Pulse Publish API

Build with Pulse Publish.

Add video publishing to your app: upload once, set a caption per network, get a person's approval, and publish through every network's official API. One base URL, bearer auth, JSON both ways.

Base URL	`https://publish.debuginit.com`
Auth	`Authorization: Bearer <DebugInit ID access token>`
Format	JSON; uploads are the raw file
Machine-readable	/openapi.json (OpenAPI 3.1): import it into Postman, Insomnia or a code generator
AI agents	MCP server at `https://publish.debuginit.com/mcp`, see MCP

Built in-house by DebugInit. Pulse Publish is not a wrapper around another posting service. The DebugInit team wrote every network connector, the approval engine and this API in Rust, and it calls each network's official API directly.

Quick start API reference MCP for agents

Before you start

Get access.

Create a DebugInit ID
Sign up at id.mypulse.buzz and verify your email. It's the identity your integration calls with.
Request API access
Tell us what you're building through debuginit.com/contact, with your verified email, the networks you need, and your web origin if you'll call from a browser.
Get your access level
We enable your ID with read or write access and connect the network accounts you'll publish to. GET /v1/me shows what you have.

Quick start · 5 minutes

From a token to a live post.

The whole journey once, end to end. Pick your language in any code box; the rest of the page follows.

Set up

Get an access token for your DebugInit ID (see Authentication) and keep it in an environment variable.

export PP=https://publish.debuginit.com
export TOKEN=…   # your DebugInit ID access token

// Node 18+, or a browser on a registered origin
const PP = "https://publish.debuginit.com";
const TOKEN = process.env.TOKEN;   // DebugInit ID access token
const H = { Authorization: "Bearer " + TOKEN };

# pip install requests
import os, requests
PP = "https://publish.debuginit.com"
H = {"Authorization": "Bearer " + os.environ["TOKEN"]}

Check your access

Expect your account, and "write": true if you can create and approve.

curl "$PP/v1/me" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/me", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/me", headers=H)
print(r.status_code, r.json())

See the connected accounts

Only networks with "connected": true can be posted to. See Connected accounts.

curl "$PP/v1/networks" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/networks", headers=H)
print(r.status_code, r.json())

Upload the video and cover

curl -X POST "$PP/v1/media?name=edge-part2.mp4" \
  -H "Authorization: Bearer $TOKEN" \
  --data-binary @edge-part2.mp4

curl -X POST "$PP/v1/media?name=edge-part2-cover.png" \
  -H "Authorization: Bearer $TOKEN" \
  --data-binary @edge-part2-cover.png

import { readFile } from "node:fs/promises";

for (const name of ["edge-part2.mp4", "edge-part2-cover.png"]) {
  const res = await fetch(PP + "/v1/media?name=" + name, { method: "POST", headers: H, body: await readFile(name) });
  console.log(res.status, await res.json());   // 201 { name, bytes }
}

for name in ["edge-part2.mp4", "edge-part2-cover.png"]:
    with open(name, "rb") as f:
        r = requests.post(f"{PP}/v1/media", params={"name": name}, headers=H, data=f)
    print(r.status_code, r.json())   # 201 {'name': ..., 'bytes': ...}

Create the post

One default caption, overrides where a network needs its own, and a time.

curl -X POST "$PP/v1/posts" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
}'

const res = await fetch(PP + "/v1/posts", {
  method: "POST",
  headers: { ...H, "Content-Type": "application/json" },
  body: JSON.stringify({
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
}),
});
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts", headers=H, json={
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
})
print(r.status_code, r.json())

Preview it for a network

The exact text that will go, and its length as that network counts it.

curl "$PP/v1/posts/edge-part2/x" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts/edge-part2/x", headers=H)
print(r.status_code, r.json())

Approve it

One call per network. This is the human step: put it behind an Approve button in your app, pressed by the person accountable for the post.

curl -X POST "$PP/v1/posts/edge-part2/x/approve" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x/approve", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts/edge-part2/x/approve", headers=H)
print(r.status_code, r.json())

Publish now, or let the time come

Approved posts go out at at. To go now:

curl -X POST "$PP/v1/posts/edge-part2/x/publish" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x/publish", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts/edge-part2/x/publish", headers=H)
print(r.status_code, r.json())

Read the result

status becomes posted with the live url, or failed with the network's own error.

curl "$PP/v1/posts/edge-part2" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts/edge-part2", headers=H)
print(r.status_code, r.json())

That's the whole API. Everything else is detail: per-network settings, connected accounts and the reference below.

Authentication

Every /v1 call except /v1/health needs Authorization: Bearer <token>, where the token is a DebugInit ID access token for an account that has API access.

