Persona — AI News Aggregator

Persona is an n8n workflow that delivers a personalized news digest by email. Users interact via a chatbot (register, login, manage preferences) and receive a daily newsletter curated by Claude AI from NewsAPI, Hacker News, and The Guardian.

Architecture

Two workflows work together:

• Persona DB — handles all database operations (auth, sessions, user CRUD). Imported first.
• Persona Main (persona.json) — the chat interface + scheduled newsletter dispatcher. Calls Persona DB as a sub-workflow.

Data is stored in three n8n Data Tables (not external databases).

---

Setup

Step 0 — Start n8n

Option A — Docker (recommended)

docker run -it --rm \
  -p 5678:5678 \
  -v ~/.n8n:/home/node/.n8n \
  n8nio/n8n

Option B — npm

npm install -g n8n
n8n start

n8n is then available at http://localhost:5678. Create your account on first launch.

---

Prerequisites

• A running n8n instance (self-hosted or cloud)
• API keys for:
  • Anthropic (Claude Haiku)
  • NewsAPI
  • The Guardian
• An SMTP account configured in n8n credentials
• Python 3 (for the ID injection script)

---

Step 1 — Create the Data Tables

In n8n, go to Data Tables and create three tables by importing the CSV files in this repo:

Table name	CSV file	Key columns
persona_users	persona_users.csv	username, passwordHash, email, interests, scheduleTime, scheduleTimezone, isActive, gdprConsent, sendNow, ...
persona_sessions	persona_sessions.csv	token, username, expiresAt
persona_login_attempts	persona_login_attempts.csv	attemptKey, count, resetAt

The CSV files contain a _schema_init placeholder row that defines the columns. You can delete it after import.

After creation, note the ID of each table — visible in the URL when you open it: /data-table/<ID>

---

Step 2 — Import Persona DB.json

1. In n8n go to Workflows → Import from file.
2. Import Persona DB.json.
3. Activate the workflow.
4. Note its workflow ID — visible in the URL: /workflow/<ID>.

---

Step 3 — Inject IDs with the Python script

The persona.json workflow file contains four placeholders that must be replaced with your real IDs before import:

Placeholder	Replaced by
{{DT_USERS}}	Data Table ID for persona_users
{{DT_SESSIONS}}	Data Table ID for persona_sessions
{{DT_LOGIN_ATTEMPTS}}	Data Table ID for persona_login_attempts
{{WORKFLOW_DB}}	Workflow ID of Persona DB

Run the script:

python replace_ids.py \
    --dt-users        <DT_USERS_ID> \
    --dt-sessions     <DT_SESSIONS_ID> \
    --dt-login-attempts <DT_LOGIN_ATTEMPTS_ID> \
    --workflow-db     <PERSONA_DB_WORKFLOW_ID> \
    persona.json "Persona DB.json"

The script edits the files in place. Use --output-dir to write to a separate directory instead:

python replace_ids.py \
    --dt-users        oU88XZmCfb87Oxbn \
    --dt-sessions     gLuygoEITlAypAmJ \
    --dt-login-attempts 4jjAebbyEijIfBx3 \
    --workflow-db     coEXen1CzEdEG3hH \
    --output-dir      ./patched \
    persona.json "Persona DB.json"

The script will warn about any remaining unfilled placeholders.

---

Step 4 — Import persona.json

1. In n8n go to Workflows → Import from file.
2. Import the (now patched) persona.json.

---

Step 5 — Configure credentials

In the imported workflows, link your credentials for each node that requires them:

Node	Credential type
Anthropic Chat Model, Message a model	Anthropic API key
Send Newsletter Email	SMTP account
Fetch NewsAPI	Insert your key directly in the URL (apiKey=YOUR_NEWSAPI_KEY)
Fetch Guardian	Insert your key directly in the URL (api-key=YOUR_GUADRIAN_API_KEY)

---

Step 6 — Set your n8n base URL

In persona.json, the Compose HTML Email node contains a hardcoded base URL used to generate the unsubscribe link:

const base = 'YOUR_N8N_BASE_URL';

Edit that node and replace YOUR_N8N_BASE_URL with your actual n8n public URL (e.g. https://n8n.example.com).

---

Step 7 — Activate both workflows

Activate Persona DB first, then Persona Main. The newsletter scheduler runs every 30 minutes and sends emails to users whose scheduled delivery time falls within the current ±15-minute window.

---

replace_ids.py reference

usage: replace_ids.py [-h] [--dt-users ID] [--dt-sessions ID]
                      [--dt-login-attempts ID] [--workflow-db ID]
                      [--output-dir DIR]
                      file [file ...]

options:
  --dt-users ID           Users datatable ID
  --dt-sessions ID        Sessions datatable ID
  --dt-login-attempts ID  Login attempts datatable ID
  --workflow-db ID        Persona DB workflow ID
  --output-dir DIR        Write results here instead of editing in place

You can pass one or more JSON files and only the flags you need — unspecified placeholders are left unchanged.

---

How the newsletter flow works

Schedule (every 30 min)
  → Get Active Users (Data Table)
  → For Each User
      → Check Send Conditions (timezone window + no recent send)
      → Fetch NewsAPI / HackerNews / Guardian (parallel)
      → Merge + deduplicate articles
      → Claude AI selects top 5 relevant articles
      → Compose HTML email
      → Send via SMTP
      → Update lastNewsletter timestamp

How the chat flow works

Chat Trigger (n8n chatbot UI or API)
  → Persona AI Agent (Claude Haiku)
      Tools: check_session / register_user / login_user /
             get_profile / update_preferences /
             delete_my_account / send_newsletter_now
  → All tools call Persona DB sub-workflow