Claw Post

Agent-native social posting from logged-in desktop Chrome. X, LinkedIn, Facebook member groups you already belong to, TikTok, Instagram, plus Reddit comments. No platform credentials shared.

Social posting infrastructure for AI agents. A paired browser extension publishes from the user's real browser session — no API keys to the social platforms needed.

Why Claw Post

• Zero credential exposure for social platforms. No OAuth tokens or social-platform passwords are sent to the agent or to Claw Post for posting. You use one Claw Post API key for API calls only; social sessions stay in the paired Chrome browser.
• Desktop Chrome, not a headless VPS. The agent can live on a server; the extension posts from the Chrome profile the human already uses. No chrome.debugger, no Playwright hunting the Post button.
• Multi-platform, one API. X, LinkedIn, Facebook (feed and groups), TikTok, and Instagram — all through a single POST /v1/jobs/tweet endpoint.
• Facebook member groups. Graph publish_to_groups is gone; Buffer/Zapier cannot publish there. Post only to groups the human already belongs to. Never auto-join, never discover→join→post.
• Native MCP support. The public MCP endpoint (https://mcp.clawpost.net/mcp) uses Streamable HTTP with OAuth 2.0. Claude, ChatGPT, Hermes, and other agents built on the official MCP SDKs (TypeScript, Python, and compatible clients) can connect — no Claw Post API key required.
• Simple setup. Sign up → install extension → click extension icon and sign in with Google (auto-pairs) → provide one API key (or connect via MCP OAuth). No per-platform credentials.
• Low cost per post. Free tier included; paid plans scale affordably. See clawpost.net for pricing.

Capabilities

Capability	Endpoint	Platforms
Post to feed or group	POST /v1/jobs/tweet	X, LinkedIn, Facebook, TikTok, Instagram
Comment / reply on a thread	POST /v1/reddit/comment	Reddit
Upload media for posts (small files)	POST /v1/media/upload	X, LinkedIn, Facebook, TikTok, Instagram
Upload large media (signed URL)	POST /v1/media/upload-url	X, LinkedIn, Facebook, TikTok, Instagram
Search for groups	POST /v1/groups/search	Facebook
Join a group	POST /v1/facebook/groups/join	Facebook
Check group membership	POST /v1/facebook/groups/status	Facebook
Poll any job result	GET /v1/jobs/:id	All

All endpoints require the clawpost-api-key header.

MCP agents: If your agent supports the Model Context Protocol, point it at https://mcp.clawpost.net/mcp and complete OAuth instead of using this REST API. Tools: list_platforms, create_post, create_reddit_comment, get_upload_url, get_post_status, and get_account_status. Use an MCP client that supports Streamable HTTP (current Anthropic / OpenAI SDK paths and the reference Python client).

Prerequisites (instruct the human)

Before any action works, the human must:

1. Sign up at clawpost.net
2. Install the Claw Post Chrome extension from the Chrome Web Store
3. Pair the extension by clicking the extension icon and signing in with Google (primary flow; auto-pairs to your account)
4. Set the CLAWPOST_API_KEY environment variable in the agent (copy the key from the Dashboard — it is not embedded in this skill file)
5. Log in to the target platform once in their browser (x.com, linkedin.com, facebook.com, tiktok.com, instagram.com, or reddit.com)

Extension scope. The Chrome extension is designed to activate on supported social sites and when communicating with the Claw Post API. Verify permissions on the Chrome Web Store listing and review the Terms of Service before installing.

If the agent gets EXTENSION_NOT_PAIRED or not_logged_in, direct the human to complete these steps. If automatic pairing fails, use the Dashboard 6-digit code flow as a fallback.

Authentication

All requests use the clawpost-api-key header:

text

clawpost-api-key: YOUR_CLAWPOST_API_KEY

The value above is a placeholder. Set the real secret via the CLAWPOST_API_KEY environment variable (or your agent’s secret store); do not paste live keys into this file or commit them to version control.

API Base URL

text

https://api.clawpost.net

This is the stable, official Claw Post API endpoint. Full documentation: https://clawpost.net/api-docs

---

1. Posting

Create a post job, then poll for completion.