Issuer	`https://id.mypulse.buzz` (OpenID configuration)
Flow	OAuth 2.1 authorization code with PKCE (S256)
Scopes	`openid email profile`
Checked	On each call (cached for a minute): the token must be valid and its verified email must have access

Apps acting for a person sign them in with DebugInit ID and send their token. Server-to-server integrations use a token for a dedicated DebugInit ID created for that service.

Keep tokens out of URLs, logs and front-end bundles. In CI, store them as secrets. A 401 means the token expired or was revoked: get a fresh one.

Access levels.

Read lets you list networks, posts, previews, media and activity. Write also lets you upload, create, delete, approve and publish. Without write access every POST and DELETE answers 403 and nothing changes. GET /v1/me returns "write". In the reference, calls that need it are marked write access.

Concepts

Posts and targets.

A post is one video with a title, a time and captions. Each network it goes to is a target: post × network. Targets are approved, published and reported one by one, so LinkedIn can be live while X still waits for approval.

Posts you create have "source": "api" and can be deleted until something is posted. Posts the workspace manages itself show "source": "config"; you can read and approve them, not delete them.

Approval#

Approving stores a fingerprint of the title, caption, settings, video and cover. Before anything is sent, the fingerprint is checked again. If it differs, approval reads changed and the target is refused until a person approves it again.

no→ approve →ok→ edit anything →changed→ approve again →ok

Lifecycle#

waiting→ time comes →due→posting→postedorfailed

A failed target keeps its error. Fix the cause (a caption that's too long, an expired sign-in) and publish it again.

Scheduling#

Approved targets go out at their at time; the scheduler checks every minute. GET /v1/me shows "schedule": "on" while scheduled publishing runs. publish sends immediately whatever the time.

Guides

Recipes for real posts.

Post to a LinkedIn Page, for an audience#

Target linkedin_page and say which Page in settings: an id, a link, a vanity name or the Page's name. A Showcase Page works the same way. An audience narrows who sees it in their feed; LinkedIn needs more than 300 matching members.

curl -X POST "$PP/v1/posts" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
  "id": "edge-page", "at": "2026-11-21T09:30:00+05:30", "title": "Why edge systems",
  "video": "edge-part2.mp4", "targets": ["linkedin_page"],
  "caption": "Why edge systems keep working when the cloud doesn't.",
  "settings": { "linkedin_page": { "page": "143896897",
    "audience": { "locations": ["India"], "seniorities": ["director", "vp", "cxo"] } } }
}'

const res = await fetch(PP + "/v1/posts", {
  method: "POST",
  headers: { ...H, "Content-Type": "application/json" },
  body: JSON.stringify({
  "id": "edge-page", "at": "2026-11-21T09:30:00+05:30", "title": "Why edge systems",
  "video": "edge-part2.mp4", "targets": ["linkedin_page"],
  "caption": "Why edge systems keep working when the cloud doesn't.",
  "settings": { "linkedin_page": { "page": "143896897",
    "audience": { "locations": ["India"], "seniorities": ["director", "vp", "cxo"] } } }
}),
});
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts", headers=H, json={
  "id": "edge-page", "at": "2026-11-21T09:30:00+05:30", "title": "Why edge systems",
  "video": "edge-part2.mp4", "targets": ["linkedin_page"],
  "caption": "Why edge systems keep working when the cloud doesn't.",
  "settings": { "linkedin_page": { "page": "143896897",
    "audience": { "locations": ["India"], "seniorities": ["director", "vp", "cxo"] } } }
})
print(r.status_code, r.json())

Audience keys: locations, seniorities (unpaid, training, entry, senior, manager, director, vp, cxo, partner, owner), industries, job_functions.

One caption per network#

caption is the default; captions overrides it for any network. Check with the preview call: on X a link always counts as 23 characters and an emoji as 2, so the raw length isn't what X sees.

curl "$PP/v1/posts/edge-part2/x" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts/edge-part2/x", headers=H)
print(r.status_code, r.json())

Pinterest Pins#

Pins have their own title (up to 100), a destination link and alt text (up to 500). The board is the one chosen when Pinterest was connected.

Response

"targets": ["pinterest"],
"captions": { "pinterest": "Six reasons edge systems matter, with real examples. Part 2 of 3." },
"settings": { "pinterest": {
  "title": "Why Edge Systems Matter: 6 Real Examples (Part 2 of 3)",
  "link": "https://wayworks.space",
  "alt_text": "Vertical explainer video on six reasons edge systems matter"
} }

Connected accounts#

Pulse Publish holds the sign-ins to each network, so your app never stores a LinkedIn or X token. GET /v1/networks lists them with connected, the account name and expires_in_days.

