About
This skill exports HubSpot workflows as versioned JSON files via the API, enabling them to be treated as code. It allows developers to track changes with diffs and fully restore workflows from the JSON backups, providing a critical safety net. Use it to backup, review, and recover automation configurations that are otherwise unprotected in the HubSpot UI.
Quick Install
Claude Code
Recommendednpx skills add TomGranot/hubspot-admin-skills -a claude-code/plugin add https://github.com/TomGranot/hubspot-admin-skillsgit clone https://github.com/TomGranot/hubspot-admin-skills.git ~/.claude/skills/workflows-as-codeCopy and paste this command in Claude Code to install this skill
Documentation
Workflows as Code
Export every workflow in the portal to JSON files, keep the exports under version control, and restore workflows from JSON when something breaks. HubSpot has no recycle bin for workflows — a deleted or mangled workflow is gone unless you have its definition. This skill gives you that safety net, plus reviewable diffs of what changed between exports.
Why This Matters
Workflows are production automation, but most portals treat them like disposable UI configuration: no backups, no change history, no review. One mis-click in the workflow editor can silently break lead routing or suppression for weeks. The v4 Automation API supports full read and create of workflow definitions, which makes a code-like lifecycle possible: export → version → diff → restore.
Prerequisites
- A HubSpot private app access token (
HUBSPOT_ACCESS_TOKENin.env) with theautomationscope (plus sensitive-data scopes if any flows touch sensitive properties) - Python 3.10+ with
uv - A place to keep exports — a git repository is ideal (
data/workflow-exports/is gitignored here by default because exports may reference internal names; move them to a private repo if you want history)
Scripts
| Stage | Script | Run with |
|---|---|---|
| Export | scripts/export.py | uv run skills/workflows-as-code/scripts/export.py |
| Restore | scripts/restore.py | uv run skills/workflows-as-code/scripts/restore.py <export-file.json> |
export.py lists all flows via GET /automation/v4/flows, fetches each full definition via the batch read endpoint (POST /automation/v4/flows/batch/read, falling back to per-flow GETs), and writes one JSON file per workflow plus a manifest. restore.py recreates a workflow from an export file via POST /automation/v4/flows — always disabled, under a "(restored)" name, for review before enabling.
Execution Pattern
Stage 1: Plan
- Decide export cadence: before any workflow cleanup (mandatory — see
/cleanup-workflows), after building new workflows, and on a schedule (monthly pairs well with/quarterly-database-cleanup). - Decide where exports live long-term (private git repo recommended).
Stage 2: Before
Nothing to prepare — the export is read-only. Note the current workflow count from Automation > Workflows for comparison.
Stage 3: Execute — Export
uv run skills/workflows-as-code/scripts/export.py
Output layout:
data/workflow-exports/<YYYY-MM-DD>/
├── manifest.csv # flowId, name, type, isEnabled, revisionId, file
├── flow-<flowId>.json # one full definition per workflow
└── ...
To diff two exports:
diff -u data/workflow-exports/2026-06-01/flow-12345.json \
data/workflow-exports/2026-07-01/flow-12345.json
(Timestamps and revision IDs will differ; meaningful changes show up in actions and enrollmentCriteria.)
Stage 3 (alternative): Execute — Restore
To recreate a deleted or broken workflow from an export:
uv run skills/workflows-as-code/scripts/restore.py data/workflow-exports/2026-06-01/flow-12345.json
The script strips server-assigned fields (id, revisionId, timestamps), renames the flow with a "(restored)" suffix to avoid name collisions, and creates it disabled. Review it in the UI, verify enrollment criteria and actions against the original, then rename and enable.
Stage 4: After
- Export: verify the manifest row count matches the portal's workflow count and spot-open one JSON file.
- Restore: open the restored workflow in Automation > Workflows, compare against the export, then enable.
Rollback
- Export is read-only — nothing to roll back.
- A restored workflow you do not want: it was created disabled; just delete it.
Technical Gotchas
- Updating in place requires the current
revisionId.PUT /automation/v4/{flowId}must include the flow's latestrevisionIdandtype— a stale revision is rejected. The restore script sidesteps this by creating a new flow instead of updating. - Some actions round-trip imperfectly. Actions whose field shapes are undocumented (e.g., copy-from-associated-object, notification recipients) export fine but may be rejected on re-create in a different portal. The restore script reports per-action errors; add rejected actions in the UI.
- v3 workflow IDs are not v4 flow IDs. If you have old tooling built on
/automation/v3/workflows, the v4 API provides workflowId → flowId mapping endpoints for migration. New tooling should use v4 exclusively — v3 is legacy. - Sensitive-data scopes. Flows that read or set sensitive properties require the corresponding sensitive-data scopes on your private app, or reads will fail.
- Exports may contain internal data. Workflow names, property names, list IDs, and notification text are all in the JSON. Treat exports like configuration secrets — keep them in a private repo.
GitHub Repository
Frequently asked questions
What is the workflows-as-code skill?
workflows-as-code is a Claude Skill by TomGranot. Skills package instructions and resources that Claude loads on demand, so Claude can perform workflows-as-code-related tasks without extra prompting.
How do I install workflows-as-code?
Use the install commands on this page: add workflows-as-code to Claude Code as a plugin, or clone its repository into your skills directory, then restart Claude so it picks up the skill.
What category does workflows-as-code belong to?
workflows-as-code is in the Meta category, tagged api and automation.
Is workflows-as-code free to use?
Yes. workflows-as-code is listed on AIMCP and free to install. It runs inside Claude, so no separate service account is required to use the skill itself.
Related Skills
This skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
This skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
This skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