Create post

text

POST https://api.clawpost.net/v1/jobs/tweet
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

X (Twitter):

json

{ "text": "Hello world!", "platform": "x" }

LinkedIn:

json

{ "text": "Hello LinkedIn!", "platform": "linkedin" }

Facebook feed:

json

{ "text": "Hello Facebook!", "platform": "facebook" }

Facebook group (use groupUrl from search results, or a known groupId):

json

{
  "text": "Hello group!",
  "platform": "facebook",
  "platformPayload": { "groupUrl": "https://www.facebook.com/groups/123456789/" }
}

TikTok (video + caption):

json

{ "text": "Caption text", "platform": "tiktok", "mediaPaths": ["<url from upload>"] }

Instagram (image or video + caption):

json

{ "text": "Caption text", "platform": "instagram", "mediaPaths": ["<url from upload>"] }

Instagram requires at least one media file (image or video). Text-only posts are not supported.

Optional fields:

• platform – "x" (default), "linkedin", "facebook", "tiktok", or "instagram"
• mediaPaths – array of media URLs (upload first via media endpoint below)
• idempotencyKey – unique string to prevent duplicate posts
• platformPayload – Facebook group targeting: { "groupId": "..." } or { "groupUrl": "..." }

Poll job status

text

GET https://api.clawpost.net/v1/jobs/:id

Status progresses: queued → processing → succeeded | failed.

• On success: response includes postUrl — the live URL of the published post. Share this with the user to confirm the post is live.
• On failure: response includes error and errorCode.
• Timing: posts execute through a real browser session. Allow 10–60 seconds before the job completes (longer for video uploads). Poll every 5–10 seconds.

---

1b. Reddit comments (browser actions)

Reply to a specific Reddit post or comment through the user's logged-in Chrome. No Reddit API key. The agent must supply the target thread/comment URL — Claw Post does not find threads for you.

text

POST https://api.clawpost.net/v1/reddit/comment
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

json

{
  "text": "Useful reply text",
  "targetUrl": "https://www.reddit.com/r/example/comments/abc123/title/"
}

• targetUrl must be a Reddit URL whose path includes /comments/ (post or comment permalink).
• Optional: idempotencyKey.
• Poll GET /v1/jobs/:id the same way as posts. On success, postUrl is the live comment permalink.
• Prerequisites: extension paired, Chrome open, user logged into Reddit in that browser.

---

2. Media upload

Upload media before posting. The returned url goes into the mediaPaths array. Media files are automatically deleted after 7 days — you do not need to manage cleanup.

Small files (images, short clips — under ~30 MB)

Send the file body directly to the API:

text

POST https://api.clawpost.net/v1/media/upload
clawpost-api-key: YOUR_CLAWPOST_API_KEY
Content-Type: multipart/form-data
Body: file=<media file>

Response:

json

{ "url": "https://storage.googleapis.com/..." }

Large files (videos and any file over ~30 MB) — two-step

For large files, get a signed upload URL first, then PUT the file directly to storage. This bypasses all proxy limits and works for files of any size.

Step 1 — request an upload URL:

text

POST https://api.clawpost.net/v1/media/upload-url
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

json

{ "filename": "video.mp4" }

contentType is optional — the API infers it automatically from the filename extension. Pass platform to validate the file type before uploading (e.g. "platform": "x" rejects audio files immediately).

Response:

json

{
  "uploadUrl": "https://storage.googleapis.com/...?X-Goog-Signature=...",
  "url": "https://storage.googleapis.com/...",
  "resolvedContentType": "video/mp4"
}

Step 2 — PUT the file directly to uploadUrl using the resolved content type (no clawpost-api-key header — the URL is self-authorised):

text

PUT <uploadUrl>
Content-Type: <resolvedContentType from Step 1>
Body: <raw file bytes>

The signed URL will reject PUT requests with incorrect Content-Type headers (403). Always use the resolvedContentType value returned in Step 1.

Expect a 200 or 204 response from Google Storage. The upload window is 1 hour.

Step 3 — use the url in the job exactly as with small files:

json

{ "text": "Check this out!", "platform": "tiktok", "mediaPaths": ["<url from step 1>"] }

The url is valid for 24 hours — sufficient for any normal job queue time.

---

3. Facebook Group Search

Out of scope for the recommended workflow. Post only to a groupUrl the human gave.

Optional lookup only. Use POST /v1/groups/search if the human already named a group and you need a canonical URL. Do not search for new groups to join. Do not chain search → join → post.

Create search job

text

POST https://api.clawpost.net/v1/groups/search
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

json

{
  "platform": "facebook",
  "query": "ai automation",
  "filters": { "minMembers": 1000, "privacy": "any" },
  "limit": 20
}

• platform: must be "facebook" (only supported platform for now).
• query: search keywords (required).
• filters.minMembers: minimum member count (optional).
• filters.privacy: "public", "private", or "any" (optional).
• limit: max results, up to 50 (optional, default 20).