To connect an account, or reconnect one before it expires, call signin.
Send the account's owner to the returned open URL. They approve on the network's own screen; the network returns to Pulse Publish, which shows "Connected".
Check GET /v1/networks again. A network not yet enabled for your access answers 409: ask us to enable it.

curl -X POST "$PP/v1/networks/linkedin/signin" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks/linkedin/signin", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/networks/linkedin/signin", headers=H)
print(r.status_code, r.json())

Call from a browser#

Browsers may call the API from the web origins registered for your integration (CORS); tell us yours when you request access. Use the signed-in person's own DebugInit ID token.

JavaScript

// inside the Pulse app: token = the signed-in person's DebugInit ID access token
const res = await fetch("https://publish.debuginit.com/v1/posts", {
  headers: { Authorization: "Bearer " + token },
});
if (res.status === 401) return signInAgain();
const { posts } = await res.json();

Automate from CI#

Render the video in CI, upload it, queue it, and leave approval to a person. Keep the token in the CI's secrets.

GitHub Actions

# .github/workflows/queue-video.yml
on: { workflow_dispatch: {} }
jobs:
  queue:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - run: |
          curl -fsS -X POST "$PP/v1/media?name=weekly.mp4" -H "Authorization: Bearer $TOKEN" --data-binary @out/weekly.mp4
          curl -fsS -X POST "$PP/v1/posts" -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
            -d @post.json   # approval stays with a person
        env:
          PP: https://publish.debuginit.com
          TOKEN: ${{ secrets.PULSE_PUBLISH_TOKEN }}

MCP · Model Context Protocol

Let an AI agent do the busywork.

Pulse Publish is also an MCP server. Claude, Cursor, VS Code or any MCP client can check accounts, draft captions within each network's limits, upload and queue posts, and read results, then stop and ask a person before anything is approved or published.

Your AI clientClaude · Cursor · VS Code

Pulse Publishsame rules as the REST API

11 networksLinkedIn, X, YouTube…

Endpoint	`https://publish.debuginit.com/mcp`
Transport	Streamable HTTP, JSON responses, stateless
Protocol	`2026-07-28` and `2025-11-25` · `2025-06-18` · `2025-03-26`
Auth	`Authorization: Bearer <DebugInit ID access token>`
Offers	14 tools · 4 resources and 2 templates · 3 prompts

See an agent at work#

A real sequence of tool calls. The agent does the legwork and hands the decision back to you.

AI client · connected to pulse-publish

YouQueue the edge-systems video for LinkedIn and X, tomorrow 9:30, captions in our voice.

whoamiwrite access · schedule on
list_networkslinkedin ✓ · x ✓ · pinterest ✓
list_mediaedge-part2.mp4 · edge-part2-cover.png
create_postedge-part2 queued for Tue 09:30 · approval: no
preview_post · x214 of 280 characters (links count as 23) · fits
preview_post · linkedin1,180 of 3,000 characters · fits

Approval needed. approve_post publishes in your name, so the agent stops and shows you both captions.

approve_post · xapproval: ok
approve_post · linkedinapproval: ok

AgentApproved. Both go out Tue 09:30; I'll check get_post for the live links after.

AgentNothing was approved. The post waits in your queue until you say so.

The human stays in charge. approve_post and publish_now are marked as write tools, and the server's instructions tell agents never to call them without your go-ahead. Without write access they refuse anyway.

Connect your client#

Pick your client, paste, and set your token. Then ask it something like "what's waiting for approval in Pulse Publish?"

# Claude Code
claude mcp add --transport http pulse-publish https://publish.debuginit.com/mcp \
  --header "Authorization: Bearer $TOKEN"

// .cursor/mcp.json
{
  "mcpServers": {
    "pulse-publish": {
      "url": "https://publish.debuginit.com/mcp",
      "headers": { "Authorization": "Bearer ${env:PULSE_PUBLISH_TOKEN}" }
    }
  }
}

// .vscode/mcp.json
{
  "inputs": [{ "type": "promptString", "id": "pp-token", "description": "Pulse Publish token", "password": true }],
  "servers": {
    "pulse-publish": {
      "type": "http",
      "url": "https://publish.debuginit.com/mcp",
      "headers": { "Authorization": "Bearer ${input:pp-token}" }
    }
  }
}

// Claude Desktop and other stdio-only clients: claude_desktop_config.json
{
  "mcpServers": {
    "pulse-publish": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://publish.debuginit.com/mcp", "--header", "Authorization:Bearer ${PULSE_PUBLISH_TOKEN}"],
      "env": { "PULSE_PUBLISH_TOKEN": "…" }
    }
  }
}

Tool explorer#

Every tool runs the same code as its REST call. Pick one to see its schema, hints and an example call.

