OpenTweet X Poster

Post to X (Twitter) using the OpenTweet API. Create tweets, schedule posts, publish threads, upload media, run an evergreen queue, search inspiration tweets,...

Security Scan

The skill's requirements and runtime instructions match its stated purpose (posting to X via the OpenTweet API); nothing requested or instructed is disproportionate or unrelated to that purpose.

Purpose & Capability

Name/description describe posting, scheduling, media uploads, threads, analytics and the skill only requires an OPENTWEET_API_KEY and documents REST calls to https://opentweet.io — this credential and the described endpoints are consistent with the claimed purpose.

Instruction Scope

SKILL.md contains explicit API requests (GET/POST/PUT) to OpenTweet endpoints, guidance to check subscription/limits, and instructions for uploads and scheduling. It does not instruct reading local files, unrelated environment variables, or sending data to non-OpenTweet endpoints. Note: the actions allowed (posting, scheduling, auto-retweeting, evergreen queues) are powerful and will publish to user accounts when used.

Install Mechanism

No install specification or code is bundled (instruction-only). No downloads or package installs are requested, so there is no install-time code risk.

Credentials

Only OPENTWEET_API_KEY is required and declared as the primary credential. That is proportionate for an API-based posting/scheduling skill; no unrelated secrets or system paths are requested.

Persistence & Privilege

always:false and default model invocation settings mean the skill is not force-included system-wide. It does not request modifying other skills or system configs. The skill will be able to act (post/schedule) using the provided API key per normal platform behavior.

Assessment

This skill appears coherent with its purpose, but it can publish and schedule posts on your behalf — only provide an OPENTWEET_API_KEY you trust and that is scoped appropriately. If possible, test with a non-critical or throwaway X account first. Understand that the skill can publish immediately, schedule posts, enable auto-retweets, and access analytics for connected accounts; revoke the API key if you see unexpected activity. Verify OpenTweet's service and terms (https://opentweet.io) before granting access.

You can post to X (Twitter) using the OpenTweet REST API. All requests go to https://opentweet.io with the user's API key.

Authentication

Every request needs this header:

Authorization: Bearer $OPENTWEET_API_KEY
Content-Type: application/json

For file uploads, use Content-Type: multipart/form-data instead.

Before You Start

ALWAYS verify the connection first:

GET https://opentweet.io/api/v1/me

Returns subscription status, daily post limits, post counts, and connected X accounts. Check subscription.has_access is true and limits.remaining_posts_today > 0 before scheduling or publishing.

Multi-Account Support

Pro users get 1 X account, Advanced 3, Agency 10. Use the x_account_id parameter to target a specific account.

List connected accounts

GET https://opentweet.io/api/v1/accounts

Returns: { "accounts": [{ "id": "...", "x_handle": "@handle", "x_name": "Display Name", "is_primary": true, "nickname": null }] }

Using x_account_id

Add x_account_id to any POST/PUT body or GET query parameter to target a specific X account:

• Creating posts: { "text": "...", "x_account_id": "account_id_here" }
• Listing posts: GET /api/v1/posts?x_account_id=account_id_here
• Batch schedule: { "schedules": [...], "x_account_id": "account_id_here" }
• Analytics: GET /api/v1/analytics/overview?x_account_id=account_id_here
• Evergreen: GET /api/v1/evergreen/posts?x_account_id=account_id_here
• Best-times analyze: POST /api/v1/analytics/best-times/analyze body { "x_account_id": "..." }

When x_account_id is omitted, the primary account is used. Single-account users never need to specify it.

Post Management

Create a tweet

POST https://opentweet.io/api/v1/posts
Body: { "text": "Your tweet text" }

Optionally add "scheduled_date": "2026-05-01T10:00:00Z" to schedule it (requires active subscription, date must be in the future).

Create and publish immediately (one step)

POST https://opentweet.io/api/v1/posts
Body: { "text": "Hello from the API!", "publish_now": true }

Creates the post AND publishes to X in one request. Cannot combine with scheduled_date or bulk posts. Response includes status: "posted", x_post_id, and url (the real X post URL) on success.

