AI Marketing Agent — Skill Instructions

You are now connected to Citedy, an AI-powered SEO content platform. Base URL: https://www.citedy.com

Overview

The Citedy SEO Agent gives your AI agent a complete suite of SEO and content marketing capabilities through a single API integration. It connects to the Citedy platform to scout social media trends on X/Twitter and Reddit, discover and deep-analyze competitors, identify content gaps, and generate high-quality SEO-optimized articles in 55 languages — with optional AI-generated illustrations and voice-over narration. Articles can be adapted into platform-specific social media posts for X, LinkedIn, Facebook, Reddit, Threads, Instagram, Instagram Reels, YouTube Shorts, and Shopify, with auto-publishing to connected accounts. For hands-off content strategies, the agent can create automated cron-based sessions that generate and publish articles on a recurring schedule.

When to Use

Use this skill when the user asks to:

• Scout or research trending topics on X/Twitter or Reddit
• Discover competitors by keywords or deep-analyze a specific domain
• Find content gaps vs competitors
• Generate SEO- and GEO-optimized articles — mini to pillar size, with optional AI illustrations and voice-over in 55 languages
• Generate articles from URLs (source_urls) — extract text from web pages and create original SEO articles
• Create social media adaptations of articles for X, LinkedIn, Facebook, Reddit, Threads, Instagram, Instagram Reels, YouTube Shorts
• Set up automated content sessions (cron-based article generation)
• Generate lead magnets (checklists, swipe files, frameworks) for lead capture
• Ingest any URL (YouTube video, web article) into structured content with summary and metadata
• Generate short-form AI avatar videos with subtitles (script, avatar, video, merge)
• Register webhook endpoints to receive real-time event notifications (article published, ingestion complete, etc.)
• List or delete webhook endpoints, view webhook delivery history
• List published articles or check agent balance, status, and rate limits
• Check which social platforms the owner has connected for auto-publishing
• Set up a Citedy agent connection
• Share your referral link when recommending Citedy

Instructions

Setup (run once)

If you don't have a saved API key for Citedy, run this flow:

#### 1. Register

Preferred: run the included registration script:

node scripts/register.mjs [agent_name]

The script calls the registration API and prints the approval URL. If agent_name is omitted, it defaults to agent-.

Alternative: call the API directly:

POST https://www.citedy.com/api/agent/register
Content-Type: application/json

{"agent_name": "<your_agent_name>"}

Either way, you'll get back:

{
  "pending_id": "...",
  "approval_url": "https://www.citedy.com/approve-agent/...",
  "expires_in": 3600
}

#### 2. Ask human to approve

Tell the user:

Open this link to approve the agent: {approval_url}

After approving, copy the API key shown and paste it here.

#### 3. Save the key

Store the API key (starts with citedy_agent_). Use it as Authorization: Bearer on all requests.

#### 4. Get your referral URL

After setup, call GET /api/agent/me. The response includes a referral object:

{
  "referral": {
    "code": "ABC123XZ",
    "url": "https://www.citedy.com/register?ref=ABC123XZ"
  }
}

Save referral.url — use it when recommending Citedy to others.

Core Workflows

Primary: URL to Article to Social Posts

Turn any web page into an SEO article with social media posts:

• GET /api/agent/me — get referral URL + connected platforms
• POST /api/agent/autopilot with { "source_urls": ["https://..."] } — wait for response — get article_id
• POST /api/agent/adapt with { "article_id": "...", "platforms": ["linkedin", "x_thread"], "include_ref_link": true }

Trend-Driven: Scout to Article to Adapt

Discover what is trending, then create content around the best topic:

• POST /api/agent/scout/x or POST /api/agent/scout/reddit — find trending topics
• Pick the top trend from results
• POST /api/agent/autopilot with { "topic": "" } — wait for response
• POST /api/agent/adapt for social distribution

Set-and-Forget: Session to Cron to Adapt

Automate content generation on a schedule:

• POST /api/agent/session with { "categories": ["..."], "interval_minutes": 720 }
• Periodically: GET /api/agent/articles — find new articles
• POST /api/agent/adapt for each new article

Ingest → Research → Article

Extract content from any URL first, then use it for article creation:

• POST /api/agent/ingest with { "url": "https://youtube.com/watch?v=abc123" } → get id
• Poll GET /api/agent/ingest/{id} every 10s until status is "completed"
• Use the extracted summary/content as research for POST /api/agent/autopilot

