The workflow
The four-stage loop Penling runs — brief, focus areas, plan, build — and why each stage matters before the next begins.
Penling runs a four-stage loop for every initiative. Each stage produces a concrete artefact that the next stage builds from. Skipping a stage is possible — but it reintroduces the ambiguity the loop is designed to eliminate.
Stage 1: Brief
A brief is a short, plain-language description of the initiative. It answers three questions:
- What are we building? A description of the intended outcome.
- Why does it matter? The motivation — user pain, business need, or technical requirement.
- What's out of scope? At least one thing you are explicitly not doing.
The brief is the seed. Penling's AI reads it and can propose a set of focus areas if you ask it to. The quality of that proposal depends on the quality of the brief. Remember — vague in, vague out.
Stage 2: Focus areas
Focus areas decompose the brief into reviewable, testable slices of work. Each focus area has a four-part spec:
- Definition — what this slice is, in one or two sentences
- Results — the observable outcomes that must be true when it's done
- Boundaries — what is explicitly out of scope for this slice
- Conditions — what thing must remain true once the work has been done
Penling's AI can suggest an initial set of focus areas from the brief. You accept, edit, or rewrite them. The goal at this stage is to reach alignment — every stakeholder should be able to read the focus areas and agree on what "done" means.
Stage 3: Plan
A plan turns the set of focus areas into an ordered build sequence. Penling generates a plan that:
- Groups the work in a focus area into logical phases
- Surfaces dependencies between slices
- Identifies the critical path
- Raises questions on aspects of the focus that might need clarification
- Highlights risks that you can ignore or acknowledge
You can modify the generated plan in whatever way you see fit. Once your changes have been made, publish the plan as a versioned build.
The plan is the handoff artefact — it's what your coding agent receives when work begins.
Stage 4: Build
With a plan in place, the initiative moves into active development. In build mode, Penling tracks:
- Which focus actions (tasks) are in progress, blocked, or complete
- Clarifications raised by the team or agent
- Verifications (tests) carried out against the task acceptance criteria
- Commits & PR's
The build stage is where Penling connects to your coding environment via the MCP server — plans flow directly to Claude Code, Cursor, or any MCP-compatible agent.
The loop, not the line
The four stages aren't a waterfall. You'll often return to an earlier stage: a clarification during build reveals a gap in a focus area, which requires a plan update. Penling is designed for this — every artefact is editable, every stage is revisitable, and the history of changes is preserved.
There are very few 'required' pathways in Penling. You can choose how to use it in whichever suits your workflow best.