Who am I `whoami`

Who the token belongs to, whether writing is on (approve/publish/create need it), and whether the schedule runs.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "whoami"
  }
}

{
  "additionalProperties": false,
  "properties": {},
  "type": "object"
}

List networks `list_networks`

Every network: connected or not, the account, keys set, days until the sign-in ends, and limits.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "list_networks"
  }
}

{
  "additionalProperties": false,
  "properties": {},
  "type": "object"
}

List the queue `list_posts`

Every post with, per network, approval (no · ok · changed), status (waiting · due · posting · posted · failed), and the live url or the error.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "list_posts"
  }
}

{
  "additionalProperties": false,
  "properties": {},
  "type": "object"
}

Get a post `get_post`

One post with its per-network approval, status, url and error.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2"
    },
    "name": "get_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    }
  },
  "required": [
    "id"
  ],
  "type": "object"
}

Preview a post for one network `preview_post`

Exactly what would be sent to that network: title, caption, its length as the network counts it, settings, whether it fits the limits, and the approval state.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2",
      "network": "x"
    },
    "name": "preview_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    },
    "network": {
      "description": "network id",
      "enum": [
        "linkedin",
        "linkedin_page",
        "facebook",
        "instagram",
        "threads",
        "bluesky",
        "mastodon",
        "pinterest",
        "x",
        "youtube",
        "tiktok"
      ],
      "type": "string"
    }
  },
  "required": [
    "id",
    "network"
  ],
  "type": "object"
}

List media `list_media`

Videos and covers uploaded to the server (names to use in create_post).

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "list_media"
  }
}

{
  "additionalProperties": false,
  "properties": {},
  "type": "object"
}

Recent activity `recent_activity`

The latest log lines: who approved what, what was posted, errors.

read-only

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "recent_activity"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "lines": {
      "default": 50,
      "maximum": 500,
      "minimum": 1,
      "type": "integer"
    }
  },
  "type": "object"
}

Upload a video or cover `upload_media`

Upload a file (base64) of up to 25 MB. Larger files: use the REST upload (POST /v1/media). The name can't already exist.

changes datareaches networks

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "data_base64": "iVBORw0KGgo…",
      "name": "edge-part2-cover.png"
    },
    "name": "upload_media"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "data_base64": {
      "contentEncoding": "base64",
      "type": "string"
    },
    "name": {
      "description": "file name ending in .mp4 .mov .png .jpg or .jpeg",
      "pattern": "^[A-Za-z0-9][A-Za-z0-9._-]{0,119}$",
      "type": "string"
    }
  },
  "required": [
    "name",
    "data_base64"
  ],
  "type": "object"
}

Create a post `create_post`

Queue a video. It goes nowhere until a person approves each network.

changes datareaches networks

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "at": "2026-11-20T09:30:00+05:30",
      "caption": "Why edge systems keep working when the cloud doesn't.",
      "id": "edge-part2",
      "targets": [
        "linkedin",
        "x"
      ],
      "title": "Why edge systems",
      "video": "edge-part2.mp4"
    },
    "name": "create_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "at": {
      "description": "RFC 3339 with offset, e.g. 2026-11-20T09:30:00+05:30",
      "format": "date-time",
      "type": "string"
    },
    "caption": {
      "description": "default text for every network",
      "type": "string"
    },
    "captions": {
      "additionalProperties": {
        "type": "string"
      },
      "description": "per-network text, keyed by network id",
      "type": "object"
    },
    "cover": {
      "description": "name from list_media",
      "type": "string"
    },
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    },
    "settings": {
      "additionalProperties": {
        "type": "object"
      },
      "description": "per-network options, e.g. linkedin_page.page / audience, pinterest.title / link / alt_text",
      "type": "object"
    },
    "targets": {
      "items": {
        "description": "network id",
        "enum": [
          "linkedin",
          "linkedin_page",
          "facebook",
          "instagram",
          "threads",
          "bluesky",
          "mastodon",
          "pinterest",
          "x",
          "youtube",
          "tiktok"
        ],
        "type": "string"
      },
      "minItems": 1,
      "type": "array"
    },
    "title": {
      "type": "string"
    },
    "video": {
      "description": "name from list_media",
      "type": "string"
    }
  },
  "required": [
    "id",
    "at",
    "title",
    "video",
    "targets",
    "caption"
  ],
  "type": "object"
}

Delete a post `delete_post`

Delete a post created through the API, only while nothing has been posted.

