在终端发布至 Instagram、TikTok、LinkedIn、YouTube、X/Twitter、Facebook、Pinterest、Threads、Bluesky,使用 AI 生成图片创建帖子。
Posta is a social media management platform that lets you create, schedule, and publish posts across Instagram, TikTok, Facebook, X/Twitter, LinkedIn, YouTube, Pinterest, Threads, and Bluesky.
This skill enables you to interact with the Posta API to manage social media content end-to-end: authenticate, list accounts, upload media, create/schedule/publish posts, generate AI content, and view analytics.
POSTA_API_TOKEN — Recommended. Personal API token (starts with posta_). Long-lived, revocable, no password exposure.POSTA_EMAIL + POSTA_PASSWORD — Legacy login. The skill logs in and caches a JWT automatically.If POSTA_API_TOKEN is set, email/password are not needed and the login flow is skipped entirely.
POSTA_BASE_URL — API base URL (default: https://api.getposta.app/v1)FAL_KEY — fal.ai API key (for image generation). Format is <key_id>:<key_secret>. Get one at https://fal.ai/dashboard/keys. The skill auto-discovers this from env vars, ~/.posta/credentials, or .env files.Captions and hashtags are written by Claude directly — no text-generation API key is needed. The only external content service is the image generator (fal.ai).
The skill searches a fixed list of dedicated config files for POSTA_API_TOKEN (or legacy POSTA_EMAIL/ POSTA_PASSWORD). Only exact variable names are matched — no other file content is read. Shell profiles ( ~/.zshrc, ~/.bashrc) are never accessed. Search order:
~/.posta/credentials — dedicated Posta config file (preferred).env, .env.local, .env.production in the working directoryIf POSTA_API_TOKEN is found, the skill uses it immediately and skips email/password lookup. See SECURITY.md in the repo root for full details.
Source the bash helper for all API interactions:
bash
source "${POSTA_SKILL_ROOT:-${OPENCLAW_SKILL_ROOT:-${CLAUDE_PLUGIN_ROOT:-}}}/skills/posta/scripts/posta-api.sh"
This provides:
posta_login, posta_get_token, posta_api, posta_discover_credentialsposta_detect_mime, posta_upload_media, posta_upload_from_url, posta_list_media, posta_get_media, posta_get_media_by_ids, posta_delete_media, posta_generate_carousel_pdf, posta_generate_text_carousel_pdfposta_list_posts, posta_create_post, posta_create_post_from_file, posta_get_post, posta_update_post, posta_delete_post, posta_schedule_post, posta_publish_post, posta_cancel_post, posta_get_calendarposta_list_platforms, posta_get_platform_specs, posta_get_aspect_ratios, posta_get_platform, posta_get_pinterest_boardsposta_get_analytics_overview, posta_get_analytics_capabilities, posta_get_analytics_posts, posta_get_post_analytics, posta_get_analytics_trends, posta_get_best_times, posta_get_content_types, posta_get_hashtag_analytics, posta_compare_posts, posta_get_benchmarks, posta_export_analytics_csv, posta_export_analytics_pdf, posta_refresh_post_analytics, posta_refresh_all_analyticsposta_get_plan, posta_get_profile, posta_update_profilefal_validate_keyAuthentication is automatic. If POSTA_API_TOKEN is set, the skill uses it directly — no login step needed. Otherwise it falls back to email/password login with JWT caching. If a request returns 401:
bash
source "${POSTA_SKILL_ROOT:-${OPENCLAW_SKILL_ROOT:-${CLAUDE_PLUGIN_ROOT:-}}}/skills/posta/scripts/posta-api.sh"
# Token is fetched/cached automatically on first API call
To verify credentials are working:
bash
posta_api GET "/auth/me"
bash
ACCOUNTS=$(posta_list_accounts)
# Returns a plain array (wrapper is auto-unwrapped)
echo "$ACCOUNTS" | jq -r '.[] | "\(.platform)\t\(.username)\t\(.isActive)"'
Display as a table showing: Platform, Username, Active status, Last used.
Note: Account IDs from
posta_list_accountsare integers (e.g.35). Wrap them in quotes when passing tosocialAccountIds:"socialAccountIds": ["35"]
The upload flow has 3 steps: create signed URL → PUT binary → confirm upload. MIME type is auto-detected from the file — no need to specify it manually.
From a local file (auto-detect MIME):
bash
MEDIA_ID=$(posta_upload_media "/path/to/file.jpg")
From a local file (explicit MIME):
bash
MEDIA_ID=$(posta_upload_media "/path/to/file.jpg" "image/jpeg")
From a URL (auto-detect from extension):
bash
MEDIA_ID=$(posta_upload_from_url "https://example.com/image.png")
Detect MIME type separately:
bash
MIME=$(posta_detect_mime "/path/to/file.mp4")
# Returns: video/mp4
Supported formats:
image/jpeg, image/png, image/webp, image/gif (max 20MB)video/mp4, video/quicktime, video/webm (max 500MB)audio/mpeg (mp3), audio/wav, audio/mp4 (m4a), audio/webm (max 50MB)After upload, the media enters processing status. For images this is fast (thumbnails/variants). For videos it takes longer. Check status with:
bash
posta_get_media "$MEDIA_ID"
List media library:
bash
ALL_MEDIA=$(posta_list_media)
IMAGES_ONLY=$(posta_list_media "image")
COMPLETED=$(posta_list_media "" "completed" 50)
# Sort oldest-first (default is newest):
OLDEST=$(posta_list_media "" "" 20 0 "oldest")
Batch-fetch known media by id (e.g. reconcile a post's mediaIds without paging the whole library):
