ps-build
Skill1 starsChanged 5 months agoMITView on GitHub
Build a new Process Street workflow from a description, document, or structured plan. Use when the user wants to create a workflow/template/process in Process Street.
- Reads credentials
In this repository12 other files it gives its agents.
- data-setSkill
- demo-buildSkill
- chrome-recording-to-workflowSkill
- logic-rulesSkill
- ps-auditSkill
- ps-editSkill
- ps-organizeSkill
- workflow-designSkill
- integration-scopingSkill
- ps-searchSkill
- ps-best-practicesSkill
- ps-helpSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 47
Use $task:Step Name and $widget:Step Name:Field Label symbolic refs.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Build a Process Street workflow
Turn a description (or pasted process / document) into a live Process Street workflow via the public API.
CLI: python3 "${CLAUDE_PLUGIN_ROOT}/scripts/process_street.py" <command> (when installed as a plugin) or python3 scripts/process_street.py from the repo. Needs PROCESS_STREET_API_KEY set. Capability + field-type detail: reference/api-reference.md. Design guidance: reference/best-practices.md.
Use the CLI even if the Process Street MCP is connected.
build-workfloworchestrates the whole multi-step build (tasks, widgets, logic, rules) in one call and records a rollback target — raw MCP calls don't. Don't substitute MCP create-ops for the build flow.
1. Understand the process
Get: the workflow name and outcome, the steps in order (who does what), decision points (branches), approvals/sign-offs, data to capture (and at which step), SLAs, stop gates, and anything conditional. If the user pasted a document or transcript, extract this from it. Ask one clarifying question only if something essential is missing.
2. Draft a build plan (JSON)
Apply best practices (reference/best-practices.md): outcome-based name, 6–15 verb-named steps, the right field type per datum (Select for fixed choices, Date, Number, Members, Email — not all free text), conditional logic to hide irrelevant steps, stops/approvals only where they earn the friction, dynamic due dates, an owner per task.
Build-plan shape (see api-reference.md for all 17 field types, due-date/assignment rules, content widgets):
{
"name": "Customer Onboarding",
"folderId": "<folderId>",
"shareLevel": "Run",
"tasks": [
{"name": "Intake", "taskType": "Standard", "dueOffset": "P1D", "widgets": [
{"fieldType": "Text", "label": "Company Name"},
{"fieldType": "Select", "label": "Plan Tier", "options": ["Starter", "Pro", "Enterprise"]},
{"fieldType": "Number", "label": "Seats", "constraints": {"minValue": 1, "decimalPlaces": 0}}
]},
{"name": "Manager Approval", "taskType": "Approval", "stop": true,
"assignmentRules": [{"email": "manager@company.com"}],
"dueDateRule": {"sourceType": "PreviousTaskCompletedDate", "offsetDirection": "After", "dueOffset": "P1D", "businessDaysHandlingMode": "BusinessDaysOnly"}}
],
"logicRules": [
{"action": "Show",
"condition": {"or": [{"and": [{"widgetId": "$widget:Intake:Plan Tier", "operator": "Is", "value": "Enterprise"}]}]},
"targets": {"taskIds": ["$task:Enterprise Setup"]}}
]
}Use $task:Step Name and $widget:Step Name:Field Label symbolic refs. Logic operators: Is, IsNot, HasAnyValue, HasNoValue, IsGreaterThan, IsLessThan.
You need a folderId. List folders: process_street.py folders. Create one: process_street.py folder-create --name "...".
3. Confirm, then build
Show the user a plain-English summary (steps, types, fields, logic, what's manual). Get a yes. Then:
echo '<plan-json>' | python3 "${CLAUDE_PLUGIN_ROOT}/scripts/process_street.py" build-workflow [--draft-only]Use --draft-only if the plan has AI or Code tasks (they can't publish until configured). Then configure them and publish:
- AI:
task-config <wf> <rev> <taskId> --type Ai --prompt "..." - Code:
task-config <wf> <rev> <taskId> --type Code --code 'outputData["x"]=inputData["y"]' --input-mapping '{"y":"<fieldId>"}' --output-mapping '{"x":"<fieldId>"}' - Approval (so it's completable in a run):
task-config <wf> <rev> <approvalTaskId> --type Approval --subject-task-ids <taskId> - Then:
publish-revision <wf> <rev>
4. Polish (recommended)
Make it look professional and on-brand with add-media <wf> <rev> <taskId> (create a draft first if already published):
--logo-domain <client-domain>(company logo),--text '## Welcome…'on step 1--stock "keyword,keyword"topic image,--video <youtube-url>,--embed <url>(a relevant live resource) Make assets relevant to the use case. Brand colors/cover/fonts are UI-only — mention that if asked.
5. Verify
process_street.py template-export <workflowId> — confirm steps, fields, logic. Give the user the workflow URL: https://app.process.st/workflows/<workflowId>.
Manual (UI-only) items to flag
Required-field flag, brand theming (colors/cover/fonts), org/SSO settings, deep native-integration config. File fields are created as MultiFile.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