changes datadestructiveidempotentreaches networks

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2"
    },
    "name": "delete_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    }
  },
  "required": [
    "id"
  ],
  "type": "object"
}

Approve a post for a network `approve_post`

Approve exactly the current content for one network. Only when the user has clearly approved it.

changes dataidempotentreaches networksneeds your go-ahead

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2",
      "network": "x"
    },
    "name": "approve_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    },
    "network": {
      "description": "network id",
      "enum": [
        "linkedin",
        "linkedin_page",
        "facebook",
        "instagram",
        "threads",
        "bluesky",
        "mastodon",
        "pinterest",
        "x",
        "youtube",
        "tiktok"
      ],
      "type": "string"
    }
  },
  "required": [
    "id",
    "network"
  ],
  "type": "object"
}

Withdraw approval `unapprove_post`

Stop a post on one network before it goes out.

changes dataidempotentreaches networks

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2",
      "network": "x"
    },
    "name": "unapprove_post"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    },
    "network": {
      "description": "network id",
      "enum": [
        "linkedin",
        "linkedin_page",
        "facebook",
        "instagram",
        "threads",
        "bluesky",
        "mastodon",
        "pinterest",
        "x",
        "youtube",
        "tiktok"
      ],
      "type": "string"
    }
  },
  "required": [
    "id",
    "network"
  ],
  "type": "object"
}

Publish now `publish_now`

Publish an approved post to one network right away instead of at its time. Public and can't be undone from here: only on the user's explicit go-ahead.

changes datadestructivereaches networksneeds your go-ahead

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "id": "edge-part2",
      "network": "x"
    },
    "name": "publish_now"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "id": {
      "description": "post id",
      "pattern": "^[a-z0-9-]{1,64}$",
      "type": "string"
    },
    "network": {
      "description": "network id",
      "enum": [
        "linkedin",
        "linkedin_page",
        "facebook",
        "instagram",
        "threads",
        "bluesky",
        "mastodon",
        "pinterest",
        "x",
        "youtube",
        "tiktok"
      ],
      "type": "string"
    }
  },
  "required": [
    "id",
    "network"
  ],
  "type": "object"
}

Start a network sign-in `start_network_signin`

Returns the URL a person opens to connect a network (the network's own consent screen).

changes datareaches networks

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "network": "x"
    },
    "name": "start_network_signin"
  }
}

{
  "additionalProperties": false,
  "properties": {
    "network": {
      "description": "network id",
      "enum": [
        "linkedin",
        "linkedin_page",
        "facebook",
        "instagram",
        "threads",
        "bluesky",
        "mastodon",
        "pinterest",
        "x",
        "youtube",
        "tiktok"
      ],
      "type": "string"
    }
  },
  "required": [
    "network"
  ],
  "type": "object"
}

Resources and prompts#

Resource	What
`pulse-publish://guide`	The workflow and the approval rule, written for an AI client
`pulse-publish://posts` · `…/posts/{id}` · `…/posts/{id}/{network}`	The queue, one post, one network's preview
`pulse-publish://networks` · `pulse-publish://media`	Connected accounts and limits · uploaded files

Prompt	What
`draft_captions`	One caption per network for a topic, inside each network's limits
`review_queue`	What's waiting for approval, what failed and why, what goes out next
`queue_video`	Upload, create and preview a post, then stop for your approval

Protocol details#

2026-07-28 clients send _meta["io.modelcontextprotocol/protocolVersion"] and clientCapabilities on every request, plus the MCP-Protocol-Version, Mcp-Method and (for tools/call, resources/read, prompts/get) Mcp-Name headers. A mismatch answers 400 with -32020. server/discover describes the server.
Older clients call initialize; the server answers with the version they asked for when it supports it, otherwise 2025-11-25. Notifications get 202. No session id is issued.
Unsupported versions: 400 with -32022 and the supported list. Unknown methods: -32601. One JSON-RPC message per POST; GET and DELETE answer 405.
401s carry WWW-Authenticate pointing to the protected-resource metadata (RFC 9728). Browser origins must be registered; others get 403.

curl

# a raw 2026-07-28 call, for testing
curl "$PP/mcp" -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" \
  -H "MCP-Protocol-Version: 2026-07-28" -H "Mcp-Method: tools/call" -H "Mcp-Name: list_posts" \
  -d '{ "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "list_posts", "arguments": {},
        "_meta": { "io.modelcontextprotocol/protocolVersion": "2026-07-28", "io.modelcontextprotocol/clientCapabilities": {} } } }'

API reference

Every call.

