11. Importing the Workflows into n8n¶

← Master Career Dossier · Home →

This guide walks you through importing Maestro's 31 workflows into your own n8n instance. It assumes you've already started n8n (see Installation, Step 7) and can open it at http://localhost:5678.

⏱️ Time: 20–30 minutes. Most of it is the import clicks; the wiring at the end is quick.

What you're importing¶

The workflows/ folder contains 31 workflows, organized into five subfolders by role:

workflows/
├── entry/        (4)  the pipelines you trigger — discovery, build, refine, cover letters
├── shared/       (5)  helpers every pipeline calls — Call LLM, Data Loader, etc.
├── agents/      (11)  the AI agents (Agent 1 … Agent 11)
├── sources/      (4)  job-board connectors — Greenhouse, Lever, Ashby, JSearch
└── recorders/    (7)  the workflows that save results to Google Sheets

These aren't independent — the entry workflows call the agents, sources, recorders, and shared helpers as sub-workflows. That dependency shapes the import order below.

Before you start¶

You need n8n running and reachable at http://localhost:5678. If you haven't set up the database, Google credentials, or API keys yet, do Installation Steps 3–9 first — the workflows will import without them, but they won't run until credentials are connected.

Step 1 — Import every workflow¶

n8n imports one file at a time. You'll repeat this for all 31 files. Import order matters because sub-workflows must exist before the workflows that reference them resolve cleanly. Import in this order:

shared/ (5 files)
recorders/ (7 files)
agents/ (11 files)
sources/ (4 files)
entry/ (4 files) — last, since these call everything above

💡 If you import out of order, it's not fatal — references resolve by workflow once everything is present. But importing helpers first means fewer "missing workflow" warnings along the way.

For each file:

In n8n, click Overview (or Workflows) in the left sidebar.
Click the Create Workflow dropdown (top-right) → Import from File…
Select the .json file from the workflows/ subfolder.
The workflow opens. Click Save (top-right).

Import from File

In n8n, click Create workflow ▾ → Import from File… and select the .json file.

Repeat until all 31 are imported. After this, your workflow list shows all 31 by name (Agent 1 – Resume Builder, Application Orchestrator, Call LLM, and so on).

Step 2 — Publish every workflow¶

Importing and saving is not enough. n8n only lets one workflow call another if the callee is Active (published). If you skip this, entry workflows fail at runtime with "could not find workflow" errors.

For each of the 31 workflows, open it and toggle it Active using the switch in the top-right.

Activate every workflow

After importing, toggle each workflow from Inactive to Active (top-right). All 31 must be Active — sub-workflows that aren't published are invisible to their callers.

📌 All 31 must be Active, not just the entry workflows. The agents, sources, recorders, and shared helpers are all called as sub-workflows, and a sub-workflow that isn't published is invisible to its caller.

A fast way to verify: in the workflow list, every row should show a green Active badge.

Step 3 — Reconnect the Google credential¶

The exported workflows reference a Google service-account credential by an internal ID that won't exist in your n8n. After import, the Google Sheets and Google Drive nodes will show the credential as missing or unset. You fix this once, then re-select it where needed.

First create the credential if you haven't: Credentials → Add Credential → Google Service Account API, and paste your service account's client_email and private_key. (See Installation Step 9.)
Open any workflow that has a Google Sheets or Drive node showing a red "credential" warning.
Click the node, open its Credential dropdown, and select your Google credential.
Save.

A node with a missing credential

A node showing a ⚠ on its Credential dropdown (e.g. a Google Sheets node) needs you to re-select your credential. Open the node and pick your Google service account.

💡 You don't have to touch every node by hand. n8n shares a credential across all nodes that use it — once you select your Google credential on one Sheets node, re-open the others and they'll usually offer it as the default. Work through any that still show a warning.

Which workflows have Google nodes? Mainly the recorders (they write to Sheets), the sources and Job Fetcher (some read config), Data Loader and Prompt Loader (read config/prompts), and the Application Recorder (writes to Drive). Check each for a credential warning.

Step 4 — Connect the AI provider credential¶

The agents call Anthropic / OpenAI / Gemini through HTTP nodes inside the Call LLM workflow, using a Header Auth credential.

Credentials → Add Credential → Header Auth.
For Anthropic: header name x-api-key, value = your sk-ant-… key.
Open Call LLM, find the Anthropic HTTP Request node, and select this credential.
If you use OpenAI or Gemini, repeat with their headers (Authorization: Bearer … / x-goog-api-key) on the respective nodes.

See Configuration → AI provider credentials for the exact header values.

Step 5 — Set the webhook secret on the entry workflows¶

The four trigger workflows are protected by a shared secret so only your dashboard and scheduler can fire them. Each entry workflow's Webhook node needs a Header Auth credential whose value matches the MAESTRO_WEBHOOK_SECRET in your dashboard's .env.local.

Credentials → Add Credential → Header Auth.
Header name: X-Maestro-Secret; value: the same string you set for MAESTRO_WEBHOOK_SECRET.
Open each of these and select the credential on the Webhook node:
Job Discovery (/webhook/run-discovery)
Application Orchestrator (/webhook/build-application)
Application Refinement (/webhook/refine-resume)
Cover Letter Generation & Refinement (/webhook/generate-cover-letter)

⚠️ If this secret doesn't match what the dashboard/scheduler send, triggering returns 503 and the scheduler silently skips. This is the most common "nothing happens when I click Build" cause.

Step 6 — Point the database at your Sheet¶

A few workflows carry a placeholder where a Google Sheet/Drive ID belongs. The most important is the Run Error Handler, which holds the database ID in a Set node (it can't read it from the run, by design).

Open Run Error Handler.
Find the Error Handler Config node (a Set node) holding database_id (value shows YOUR_GOOGLE_SHEETS_DATABASE_ID).
Replace it with your actual database Sheet ID (the long string from your Sheet's URL).
Save.

💡 Most other workflows read the database ID from config or from the data passed in at runtime, so they don't need editing. If any node shows a YOUR_… placeholder, replace it with your real ID the same way.

Step 7 — Refresh sub-workflow schemas (if needed)¶

If you later edit a sub-workflow's inputs, the workflows that call it cache the old schema. After importing, if an entry workflow seems to pass empty or wrong data to a sub-workflow, open the calling node (an Execute Workflow node) and click the refresh ⟳ icon to re-pull the sub-workflow's input schema.

You usually won't need this on a fresh import, but it's the fix if data isn't flowing between workflows.

Step 8 — Verify¶

A quick end-to-end check:

Open Job Discovery and click Execute Workflow (the manual-run button). It should complete without red error nodes. Check your Sheet's jobs tab — new rows should appear (depending on your watchlist_companies).
If discovery works, the shared helpers, sources, an agent (Agent 10 + Agent 8), and a recorder all ran — that exercises most of the system.

If a node errors, open the execution, read the red node's message, and cross-reference Troubleshooting. The usual first-import issues are: a workflow left Inactive, a Google credential not re-selected, or a missing AI provider key.

Quick checklist¶

[ ] All 31 workflows imported
[ ] All 31 set to Active
[ ] Google service-account credential created and selected on Sheets/Drive nodes
[ ] AI provider (Header Auth) credential selected in Call LLM
[ ] X-Maestro-Secret credential on the entry workflows' Webhook nodes
[ ] database_id placeholder replaced in Run Error Handler
[ ] Job Discovery manual run completes and writes to the jobs tab

Once all eight are checked, the pipeline is live and the dashboard can drive it.

← Master Career Dossier · Home →