Create a tweet with media

POST https://opentweet.io/api/v1/posts
Body: {
  "text": "Check out this screenshot!",
  "media_urls": ["https://url-from-upload-endpoint"]
}

Upload media first via POST /api/v1/upload, then pass the returned URL(s) in media_urls.

Create a thread

POST https://opentweet.io/api/v1/posts
Body: {
  "text": "First tweet of the thread",
  "is_thread": true,
  "thread_tweets": ["Second tweet", "Third tweet"]
}

Create a thread with per-tweet media

POST https://opentweet.io/api/v1/posts
Body: {
  "text": "Thread intro with image",
  "is_thread": true,
  "thread_tweets": ["Second tweet", "Third tweet"],
  "media_urls": ["https://intro-image-url"],
  "thread_media": [["https://img-for-tweet-2"], []]
}

thread_media is an array of arrays. Each inner array contains media URLs for the corresponding tweet in thread_tweets. Use [] for tweets with no media.

Post to an X Community

POST https://opentweet.io/api/v1/posts
Body: {
  "text": "Shared with the community!",
  "community_id": "1234567890",
  "share_with_followers": true
}

Auto-retweet a post

POST https://opentweet.io/api/v1/posts
Body: {
  "text": "This will get a boost.",
  "scheduled_date": "2026-05-01T10:00:00Z",
  "auto_retweet_enabled": true,
  "auto_retweet_offset_minutes": 240
}

After the post publishes, OpenTweet automatically retweets it from the same account auto_retweet_offset_minutes later. Works on PUT too. Range: 1–10080 minutes (up to 7 days). Both fields can also be set via PUT /api/v1/posts/{id}.

Bulk create (up to 50 posts)

POST https://opentweet.io/api/v1/posts
Body: {
  "posts": [
    { "text": "Tweet 1", "scheduled_date": "2026-05-01T10:00:00Z" },
    { "text": "Tweet 2", "scheduled_date": "2026-05-01T14:00:00Z" }
  ]
}

Schedule a post

POST https://opentweet.io/api/v1/posts/{id}/schedule
Body: { "scheduled_date": "2026-05-01T10:00:00Z" }

The date must be in the future. Use ISO 8601 format.

Publish immediately

POST https://opentweet.io/api/v1/posts/{id}/publish

No body needed. Posts to X right now. Response includes status: "posted", x_post_id, and url (the real X post URL).

Batch schedule (up to 50 posts)

POST https://opentweet.io/api/v1/posts/batch-schedule
Body: {
  "schedules": [
    { "post_id": "id1", "scheduled_date": "2026-05-02T09:00:00Z" },
    { "post_id": "id2", "scheduled_date": "2026-05-03T14:00:00Z" }
  ],
  "community_id": "optional-community-id",
  "share_with_followers": true,
  "x_account_id": "optional-account-id"
}

List posts

GET https://opentweet.io/api/v1/posts?status=scheduled&page=1&limit=20

Status options: scheduled, posted, draft, failed, evergreen (returns evergreen pool source posts).

Get a post

GET https://opentweet.io/api/v1/posts/{id}

Update a post

PUT https://opentweet.io/api/v1/posts/{id}
Body: {
  "text": "Updated text",
  "media_urls": ["https://..."],
  "scheduled_date": "2026-05-01T10:00:00Z",
  "auto_retweet_enabled": true,
  "auto_retweet_offset_minutes": 120
}

All fields optional. Cannot update already-published posts. Set scheduled_date to null to unschedule (convert back to draft).

Delete a post

DELETE https://opentweet.io/api/v1/posts/{id}

Default: if the post was already published, OpenTweet also deletes it from X. To delete only locally and leave the X post live, append ?delete_from_x=false. Response includes x_deleted and (if it failed) x_delete_error.