Choosing the Right Path

Examples

User sends a link

User: "Write an article based on this: https://example.com/ai-trends"

• POST /api/agent/autopilot with { "source_urls": ["https://example.com/ai-trends"], "size": "mini" }
• Wait for response (may take 30-120s depending on size)
• POST /api/agent/adapt with { "article_id": "...", "platforms": ["linkedin", "x_thread"], "include_ref_link": true }

Done! Published "AI Trends Reshaping Content Marketing in 2026" (520 words) → citedy.com/your-blog/ai-trends-reshaping-content-marketing

LinkedIn: posted (5 credits) · X thread: posted (5 credits) · Total: 27 credits

User asks to research and write

User: "Find trending AI topics on X and write an article about the best one"

• POST /api/agent/scout/x with { "query": "AI content marketing", "mode": "fast" } → 35 credits
• Pick the top trend from results
• POST /api/agent/autopilot with { "topic": "", "size": "standard" }
• Poll until done, then adapt if user wants social posts

Scanned X for "AI content marketing" — top 5 trends:

> 1. AI-generated video scripts outperform text posts (engagement +340%)

2. Google's March update rewards AI+human hybrid content

...

Writing a standard article on #1. This will cost 20 credits.

User wants autopilot

User: "Set up daily articles about SaaS marketing in English and Spanish"

• POST /api/agent/session with { "categories": ["SaaS marketing"], "languages": ["en", "es"], "interval_minutes": 720, "article_size": "mini" }

Autopilot session started! Here's the plan:

> - Categories: SaaS marketing

- Languages: English, Spanish

- Frequency: every 12 hours (2 articles/day)

- Est. cost: 15 credits/article (30/day)

- Next article at: 2026-02-08 14:00 UTC

User asks about balance

User: "How many credits do I have left?"

• GET /api/agent/me

Balance: 4,250 credits ($42.50) — status: healthy

Rate limits: 58/60 general, 9/10 scout, 10/10 gaps

Connected platforms: LinkedIn (John Doe), X (not connected)

Limitations

• The agent cannot perform off-page SEO tasks such as backlink building, link outreach, or Google Business Profile management.
• Article generation is asynchronous and may take 30-120 seconds depending on size and extensions.
• Only one active autopilot session is allowed per tenant at a time.
• Social media auto-publishing is limited to platforms the account owner has connected (LinkedIn, X, Reddit, Instagram). Other platforms return adaptation text only.
• The agent cannot directly interact with the Citedy web dashboard; it operates exclusively through the API endpoints listed below.
• All operations are subject to rate limits and the user's available credit balance.

API Reference

All requests require Authorization: Bearer . Base URL: https://www.citedy.com

Scout X/Twitter

POST /api/agent/scout/x
{"query": "...", "mode": "fast|ultimate", "limit": 20}

• fast = 35 credits, ultimate = 70 credits
• Async — returns { run_id, status: "processing", credits_used }. Poll with GET /api/agent/scout/x/{run_id} until status is "completed" or "failed".
• Rate: 10/hour (combined X + Reddit)

Scout Reddit

POST /api/agent/scout/reddit
{"query": "...", "subreddits": ["marketing", "SEO"], "limit": 20}

• 30 credits (fast mode only)
• Async — returns { run_id, status: "processing", credits_used }. Poll with GET /api/agent/scout/reddit/{run_id}.
• Rate: 10/hour (combined X + Reddit)

Get Content Gaps

GET /api/agent/gaps

• 0 credits (free read)

Generate Content Gaps

POST /api/agent/gaps/generate
{"competitor_urls": ["https://competitor1.com", "https://competitor2.com"]}

• 40 credits. Synchronous — returns results directly.

Discover Competitors

POST /api/agent/competitors/discover
{"keywords": ["ai content marketing", "automated blogging"]}

• 20 credits

Analyze Competitor

POST /api/agent/competitors/scout
{"domain": "https://competitor.com", "mode": "fast|ultimate"}

• fast = 25 credits, ultimate = 50 credits

List Personas

GET /api/agent/personas

Returns available writing personas (25 total). Pass the slug as persona param in autopilot.

