Track progress with Claude Code
This is what a day looks like when you drive Plenar from Claude Code — written as the sequence of natural-language requests you’d actually make. Every write opens a New Plan preview, so you see the impact before anything commits. The examples follow a task through the sample Atlas SaaS project so you can follow along.
Start your day: what should I work on?
Open Claude Code and ask:
Pick up where you left off, or start the next prioritized task.
Start a task
When you’re ready to begin:
Plenar claims the task for your session — your team sees a “Claimed by” indicator in the task detail panel showing who’s on it. Say yes, and you have a branch ready:
git checkout -b plenar/atla-implement-data-validation-layerThe branch convention (plenar/<task-id>-<slug>) links your code to the plan automatically.
During development: update as you learn
Estimate changed
You start and realize the task is more involved than expected:
You see the effect before it happens. If the goal would go Off Track, you’d know immediately and could cut scope instead.
Hit a blocker
The plan reflects the blocker, and you have options for what to do next.
Link a pull request
The PR is now tracked against the task for traceability.
Finish a task
After applying, you know exactly what’s next — the plan already accounts for your completion.
Add new scope
You discover you need more work:
You see the cost of the new scope before committing.
Create a full plan from scratch
Starting fresh? Describe the whole structure in one request:
The agent returns a full preview with dates, assignments, and goal ETAs.
Check schedule health
At any point:
The agent reports overall health, each goal’s ETA and buffer, and any warnings or pending changes.
Estimate with data
Not sure how long something will take? Ask for comparable past work:
The full loop at a glance
Ask what’s next, start a task, update it as you learn, complete it — then the next task is already waiting. Every write opens a New Plan preview — you always see the impact before committing. The plan stays accurate because keeping it current is part of your normal flow, not a separate chore.
Common questions
Do I have to confirm every change? Yes — each write returns a New Plan preview and applies only when you say so. Nothing changes the Plan of Record silently.
How does my team know I’m on a task? Starting a task claims it for your session; a “Claimed by” indicator shows in the task detail panel.
What if I don’t know the estimate? Ask the agent for exemplars (plenar_get_exemplars) — it surfaces similar completed tasks with their actual durations so you can anchor to history.
Does completing a task move other work? It recomputes the plan with your actual duration, which can let dependent tasks start earlier — shown in the preview before you apply.
What to do next
- Connect Claude Code — wire up the integration if you haven’t.
- Handle changes — playbooks for the situations above.
- Manage your schedule — the edit → preview → apply loop in depth.
- How the plan learns — how these updates sharpen future estimates.