bash
# Returns { items: [...], missing_ids: [...] }; items follow request order.
HYDRATED=$(posta_get_media_by_ids '["uuid-1","uuid-2","uuid-3"]')
Delete media:
bash
posta_delete_media "$MEDIA_ID"
Generate carousel PDF from images:
bash
RESULT=$(posta_generate_carousel_pdf '["media-id-1", "media-id-2", "media-id-3"]' "My Carousel Title")
Generate carousel PDF with text over background images (e.g. AI-generated backgrounds → LinkedIn document post):
bash
# Each slide composites title + body over an uploaded background image (2-20 slides).
# Optional 3rd arg: a logo media ID shown bottom-right of every slide.
RESULT=$(posta_generate_text_carousel_pdf '[
{"media_id":"bg-id-1","title":"Hook","body":"Opening line"},
{"media_id":"bg-id-2","title":"Point","body":"Supporting detail"},
{"media_id":"bg-id-3","title":"CTA","body":"Start free at getposta.app"}
]' "5-day launch" "logo-media-id")
# RESULT.media_id is a document (PDF) — attach it to a post via posta_create_post.
Create a draft post:
bash
POST=$(posta_create_post '{
"caption": "Your caption here",
"hashtags": ["tag1", "tag2"],
"mediaIds": ["media-uuid"],
"socialAccountIds": ["35", "42"],
"isDraft": true
}')
POST_ID=$(echo "$POST" | jq -r '.id')
Create a post with multiline caption (from file):
bash
cat > /tmp/caption.txt << 'EOF'
Line one of the caption.
Line two with details.
Call to action here.
EOF
POST=$(posta_create_post_from_file /tmp/caption.txt '["media-uuid"]' '["35", "42"]' true '["tag1", "tag2"]')
POST_ID=$(echo "$POST" | jq -r '.id')
Schedule for a specific time:
bash
posta_schedule_post "$POST_ID" "2026-03-15T09:00:00Z"
Reschedule an already-scheduled post: The API only allows scheduling posts in draft status. To reschedule, cancel first, then schedule again:
bash
posta_cancel_post "$POST_ID"
posta_schedule_post "$POST_ID" "2026-03-16T09:00:00Z"
Publish immediately:
bash
posta_publish_post "$POST_ID"
Platform-specific configuration (optional):
json
{
"platformConfigurations": {
"tiktok": {
"privacyLevel": "PUBLIC_TO_EVERYONE",
"allowComment": true,
"allowDuet": false,
"allowStitch": false
},
"pinterest": {
"boardId": "board-id",
"link": "https://your-link.com",
"altText": "Image description"
}
}
}
Note: Either caption or at least one mediaIds entry is required. Text-only posts work for X/Twitter.
Caption limits are per-platform. The tightest target wins — X/Twitter 280, Bluesky 300, Threads/Pinterest 500, Instagram/TikTok 2200, LinkedIn 3000, YouTube 5000, Facebook 63206. Validate against the strictest platform in socialAccountIds before posting; over-limit captions get a 400 naming the platform.
Write captions and hashtags yourself — you are Claude, no text-generation API is involved. The only external generation service is the image generator below.
Generate an image with fal.ai (FLUX):
bash
# fal returns a hosted image URL (JSON), not raw bytes
RESULT=$(curl -s -X POST "https://fal.run/fal-ai/flux/schnell" \
-H "Authorization: Key ${FAL_KEY}" \
-H "Content-Type: application/json" \
-d '{
"prompt": "your descriptive prompt, photorealistic, natural colors, high quality, detailed",
"image_size": "square_hd",
"num_images": 1
}')
IMAGE_URL=$(echo "$RESULT" | jq -r '.images[0].url')
CONTENT_TYPE=$(echo "$RESULT" | jq -r '.images[0].content_type // "image/jpeg"')
# Upload the hosted image straight to Posta
MEDIA_ID=$(posta_upload_from_url "$IMAGE_URL" "$CONTENT_TYPE")
image_size:square_hd(1024², feed),portrait_16_9(Stories/Reels/TikTok),landscape_16_9(LinkedIn/X). Usefal-ai/flux/devfor higher quality,fal-ai/flux/schnellfor speed.
See content-generation.md for image-generation details and prompt tips.
