> ## Documentation Index
> Fetch the complete documentation index at: https://docs.postsiva.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Postsiva MCP in n8n
> Connect n8n to Postsiva via MCP AI Agent or direct REST HTTP Request nodes.
n8n can drive Postsiva in three ways:
1. **Community nodes** — `n8n-nodes-postsiva` package with 43 dedicated nodes (recommended for most workflows)
2. **MCP AI Agent** — natural-language tool selection (best for flexible AI workflows)
3. **REST HTTP Request** — deterministic calls to specific endpoints (best for custom or uncovered endpoints)
All use the same workspace API key and Pro plan requirements.
***
## Prerequisites
* n8n instance (cloud or self-hosted) with MCP or HTTP Request nodes
* **Pro plan** with `mcp_enabled` (MCP) or `api_keys_enabled` (REST)
* Workspace API key: `psk_live_YOUR_KEY`
* Social accounts connected in the target workspace
***
## Option A — Community nodes (recommended)
Install **`n8n-nodes-postsiva`** from **Settings → Community nodes** for 43 ready-made nodes: posting, scheduling, drafts, inbox, AI generation, AI Watcher, and brand persona.
See the full guide: [n8n integration](/integrations/n8n).
***
## Option B — MCP AI Agent
Use when the workflow should **decide** which Postsiva action to take based on incoming data or chat input.
### MCP server config
| Field | Value |
| --------------- | ------------------------------ |
| **URL** | `https://mcp.postsiva.com/mcp` |
| **Transport** | Streamable HTTP |
| **Auth header** | `X-API-Key: psk_live_YOUR_KEY` |
In n8n's **AI Agent** node (or MCP Client tool):
1. Add MCP server **Postsiva** with the URL and header above
2. Connect the agent to your trigger (Webhook, Schedule, Slack, etc.)
3. System prompt example:
```
You manage social media via Postsiva MCP tools.
Always confirm before publishing live posts.
Use get_accounts first if platform connection is unclear.
Timestamps for scheduling must be ISO 8601 UTC.
```
### Example workflow — Webhook → Agent → Slack
```mermaid theme={null}
flowchart LR
WH[Webhook trigger]
AG[AI Agent + Postsiva MCP]
SL[Slack notify]
WH --> AG --> SL
```
**Webhook body:**
```json theme={null}
{
"instruction": "Draft a LinkedIn post about our n8n integration and save as draft"
}
```
**Agent tools used:** `idea_to_content` → `publish` with `draft: true`
### Example workflow — Scheduled content
1. **Schedule Trigger** — every Monday 08:00
2. **AI Agent** — "Generate 3 LinkedIn post ideas from our blog RSS themes and save top one as draft"
3. **HTTP Request** (optional) — notify team in Slack with draft link
***
## Option C — REST HTTP Request
Use for **fixed** pipelines where you know the exact endpoint.
### Shared credentials
Create n8n **Header Auth** credential:
| Header name | Value |
| ----------- | ------------------- |
| `X-API-Key` | `psk_live_YOUR_KEY` |
Base URL: `https://backend.postsiva.com`
### Pattern 1 — Generate and publish
**Node 1: HTTP Request — generate copy**
```
POST /unified/content/generate?workspace_id={{ $env.WORKSPACE_ID }}
Body (JSON):
{
"idea": "{{ $json.topic }}",
"platforms": ["linkedin"]
}
```
**Node 2: HTTP Request — publish**
```
POST /unified/post/text
Body (JSON):
{
"platforms": ["linkedin"],
"default_text": "{{ $json.data.linkedin.content }}"
}
```
### Pattern 2 — Sync analytics daily
```
GET /unified/analytics
Headers: X-API-Key
```
Store `$json` in Google Sheets, Airtable, or your data warehouse.
### Pattern 3 — Comment auto-reply (deterministic)
1. `GET /unified/comments?platforms=instagram&limit=5`
2. **IF** node — filter unanswered
3. `POST /unified/comments/reply?platform=instagram&comment_id={{ $json.id }}&text=Thanks for your comment!`
For AI-generated replies, use the **MCP AI Agent** with `get_comments` + `reply_to_comment` instead of hard-coded text.
***
## Community nodes vs MCP vs REST
| | Community nodes | MCP AI Agent | HTTP Request |
| ----------- | ----------------------------------------- | ------------------------------- | ------------------------ |
| Setup | Install `n8n-nodes-postsiva` + credential | MCP server + agent prompt | Per-endpoint nodes |
| Coverage | 43 Postsiva actions | Full MCP tool catalog | Any REST path |
| Flexibility | Medium — pick node + fields | High — agent picks tools | Low — fixed paths |
| Best for | Scheduled posts, inbox, watcher | Chatbots, variable instructions | Custom / debug endpoints |
See [n8n integration](/integrations/n8n) for community node workflows.
***
## Environment variables
Store secrets in n8n credentials or env vars:
```
POSTSIVA_API_KEY=psk_live_...
POSTSIVA_WORKSPACE_ID=aaa01a56-e769-44a6-87bf-cf6dfb9e7a27
POSTSIVA_BASE_URL=https://backend.postsiva.com
```
Reference in expressions: `{{ $env.POSTSIVA_API_KEY }}`
***
## Error handling
| Error | Action |
| --------------------- | -------------------------------------------- |
| `401` | Rotate or fix API key credential |
| Plan / `402` message | Upgrade workspace to Pro |
| Tool `success: false` | Branch on `$json.success`; log `$json.error` |
| Rate limits | Add Wait node; reduce schedule frequency |
Enable **Continue On Fail** on HTTP nodes only when downstream logic handles partial success (e.g. multi-platform publish).
***
## Example: RSS → AI → schedule
Trigger on new blog post URL
Postsiva node with `idea` = post title + summary, `platforms` = linkedin, threads
Postsiva node with caption from AI output and `scheduled_time` = next business day 09:00 UTC
Send confirmation email or Slack message with scheduled time
Same flow works with HTTP Request nodes — see [REST AI content](/apis/ai-content) and [n8n HTTP fallback](/integrations/n8n#http-request-fallback).
***
## Security
* Store `psk_live_` keys in n8n **Credentials**, not in workflow JSON exports
* Use scoped keys (`linkedin_posts` read-only for analytics-only workflows)
* Restrict webhook URLs with authentication
## Next
Install n8n-nodes-postsiva
MCP in Cursor
Generate endpoints
Scopes and headers