Media Upload

Upload an image or video

POST https://opentweet.io/api/v1/upload
Content-Type: multipart/form-data
Body: file=@your-image.png

Returns: { "url": "https://..." }

Supported formats: JPG, PNG, GIF, WebP (max 5MB), MP4, MOV (max 20MB).

Workflow: Upload first, then use the returned URL in media_urls or thread_media when creating/updating posts.

Evergreen Queue

The evergreen queue keeps a pool of timeless tweets and republishes them on a schedule with cooldown gaps so the same post doesn't repeat too often. Source posts stay as templates; the scheduler clones them as regular posts at the configured times. Requires an active paid subscription (not available on trial). Pro: 10 pool / 2 per day. Advanced: 999 pool / 10 per day.

Get queue settings + pool stats

GET https://opentweet.io/api/v1/evergreen/settings

Returns: enabled, posts_per_day, posting_times (["09:00","17:00"]), default_cooldown_days, plus pool counts and your plan limits.

Update queue settings

PUT https://opentweet.io/api/v1/evergreen/settings
Body: {
  "enabled": true,
  "posts_per_day": 2,
  "posting_times": ["09:00", "17:00"],
  "default_cooldown_days": 14
}

All fields optional. posting_times must be "HH:mm" strings. default_cooldown_days is 1–90. posts_per_day capped to your plan's daily limit.

List evergreen pool

GET https://opentweet.io/api/v1/evergreen/posts?page=1&limit=20&paused=false

Filter paused=true or paused=false. Each item includes cooldown_days, last_posted_at, times_posted, paused.

Add to evergreen pool

Mode 1 — convert an existing post:

POST https://opentweet.io/api/v1/evergreen/posts
Body: { "post_id": "507f1f77bcf86cd799439011", "cooldown_days": 14 }

Mode 2 — create a new evergreen post directly:

POST https://opentweet.io/api/v1/evergreen/posts
Body: {
  "text": "Timeless tweet text",
  "category": "Tips",
  "cooldown_days": 21,
  "is_thread": false,
  "media_urls": ["https://..."]
}

Get / update / remove an evergreen post

GET    https://opentweet.io/api/v1/evergreen/posts/{id}
PUT    https://opentweet.io/api/v1/evergreen/posts/{id}    # body: { "cooldown_days": 30, "paused": true }
DELETE https://opentweet.io/api/v1/evergreen/posts/{id}    # converts back to a draft (does not hard-delete)

GET also returns recent_posts — the last 5 published clones with their X URLs.

Evergreen publish history

GET https://opentweet.io/api/v1/evergreen/history?page=1&limit=20&source_id=optional

Lists published clones. Filter by source_id to see the history of a single evergreen post.

Inspiration (Search + Repurpose)

Search X for tweets and have AI rewrite them in the user's voice. Both endpoints require an active subscription. Search has a daily cap (Pro: 50/day, Advanced: 200/day, trial: 2/day). Repurpose counts against the AI generation daily quota.

Search inspiration tweets

GET https://opentweet.io/api/v1/inspiration/search?q=AI%20agents&max_results=20&sort_order=relevancy&lang=en&has_media=true&min_likes=100&min_retweets=10

Required: q. Optional filters: max_results, sort_order (relevancy or recency), lang, has_media, min_likes, min_retweets. Response includes data (tweets), meta.result_count, and usage (searches_used / remaining / daily_limit).

Repurpose a tweet with AI

POST https://opentweet.io/api/v1/inspiration/repurpose
Body: {
  "tweet_text": "Original tweet text to remix",
  "tweet_author": "@someone",
  "instructions": "Make it punchier and add a call to action",
  "tone": "casual",
  "save_as_draft": true
}

Returns repurposed.text, category, key_topics, plus draft.id when save_as_draft is true (default). Honors the user's voice profile and content pillars automatically. Optional x_account_id tags the saved draft.

Analytics

Account overview

GET https://opentweet.io/api/v1/analytics/overview

Returns posting stats (total posts, publishing rate, active days, avg posts/week, most active day/hour, threads, media posts), streaks (current, longest), trends (this week vs last, this month vs last, best month), category breakdown, and recent activity (daily counts for last 7 and 30 days).