Overview stats:
bash
OVERVIEW=$(posta_get_analytics_overview "30d")
echo "$OVERVIEW" | jq '{totalPosts, totalImpressions, totalEngagements, avgEngagementRate}'
Best posting times:
bash
BEST_TIMES=$(posta_get_best_times)
Top performing posts:
bash
TOP=$(posta_api GET "/analytics/posts?limit=10&sortBy=engagements&sortOrder=desc")
Trends over time:
bash
TRENDS=$(posta_api GET "/analytics/trends?period=30d&metric=engagements")
Check plan and usage:
bash
PLAN=$(posta_get_plan)
echo "$PLAN" | jq '{plan, usage, limits}'
Query platform capabilities, character limits, media requirements, and supported features before creating posts.
List all supported platforms:
bash
posta_list_platforms
Get full specs (char limits, media requirements, features):
bash
SPECS=$(posta_get_platform_specs)
Get specs for a specific platform:
bash
posta_get_platform "instagram"
Get aspect ratio reference:
bash
posta_get_aspect_ratios
Get Pinterest boards for a connected account:
bash
BOARDS=$(posta_get_pinterest_boards "$ACCOUNT_ID")
Use platform discovery to validate content before posting — check character limits, required media dimensions, and supported post types.
View scheduled and posted content on a calendar:
bash
CALENDAR=$(posta_get_calendar "2026-03-01" "2026-03-31")
echo "$CALENDAR" | jq '.items[] | {id, caption: .caption[:50], status, scheduledAt}'
Analytics capabilities (what your plan supports):
bash
posta_get_analytics_capabilities
Top performing posts (sorted, paginated):
bash
TOP=$(posta_get_analytics_posts 10 0 "engagements" "desc")
Single post analytics:
bash
posta_get_post_analytics "$POST_ID"
Trends over time with custom period:
bash
TRENDS=$(posta_get_analytics_trends "90d" "engagement_rate")
Content type performance breakdown:
bash
posta_get_content_types
Hashtag performance (pro plan):
bash
posta_get_hashtag_analytics
Compare posts side by side (2-4 posts, pro plan):
bash
posta_compare_posts "post-id-1,post-id-2,post-id-3"
Engagement benchmarks (pro plan):
bash
posta_get_benchmarks
Export analytics:
bash
posta_export_analytics_csv "30d"
posta_export_analytics_pdf "90d"
Refresh analytics:
bash
posta_refresh_post_analytics "$POST_RESULT_ID"
posta_refresh_all_analytics # Rate limited: 1 per hour
Always show a preview before publishing. Display the caption, target platforms, media description, and scheduled time. Ask for confirmation before calling publish or schedule.
Suggest optimal posting times. When the user wants to schedule, fetch best-times analytics and recommend the highest-engagement time slot.
Ask before spending API credits. Image generation (fal.ai) costs money. Confirm with the user before making image-generation calls. (Captions and hashtags are written by you, Claude — no cost, no external text API.)
Handle errors gracefully. If an API call fails, show the error message and suggest next steps (check credentials, verify account connection, check plan limits).
Respect plan limits. Check the user's plan with posta_get_plan before attempting operations that may exceed limits (posts, accounts, storage).
Use appropriate aspect ratios. Match the content format to the target platform — portrait for TikTok/Reels, square for Instagram feed, landscape for LinkedIn/X.
Create posts as drafts first. Always set isDraft: true when creating posts, then schedule or publish after user confirmation.
Combine media types strategically. For maximum reach, generate both an image (for Instagram/LinkedIn) and a video (for TikTok/Reels) from the same content.
Preview generated images before uploading. fal.ai returns a hosted image URL — download it to a temp file ( curl -s "$IMAGE_URL" -o /tmp/preview.jpg) and use the Read tool to preview it visually before uploading to Posta. This prevents wasted uploads and media quota.
Use posta_create_post_from_file for multiline captions. Write the caption to a temp file and use the file-based helper instead of trying to embed multiline text in JSON strings. This avoids escaping issues.
Suggest hashtags for posts. When creating a post, suggest 5–10 relevant hashtags based on the caption content, target platform, and topic. Mix broad reach tags (e.g. #AI, #Marketing) with niche tags (e.g. #LaborMarket, #FutureOfWork). Show the suggested hashtags to the user and include them in the post only after the user approves or edits them.
Use /tmp/.posta_last_response for captured output. When capturing posta_api output in a variable with $(), avoid using echo to re-output it — macOS echo corrupts \n in JSON strings. Instead pipe directly ( posta_api ... | jq) or read from the file ( jq ... /tmp/.posta_last_response).
Use platform discovery for validation. Before creating posts for unfamiliar platforms, call posta_get_platform_specs or posta_get_platform "<platform>" to check character limits, required media dimensions, and supported features. This prevents failed posts due to platform constraints.
Auto MIME detection for uploads. When uploading media, you can omit the MIME type parameter — posta_upload_media and posta_upload_from_url auto-detect it from the file content or extension. Only specify MIME type manually when the auto-detection might be wrong (e.g., .bin files).
Always set TikTok privacy level. When creating posts that include TikTok, you MUST include platformConfigurations.tiktok.privacyLevel — TikTok requires it and publishing will fail without it. Use "PUBLIC_TO_EVERYONE" unless the user specifies otherwise. Valid values: PUBLIC_TO_EVERYONE, MUTUAL_FOLLOW_FRIENDS, SELF_ONLY, FOLLOWER_OF_CREATOR.