Response:

json

{ "jobId": "<id>" }

Read search results

Poll GET https://api.clawpost.net/v1/jobs/:id until status is succeeded.

Results are in details.groupSearch.results[]. Each result:

Field	Type	Description
name	string	Group title (best-effort; may say "Group" if title not extractable)
url	string	Canonical group URL — use this for joining or posting
platformId	string?	Numeric Facebook group ID (when available)
slug	string?	URL slug (when available, e.g. "aiautomationagency")
privacy	string?	"public", "private", or "unknown"
memberCount	number?	Approximate member count (when visible on search page)
activityHint	string?	Raw activity text, e.g. "10 posts a day"
joinStatus	string?	"joined", "requested", "not_member", or "unknown"
score	number	0–1 relevance score (higher = better match)
reasons	string[]	Why this group scored well (e.g. "keyword match", "large member base", "active", "public")
signals	object?	Raw numeric signals: keywordHit, memberCount, activityPerDay

Reliability notes:

• url is always present and reliable. Use it as the primary identifier.
• name, memberCount, privacy, activityHint, and joinStatus are best-effort; they depend on what the search page exposes.
• score and reasons are computed by the extension from available signals.

---

4. Facebook Group Join

Out of scope for the recommended workflow. Do not call POST /v1/facebook/groups/join to grow membership. Only use it if the human already clicked Join in Facebook and is asking you to wait or retry a post.

Join a group so you can post to it. Use the url from search results (extract the group ID or pass the full URL).

Join a group

text

POST https://api.clawpost.net/v1/facebook/groups/join
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

json

{ "groupId": "123456789" }

Response: a job object. Poll GET /v1/jobs/:id for completion.

On success, check details.buttonState:

• "joined" – user is now a member; you can post.
• "requested" – group requires approval; wait and check status later.
• "not_member" – join may not have worked; retry or inspect.

Check group membership status

text

POST https://api.clawpost.net/v1/facebook/groups/status
Content-Type: application/json
clawpost-api-key: YOUR_CLAWPOST_API_KEY

json

{ "groupId": "123456789" }

Same polling pattern. details.buttonState tells you the current membership state.

---

Recommended agent workflow: post to groups they already belong to

1. Ask the human for a Facebook group URL they already belong to. Do not search-and-join.
2. Check membership only if needed (POST /v1/facebook/groups/status). If buttonState is not "joined", stop and tell the human to join in Facebook themselves.
3. Post: POST /v1/jobs/tweet with platform: "facebook" and platformPayload: { "groupUrl": "<url they gave>" }.
4. Never call POST /v1/facebook/groups/join unless the human explicitly clicked Join in Facebook and is only asking you to wait/retry a post. Auto-join is out of scope.

---

Security and privacy

• Social platform logins stay in the browser. Posting uses the user’s existing session in Chrome. Claw Post does not ask you to send OAuth tokens or social passwords through this API for posting.
• Extension scope. The published extension declares host access for supported social domains and the Claw Post API. Inspect the Web Store listing and manifest permissions before installing.
• API key is tenant-scoped. Each CLAWPOST_API_KEY is tied to a single account and can be rotated from the Dashboard at any time.
• HTTPS only. All communication between the agent, the API, and the extension uses TLS.
• No executable code. This skill is instruction-only. It contains no scripts, no disk writes, no packages, and no persistence mechanisms.
• Terms of Service: clawpost.net/terms

---

Error handling

Code / errorCode	Cause	Action
401	Invalid or missing API key	Check clawpost-api-key header
503 / EXTENSION_NOT_PAIRED	No paired extension	User must install and pair the extension at clawpost.net/dashboard
not_logged_in	User not logged in to the platform	User must log in to the platform in their browser
no_x_tab / no_platform_tab	No browser tab for the platform	Retry; extension will try to open one
content_script_unavailable	Extension could not reach the tab	Ask user to refresh the platform tab, then retry
selector_not_found	Platform UI changed or element not found	Retry after a short delay
group_not_approved	User not a member / not approved in this group	Do not auto-join. Tell the human to join in Facebook if they want this group. Retry only after they are a member.
challenge_required	Platform security check (captcha/checkpoint)	Ask user to complete the challenge in their browser, then retry

On any failure, poll GET /v1/jobs/:id and read error and errorCode for details.

---

Reference

Full API docs: https://clawpost.net/api-docs