Tweet engagement metrics (Advanced plan only)

GET https://opentweet.io/api/v1/analytics/tweets?period=30

Returns per-tweet engagement: likes, retweets, replies, quotes, impressions, bookmarks, engagement rate. Also includes top/worst performers, content type stats, engagement timeline, and best hours/days. Period: 7-365 days or "all".

Best posting times

GET https://opentweet.io/api/v1/analytics/best-times

Two analysis modes:

• engagement_weighted — uses real per-tweet engagement to score every hour×day cell. Returns heatmap, confidence, top_windows, best_day, best_hour, worst_day, worst_hour, insights. Only available after running an analysis.
• frequency_only — fallback based purely on when the user has posted. Returned when no engagement profile exists yet (needs ≥3 published posts).

Both modes also return legacy hour_distribution, day_distribution, best_hours, best_days keys for backward compatibility.

Trigger fresh best-times analysis

POST https://opentweet.io/api/v1/analytics/best-times/analyze
Body: {}    # optional: { "x_account_id": "..." }

Pulls the user's recent published tweets from X, computes engagement-weighted windows, and stores the profile. Has a built-in cooldown — if a recent analysis is still fresh, returns 429 with next_available_at. Returns success, profile (status ready / analyzing / insufficient_posts).

Common Workflows

First: verify your connection works:

1. GET /api/v1/me — check authenticated is true, subscription.has_access is true

Post a tweet right now (one step):

1. GET /api/v1/me — check limits.can_post is true
2. POST /api/v1/posts with { "text": "...", "publish_now": true }

Post a tweet with an image:

1. GET /api/v1/me — check limits
2. Upload: POST /api/v1/upload with the image file — get back a URL
3. Create + publish: POST /api/v1/posts with { "text": "...", "media_urls": ["<url>"], "publish_now": true }

Schedule a tweet:

1. GET /api/v1/me — check limits.remaining_posts_today > 0
2. POST /api/v1/posts with { "text": "...", "scheduled_date": "2026-05-01T10:00:00Z" } — you MUST make this HTTP call
3. Read the response JSON — confirm posts[0].status === "scheduled" and show the user the id and scheduled_date from the response

Schedule a tweet with auto-retweet boost:

1. GET /api/v1/me — check limits.remaining_posts_today > 0
2. POST /api/v1/posts with text, scheduled_date, auto_retweet_enabled: true, auto_retweet_offset_minutes: 240
3. Show the user the id and scheduled_date from the response

Schedule a week of content:

1. GET /api/v1/me — check remaining limit
2. Bulk create: POST /api/v1/posts with "posts": [...] array, each with a scheduled_date
3. Show the user the list of created post IDs and their scheduled dates from the response

Find inspiration and repurpose it:

1. GET /api/v1/inspiration/search?q=...&min_likes=500 — pick a tweet
2. POST /api/v1/inspiration/repurpose with tweet_text, tweet_author, save_as_draft: true
3. The saved draft's id can then be scheduled with POST /api/v1/posts/{id}/schedule

Set up an evergreen queue from existing drafts:

1. PUT /api/v1/evergreen/settings with { "enabled": true, "posts_per_day": 2, "posting_times": ["09:00","17:00"] }
2. For each draft to recycle: POST /api/v1/evergreen/posts with { "post_id": "...", "cooldown_days": 14 }
3. GET /api/v1/evergreen/history later to see what got published

Tune posting times based on engagement:

1. POST /api/v1/analytics/best-times/analyze — wait for profile.status: "ready" (poll if analyzing)
2. GET /api/v1/analytics/best-times — read top_windows and best_hour / best_day
3. Schedule new posts at the suggested times

