Universal

Task

Placeholder headline for testing

6 steps 2 checkpoints Multi-agent Human approval Self-correcting

Workflow steps

The exact path the AI follows, one step at a time. Numbered stations move the work forward; checkpoints pause it for a human.

1
Intake
Confirm the task has a clear title and a one-line summary.
If either is missing, pause and ask the operator. Do not
advance until both are present.
2
Analyze
planner Human approval
Write a short plan — the minimum needed to know what to build
and what could go wrong. One screen of text, not a document.

Cover:
- What we're doing and why
- Files likely to change
- Obvious risks
- How we'll verify it worked
Ask the operator to approve the plan before moving on.
3
Implement
developer
Execute the plan from the previous step. Keep changes
surgical — don't scope-creep into adjacent code.

When finished, move to verify.
4
Verify
reviewer Human approval
Run the verification you defined in the plan. Report what
you ran, what passed, what didn't. If something failed,
go back to implement and fix it before asking for approval.

Ask the operator to approve before moving on.

If rejected → returns to implement
5
Capture Learnings
If anything non-obvious came up, write a short note to:
```
docs/learnings/{YYYY-MM-DD}-{task-slug}.md
```
Use plain Markdown. Skip if nothing non-obvious happened.
6
Complete
Task done. Close it out.

ConvOps

Stop re-explaining your process. Pick a template, start a task, and let the AI drive the work through your workflow.