to completion. You see the entire plan unfold in the Tasks panel.

How planning works

1. You describe what you want in the chat panel.
2. The agent analyzes your request and creates a task plan — a structured list of steps needed to deliver the result.
3. Each task appears in the Tasks panel with a status indicator.
4. As work progresses, tasks move through states: pending, in-progress, verifying, and complete.

The Tasks panel

The Tasks panel gives you a live view of the agent’s work:

• Task list — every step the agent plans, with its current status.
• Requirement links — tasks are mapped to the requirements they fulfill.
• Tool calls — see exactly which tools the agent used for each task (edit-file, apply-variant, fetch-url, etc.).
• Verification notes — the agent documents how it verified each task.
• Cancel — stop a specific task if the agent is heading in the wrong direction.

Requirements capture

Before work begins, you can define requirements — the criteria that determine whether the change is successful. Requirements appear in the Requirements panel and serve as the agent’s checklist:

• Add requirements in plain language: “The CTA button must be green”, “Mobile layout should stack vertically”.
• Auto-detection — the agent can infer requirements from your initial request and surface them for your approval.
• Traceability — each task links back to the requirement it addresses, and test results map to requirements so you can see coverage.

Why this matters

Without planning, AI tools often produce a single monolithic change that’s hard to review or undo. With Coframe Agent‘s task planning:

• Transparency — you see what the agent intends to do before it does it.
• Control — cancel or redirect individual tasks without losing other work.
• Accountability — every change traces back to a requirement you set.
• Parallelism — the agent can delegate independent tasks to sub-agents for faster delivery.

Tip

For complex changes, start by defining 2–3 requirements. The agent will plan around them and you can track exactly which requirement each task addresses.