17 calls. Examples assume PP, TOKEN and H from the set-up step. Network ids: linkedin, linkedin_page, facebook, instagram, threads, bluesky, mastodon, pinterest, x, youtube, tiktok.

GET/v1/healthpublic

Health#

Public, no token. Use it for uptime checks.

curl "$PP/v1/health" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/health", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/health", headers=H)
print(r.status_code, r.json())

Response

{ "ok": true, "service": "pulse-publish", "version": "0.1.0", "config_error": false }

GET/v1/meread

Who am I#

Who the token belongs to, whether you have write access, and whether scheduled publishing runs.

curl "$PP/v1/me" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/me", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/me", headers=H)
print(r.status_code, r.json())

Response

{ "caller": "you@yourcompany.com", "root": false, "write": true, "owner": "Pulse", "schedule": "on" }

GET/v1/networksread

List networks#

Every network: connected or not, the account, whether its keys are set, the sign-in type and redirect URL, days until the sign-in ends, and its limits.

curl "$PP/v1/networks" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/networks", headers=H)
print(r.status_code, r.json())

Response

{ "networks": [ { "id": "linkedin", "name": "LinkedIn", "connected": true, "account": "Sanjay shah",
    "keys_set": true, "sign_in": "oauth", "redirect_url": "https://publish.debuginit.com/callback/linkedin",
    "expires_in_days": 57, "expired": false, "limits": { "caption_max_chars": 3000, … } }, … ] }

POST/v1/networks/{network}/signinwrite access

Start a sign-in#

Returns the network's consent URL. Open it in a browser; the network sends the browser back to /callback/{network}, which shows "Connected".

curl -X POST "$PP/v1/networks/linkedin/signin" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks/linkedin/signin", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/networks/linkedin/signin", headers=H)
print(r.status_code, r.json())

Response

{ "open": "https://www.linkedin.com/oauth/v2/authorization?…", "redirect_url": "https://publish.debuginit.com/callback/linkedin" }

POST/v1/networks/{network}/connectwrite access

Connect without a browser#

For Bluesky and Mastodon accounts set up for your access: they sign in with a handle and app password, no browser step.

curl -X POST "$PP/v1/networks/bluesky/connect" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks/bluesky/connect", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/networks/bluesky/connect", headers=H)
print(r.status_code, r.json())

Response

{ "connected": "bluesky", "account": "debuginit.bsky.social" }

DELETE/v1/networks/{network}write access

Forget a sign-in#

Removes the stored sign-in for one network; posting to it stops until it is connected again.

curl -X DELETE "$PP/v1/networks/x" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/networks/x", { method: "DELETE", headers: H });
console.log(res.status, await res.json());

r = requests.delete(f"{PP}/v1/networks/x", headers=H)
print(r.status_code, r.json())

Response

{ "disconnected": "x" }

GET/v1/mediaread

List media#

Videos and covers uploaded to this server.

curl "$PP/v1/media" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/media", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/media", headers=H)
print(r.status_code, r.json())

Response

{ "media": [ { "name": "edge-part2.mp4", "bytes": 22295769 } ] }

POST/v1/media?name={file}write access

Upload a video or cover#

The raw file is the body. Names use letters, digits and . - _ and end in .mp4, .mov, .png, .jpg or .jpeg. A name can't be reused (409); files can be up to 500 MB (413).

curl -X POST "$PP/v1/media?name=edge-part2.mp4" \
  -H "Authorization: Bearer $TOKEN" \
  --data-binary @edge-part2.mp4

curl -X POST "$PP/v1/media?name=edge-part2-cover.png" \
  -H "Authorization: Bearer $TOKEN" \
  --data-binary @edge-part2-cover.png

import { readFile } from "node:fs/promises";

for (const name of ["edge-part2.mp4", "edge-part2-cover.png"]) {
  const res = await fetch(PP + "/v1/media?name=" + name, { method: "POST", headers: H, body: await readFile(name) });
  console.log(res.status, await res.json());   // 201 { name, bytes }
}

for name in ["edge-part2.mp4", "edge-part2-cover.png"]:
    with open(name, "rb") as f:
        r = requests.post(f"{PP}/v1/media", params={"name": name}, headers=H, data=f)
    print(r.status_code, r.json())   # 201 {'name': ..., 'bytes': ...}

Response

{ "name": "edge-part2.mp4", "bytes": 22295769 }

GET/v1/postsread

List the queue#

Every post with, per network, its approval (no · ok · changed), status (waiting · due · posting · posted · failed), and the live url or the error.

curl "$PP/v1/posts" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts", headers=H)
print(r.status_code, r.json())

