pipefy-process-design — Claude Code skill: what it does
Advises on Pipefy workflow architecture, phases, and fields before you build, activating only for open-ended process design requests.
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a claude-codeWhat pipefy-process-design does
This skill acts as an architecture consultant when a workflow needs to be mapped from scratch. If you already supply phases, fields, or a detailed specification, the agent skips this step and routes directly to build tools.
Before proposing a layout, the agent queries visible organization workflows with search_pipes to avoid duplicate setups. It structures stages around object states rather than actions, keeps start forms minimal, and plans for edge cases with blocked states. Fixed-condition transitions get marked for automation instead of manual routing.
Output comes as a structured blueprint listing stages, field types, and a single-line impact note. The agent does not fabricate metrics. If volume or lead times are missing, it asks for them directly.
What it helps with
- architect Pipefy workflows before generating phases and cards
- inspect existing organization pipes with
search_pipesto prevent duplicate flows - determine minimal start form fields and key automation triggers
- select between linear approvals, multi-stage pipelines, and hub-and-spoke designs
Example prompts for the agent
- Help me design an expense approval process in Pipefy
- What's the best structure for an IT ticketing flow?
- How do I organize an onboarding pipeline across teams?
How to install pipefy-process-design
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a claude-code2Manually: put the files in place
~/.claude/skills/pipefy-process-design/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/pipefy-process-design/
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a codex2Manually: put the files in place
.agents/skills/pipefy-process-design/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a cursor2Manually: put the files in place
.cursor/skills/pipefy-process-design/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install pipefy/ai-toolkit pipefy-process-design2Manually: put the files in place
.github/skills/pipefy-process-design/SKILL.md1Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a gemini-cli2Platform CLI
gemini skills install https://github.com/pipefy/ai-toolkit --path plugins/pipefy/skills/pipefy-process-designManually: put the files in place
.devin/skills/pipefy-process-design/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about pipefy-process-design
How do I install pipefy-process-design in Claude Code?
Run npx skills add pipefy/ai-toolkit --skill pipefy-process-design -a claude-code.
When should this skill be used instead of building directly?
Use it when asking for architectural advice or consulting on workflow structure. When you supply an exact specification or ask to build a pipe directly, the agent skips this skill.
What does the final design output include?
It outputs phase names, start form fields with types, key automations, related processes, and a single-line impact summary.
Source text
SKILL.md · Apache-2.0
---
name: pipefy-process-design
description: >
Use this skill ONLY when the user explicitly asks for help designing
or architecting a process ("help me design", "what's the best structure",
"how do I organize this flow?"). For direct execution requests
("create a pipe for X", "build a reimbursement process", or any message
with a detailed spec), skip this skill and execute directly using the
domain skills (pipes-and-cards, automations, etc.).
tags: [pipefy, process-design, orchestration, pipes, consulting]
---
# Pipefy Process Design
Read the [MCP reference](references/mcp.md) or [CLI reference](references/cli.md) for the surface you are using. Load only the relevant reference.
This skill activates when the user wants **consulting help** to design a process — not when they want you to build one. If the user already knows what they want (gave you phases, fields, a spec, or a clear use case), **do not use this skill**. Execute directly.
---
## When to use vs when to execute
| Signal | Response |
|--------|----------|
| "Help me design a process for X" | **Use this skill** — consulting mode. |
| "What's the best structure for Y?" | **Use this skill** — architecture guidance. |
| "Create a pipe for X" | **Skip this skill** — route via `pipefy-building`, then the domain skill. |
| User provides phases, fields, or a spec | **Skip this skill** — the spec IS the plan. Route via `pipefy-building`, then build. |
---
## Prerequisites
- Understand the user's industry and use case before advising structure.
- Always search for existing pipes in the org before recommending a new one (`search_pipes`).
---
## Discovery phase
1. **Research existing org structure:**
Operation: `search_pipes pipe_name=""` (empty search returns all visible pipes)
Operation: `get_organization organization_id=<id>`
2. **Understand the process intent:**
- What triggers a new case? (form submission, email, manual)
- Who are the actors? (submitter, approver, ops team)
- What are the key decision points? (approve/reject, escalate, auto-close)
- What data needs to be tracked? (fields per phase)
- Are there related processes that should be connected? (use relations)
3. **Identify the right Pipefy components:**
| Need | Component |
|------|-----------|
| Workflow stages | Pipe + Phases |
| Structured data per stage | Phase fields |
| Reference/lookup data | Database table |
| Cross-process linkage | Pipe relation |
| Automatic actions | Automation rule |
| AI-driven processing | AI automation or AI agent |
---
## Design principles
- **Start with the outcome.** What does "done" look like for this process?
- **Name phases for states, not actions.** "Under Review" not "Review It".
- **Keep the start form minimal.** Only ask for data the requester can provide on day 1.
- **Use required fields sparingly.** Every required field is a blocker.
- **Design for the exception.** Add a "Blocked" or "On Hold" phase for edge cases.
- **Automate the obvious.** If a transition always happens under the same condition, it should be an automation, not a manual step.
---
## Common patterns
### Linear approval flow
`Submission → Under Review → Approved / Rejected → Done`
- Start form: requester fills details.
- "Under Review" phase: approver field + due date.
- Automation: notify approver on card creation.
### Multi-stage pipeline
`Intake → Triage → In Progress → Testing → Deployed`
- Each phase has role-specific fields.
- Automations advance cards on condition.
- SLA fields track time-in-phase.
### Hub-and-spoke (related processes)
- Central "intake" pipe connected via pipe relation to domain-specific pipes.
- Child cards created automatically via automation when parent moves to "Escalated".
---
## Output format
After the design consultation, produce a concise summary:
```
Process: <Name>
Phases: [list]
Start form fields: [list with types]
Key automations: [list]
Related processes: [list or "none"]
Impact: [one line: time this design returns to the team against a fully manual run of the same flow; if a phase exists only for human triage, what an automation or an AI agent would change there. Do not start a pipe diagnosis.]
Next step: [execute with pipes-and-cards skill? or more questions?]
```
Keep Impact to one line. Do not invent volume, hourly cost, or lead time; name every missing number and ask for all of them in one question. For a fuller justification, read `pipefy-process-impact`.
---
## Success criteria
- User has clarity on phases, fields, and automation triggers before execution starts.
- No duplicate pipes created (checked via `search_pipes`).
- The design is buildable with available operations (no features promised that don't exist).
## See also
- `pipefy-building` — for execution / build asks, read the router then the domain skill (do not expand this consulting skill into a build playbook).
- `pipefy-pipes-and-cards` — execute the design once finalized.
- `pipefy-automations` — add automation rules to the new pipe.
- `pipefy-process-intelligence` — analyze an existing process for improvement (distinct from designing new).
- `pipefy-process-impact` to justify a material change in full; this skill only emits the Impact line.
© pipefy · Apache-2.0 · pipefy/ai-toolkit