Important Rules

• ALWAYS call GET /api/v1/me before scheduling or publishing to check limits and connected accounts.
• For multi-account users, call GET /api/v1/accounts and pass x_account_id to target a specific account.
• CRITICAL: You MUST make the actual HTTP API call for every operation. Never skip the call and generate a response from memory or context.
• CRITICAL: Always parse and use the ACTUAL JSON response from the API. Never fabricate or assume response values.
• CRITICAL: A 4xx or 5xx HTTP status means the operation FAILED — never report success to the user on an error response.
• CRITICAL: After scheduling a post, always show the user the id field from the API response. If you cannot show a real 24-character MongoDB ObjectId from the response, the call was not made.
• Post IDs are always 24-character MongoDB ObjectIds (e.g. "507f1f77bcf86cd799439011"), never short strings.
• Every post response includes a status field: "draft", "scheduled", "posted", or "failed".
• Published posts include a url field with the real X post URL. Always use this URL — never construct your own.
• To verify a post was published, check: status is "posted" AND url is present.
• To verify a post was scheduled, check: status is "scheduled" AND scheduled_date is present in the response.
• Tweet max length: 280 characters (per tweet in a thread).
• Bulk limit: 50 posts per request (create or batch-schedule).
• Rate limit: 60 requests/minute, 1,000/day (Pro); 300/min, 10,000/day (Advanced).
• Dates must be ISO 8601 and in the future — past dates are rejected.
• Active subscription required to schedule, publish, use evergreen, search inspiration, or repurpose. Creating drafts is free.
• Including scheduled_date or publish_now in POST /api/v1/posts requires a subscription.
• Upload media before creating posts — use the returned URL in media_urls or thread_media.
• Media limits: 5MB for images (JPG, PNG, GIF, WebP), 20MB for videos (MP4, MOV).
• URL-containing posts have a separate, plan-based daily cap. A 429 with a urlLimit payload means the post was saved as a draft instead of published.
• Tweet engagement analytics require the Advanced plan (returns 403 on Pro).
• Evergreen queue is not available during trial.
• auto_retweet_offset_minutes must be 1–10080 (up to 7 days) when auto_retweet_enabled is true.
• 403 = no subscription / X not connected, 429 = rate limit, daily post limit, URL post cap, or evergreen pool full.
• Check response status codes: 201=created, 200=success, 4xx=client error, 5xx=server error.

Safety Guardrails

Publishing is irreversible — once a tweet is posted to X it cannot be undone via the API (DELETE removes it locally and from X, but reposts are not the same tweet).

Confirm before publishing

• Before calling /publish or using publish_now: true, always tell the user which post(s) you are about to publish and ask for confirmation.
• Show the tweet text (truncated if long) and the post ID so the user can verify.

Scheduled posts ≠ ready to publish

• If a post has a scheduled_date in the future, it is meant to be published at that time by the scheduler — not right now.
• NEVER call /publish on a post that has a future scheduled_date unless the user explicitly asks you to publish it immediately.
• When the user asks to "publish" posts, clarify whether they want to publish NOW or schedule for later. Default to scheduling if dates are provided.

Evergreen sources are not regular drafts

• A post with isEvergreen: true is a recurring template. The scheduler publishes clones, not the source itself.
• NEVER call /publish directly on an evergreen source post (the API will reject it). Add it to the queue with POST /api/v1/evergreen/posts and let the scheduler run.

Batch operations — go slow

• When creating or scheduling more than 5 posts, summarize the batch (count, date range, first/last tweet previews) and ask the user to confirm before proceeding.
• Never bulk-create AND immediately publish in one go. Create as drafts or scheduled posts first, let the user review, then publish only on confirmation.
• When using batch-schedule, show the user the list of dates before sending the request.

Don't loop publish calls

• Never loop through a list of posts calling /publish on each one without explicit user approval for the full list.
• If the user asks to "publish all my drafts" or similar, list them first and get confirmation.

AI-generated content needs a review pass

• Before saving repurposed tweets as auto-scheduled posts, show the user the AI output and let them edit. The repurpose endpoint already saves to drafts by default — keep save_as_draft: true unless the user has reviewed.

Full API docs

For complete documentation: https://opentweet.io/api/v1/docs