Response

{ "posts": [ { "id": "edge-part2", "at": "2026-11-20T04:00:00Z", "title": "Why edge systems", "source": "api",
    "video": "edge-part2.mp4", "cover": "edge-part2-cover.png",
    "targets": [ { "network": "x", "approval": "ok", "status": "posted", "url": "https://x.com/…/status/…",
                   "approved_at": "…", "posted_at": "…", "error": null } ] } ] }

POST/v1/postswrite access

Create a post#

Fields: id (a–z, 0–9 and -, unique) · at (RFC 3339 time with its offset) · title · video and cover (names from /v1/media) · targets (network ids) · caption (the default text) · captions (per-network text) · settings (per-network options, see the guides).

curl -X POST "$PP/v1/posts" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
}'

const res = await fetch(PP + "/v1/posts", {
  method: "POST",
  headers: { ...H, "Content-Type": "application/json" },
  body: JSON.stringify({
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
}),
});
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts", headers=H, json={
  "id": "edge-part2",
  "at": "2026-11-20T09:30:00+05:30",
  "title": "Why edge systems",
  "video": "edge-part2.mp4",
  "cover": "edge-part2-cover.png",
  "targets": ["linkedin", "x", "pinterest"],
  "caption": "Part 2 · Why edge systems keep working when the cloud doesn't.",
  "captions": { "x": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing" },
  "settings": { "pinterest": { "title": "Why Edge Systems Matter (Part 2 of 3)", "link": "https://wayworks.space" } }
})
print(r.status_code, r.json())

Response

201 — the new post, in the same shape as GET /v1/posts/{id}

GET/v1/posts/{id}read

Get one post#

One post, same shape as in the list. Poll this after publishing.

curl "$PP/v1/posts/edge-part2" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts/edge-part2", headers=H)
print(r.status_code, r.json())

Response

{ "id": "edge-part2", "targets": [ … ] }

DELETE/v1/posts/{id}write access

Delete a post#

Only posts created through the API, and only until something has been posted.

curl -X DELETE "$PP/v1/posts/edge-part2" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2", { method: "DELETE", headers: H });
console.log(res.status, await res.json());

r = requests.delete(f"{PP}/v1/posts/edge-part2", headers=H)
print(r.status_code, r.json())

Response

{ "deleted": "edge-part2" }

GET/v1/posts/{id}/{network}read

Preview for one network#

Exactly what would be sent: title, caption, its length as that network counts it, settings, the limits check, and the fingerprint an approval would cover.

curl "$PP/v1/posts/edge-part2/x" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/posts/edge-part2/x", headers=H)
print(r.status_code, r.json())

Response

{ "network": "x", "title": "Why edge systems", "caption": "Part 2 · Why edge systems. Part 1: https://lnkd.in/… #EdgeComputing",
  "caption_length": 74, "caption_max": 280, "fits": true, "settings": {}, "fingerprint": "9f2c…" }

POST/v1/posts/{id}/{network}/approvewrite access

Approve#

Approves exactly the current content for one network. If anything changes later, approval becomes "changed" and nothing goes out.

curl -X POST "$PP/v1/posts/edge-part2/x/approve" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x/approve", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts/edge-part2/x/approve", headers=H)
print(r.status_code, r.json())

Response

{ "id": "edge-part2", "targets": [ { "network": "x", "approval": "ok", "status": "waiting" } ] }

POST/v1/posts/{id}/{network}/unapprovewrite access

Withdraw approval#

Stops a post that hasn't gone out yet.

curl -X POST "$PP/v1/posts/edge-part2/x/unapprove" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x/unapprove", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts/edge-part2/x/unapprove", headers=H)
print(r.status_code, r.json())

Response

{ "id": "edge-part2", "targets": [ { "network": "x", "approval": "no" } ] }

POST/v1/posts/{id}/{network}/publishwrite access

Publish now#

Publishes now instead of at its time. It must be approved. Answers 202 at once; follow the result with GET /v1/posts/{id}.

curl -X POST "$PP/v1/posts/edge-part2/x/publish" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/posts/edge-part2/x/publish", { method: "POST", headers: H });
console.log(res.status, await res.json());

r = requests.post(f"{PP}/v1/posts/edge-part2/x/publish", headers=H)
print(r.status_code, r.json())

Response

202 { "publishing": "edge-part2/x", "check": "/v1/posts/edge-part2" }

GET/v1/activity?lines=50read

Activity log#

The latest log lines: who approved what, what was posted, errors. Up to 500 lines.

curl "$PP/v1/activity?lines=50" \
  -H "Authorization: Bearer $TOKEN"

const res = await fetch(PP + "/v1/activity?lines=50", { headers: H });
console.log(await res.json());

r = requests.get(f"{PP}/v1/activity?lines=50", headers=H)
print(r.status_code, r.json())

Response

{ "lines": [ "2026-11-20 09:30:02 x: posted edge-part2 → https://x.com/…", … ] }

Errors and status codes.

Errors are JSON with one plain sentence; when a network refused something, its own message is included.

Response

409 { "error": "edge-part2/x changed since it was approved — approve again" }
502 { "error": "x: HTTP 403 Forbidden: {\"detail\":\"You are not permitted to perform this action.\"}" }

Code	Meaning	What to do
200 · 201 · 202	Done · created · started	For 202, poll `GET /v1/posts/{id}`
400	Something in the request is wrong	The message names the field
401	No token, or it expired or was revoked	Get a fresh DebugInit ID token
403	Your access is read-only, or your DebugInit ID isn't enabled	Request access
404	No such post, network or file	Check the id
409	Not possible now: not approved, changed since approval, already posted, name taken, network not enabled	Approve again, rename, or ask us
413	Upload over 500 MB	Compress the file
502	DebugInit ID couldn't be reached	Retry with backoff

Network limits.

The limits the API checks before anything is sent. The preview call checks them for you.

Id	Network	Caption	Title	Video	Notes
`linkedin`	LinkedIn	3000	200	30 min	custom cover
`linkedin_page`	LinkedIn Page	3000	200	30 min	custom cover
`facebook`	Facebook Page	63206	255	90 s	custom cover
`instagram`	Instagram	2200	—	15 min	custom cover
`threads`	Threads	500	—	5 min	public video URL
`bluesky`	Bluesky	300	—	3 min	—
`mastodon`	Mastodon	500	—	—	custom cover, links count as 23
`pinterest`	Pinterest	800	100	15 min	custom cover
`x`	X	280	—	20 min	links count as 23, emoji count as 2
`youtube`	YouTube Shorts	5000	100	3 min	custom cover
`tiktok`	TikTok	2200	—	10 min	—

Before you ship

Integration checklist.

What a solid integration handles. Ticks are remembered in this browser.

Tokens refresh. On a 401, get a fresh DebugInit ID token and retry once.
Preview before approval. Show the person the preview_post text and limits_ok for each network.
A real person approves. Approval sits behind a button, never in an automatic path.
"changed" is handled. After an edit, ask for approval again instead of retrying the publish.
Results are polled. After a 202, poll GET /v1/posts/{id} until posted or failed; store the live url.
Failures show the network's words. Surface error to the person; it says what to fix.
Ids are yours. Use stable post ids so a retried create returns "already exists" instead of a duplicate.
Expiring accounts. Watch expires_in_days in /v1/networks and prompt a reconnect early.
Tokens stay secret. Not in URLs, logs or front-end bundles.

Build with Pulse Publish.

Get access.

Create a DebugInit ID

Request API access

Get your access level

From a token to a live post.

Set up

Check your access

See the connected accounts

Upload the video and cover

Create the post

Preview it for a network

Approve it

Publish now, or let the time come

Read the result

Authentication

Access levels.

Posts and targets.

Approval#

Lifecycle#

Scheduling#

Recipes for real posts.

Post to a LinkedIn Page, for an audience#

One caption per network#

Pinterest Pins#

Connected accounts#

Call from a browser#

Automate from CI#

Let an AI agent do the busywork.

See an agent at work#

Connect your client#

Tool explorer#

Who am I whoami

List networks list_networks

List the queue list_posts

Get a post get_post

Preview a post for one network preview_post

List media list_media

Recent activity recent_activity

Upload a video or cover upload_media

Create a post create_post

Delete a post delete_post

Approve a post for a network approve_post

Withdraw approval unapprove_post

Publish now publish_now

Start a network sign-in start_network_signin

Resources and prompts#

Protocol details#

Every call.

Health#

Who am I#

List networks#

Start a sign-in#

Connect without a browser#

Forget a sign-in#

List media#

Upload a video or cover#

List the queue#

Create a post#

Get one post#

Delete a post#

Preview for one network#

Approve#

Withdraw approval#

Publish now#

Activity log#

Errors and status codes.

Network limits.

Integration checklist.

Who am I `whoami`

List networks `list_networks`

List the queue `list_posts`

Get a post `get_post`

Preview a post for one network `preview_post`

List media `list_media`

Recent activity `recent_activity`

Upload a video or cover `upload_media`

Create a post `create_post`

Delete a post `delete_post`

Approve a post for a network `approve_post`

Withdraw approval `unapprove_post`

Publish now `publish_now`

Start a network sign-in `start_network_signin`