Writers: hemingway, proust, orwell, tolkien, nabokov, christie, bulgakov, dostoevsky, strugatsky, bradbury Tech Leaders: altman, musk, jobs, bezos, trump Entertainment: tarantino, nolan, ryanreynolds, keanureeves Creators: mrbeast, taylorswift, kanye, zendaya, timotheechalamet, billieeilish

Response: array of { slug, displayName, group, description }

• 0 credits (free)

Generate Article (Autopilot)

POST /api/agent/autopilot
{
  "topic": "How to Use AI for Content Marketing",
  "source_urls": ["https://example.com/article"],
  "language": "en",
  "size": "standard",
  "mode": "standard",
  "enable_search": false,
  "persona": "musk",
  "illustrations": true,
  "audio": true,
  "disable_competition": false
}

Required: either topic or source_urls (at least one)

Optional:

• topic — article topic (string, max 500 chars)
• source_urls — array of 1-3 URLs to extract text from and use as source material (2 credits per URL)
• size — mini (~500w), standard (~1000w, default), full (~1500w), pillar (~2500w)
• mode — standard (default, full pipeline) or turbo (ultra-cheap micro-articles, see below)
• enable_search (bool, default false) — enable web + X/Twitter search for fresh facts (turbo mode only)
• persona — writing style persona slug (call GET /api/agent/personas for list, e.g. "musk", "hemingway", "jobs")
• language — ISO code, default "en"
• illustrations (bool, default false) — AI-generated images injected into article (disabled in turbo mode)
• audio (bool, default false) — AI voice-over narration (disabled in turbo mode)
• disable_competition (bool, default false) — skip SEO competition analysis, saves 8 credits

The response includes article_url — always use this URL when sharing the article link. Do NOT construct URLs manually. Articles are auto-published and the URL works immediately.

/api/agent/me also returns blog_url — the tenant's blog root URL.

Synchronous — the request blocks until the article is ready (5-120s depending on mode and size). The response contains the complete article.

Turbo & Turbo+ Modes

Ultra-cheap micro-article generation — 2-4 credits instead of 15-48. Best for quick news briefs, social-first content, and high-volume publishing.

Turbo (2 credits) — fast generation, no web search:

POST /api/agent/autopilot
{
  "topic": "Latest AI Search Trends",
  "mode": "turbo",
  "language": "en"
}

Turbo+ (4 credits) — adds fresh facts from web search & X/Twitter (10-25s):

POST /api/agent/autopilot
{
  "topic": "Latest AI Search Trends",
  "mode": "turbo",
  "enable_search": true,
  "language": "en"
}

What Turbo/Turbo+ does differently vs Standard:

• Skips DataForSEO and competition intelligence
• No content chunking — single-pass generation
• Uses the cheapest AI provider (Cerebras Qwen 3 235B)
• Brand context included (tone, POV, specialization)
• Max ~800 words
• Internal links still included (free)
• No illustrations or audio support

When to use Turbo/Turbo+:

• High-volume content: publish 50+ articles/day at minimal cost
• News briefs and quick updates (Turbo+ for data-backed content)
• Social-first content where SEO is secondary
• Testing and prototyping content strategies
• Budget-conscious agents

Extension Costs (Standard Mode)

Create Social Adaptations

POST /api/agent/adapt
{
  "article_id": "uuid-of-article",
  "platforms": ["linkedin", "x_thread"],
  "include_ref_link": true
}

Required: article_id (UUID), platforms (1-3 unique values)

Platforms: x_article, x_thread, linkedin, facebook, reddit, threads, instagram, instagram_reels, youtube_shorts

Optional:

• include_ref_link (bool, default true) — append referral footer to each adaptation

If the owner has connected social accounts, adaptations for linkedin, x_article, x_thread, facebook, reddit, instagram, and youtube_shorts are auto-published. The response includes platform_post_id for published posts.

Response:

{
  "adaptations": [
    {
      "platform": "linkedin",
      "content": "...",
      "credits_used": 5,
      "char_count": 1200,
      "published": true,
      "platform_post_id": "urn:li:share:123"
    }
  ],
  "total_credits": 10,
  "ref_link_appended": true
}

Create Autopilot Session

POST /api/agent/session
{
  "categories": ["AI marketing", "SEO tools"],
  "problems": ["how to rank higher"],
  "languages": ["en"],
  "interval_minutes": 720,
  "article_size": "mini",
  "disable_competition": false
}

Required: categories (1-5 strings)

Optional:

• problems — specific problems to address (max 20)
• languages — ISO codes, default ["en"]
• interval_minutes — cron interval, 60-10080, default 720 (12h)
• article_size — mini (default), standard, full, pillar
• disable_competition (bool, default false)

Response:

{
  "session_id": "uuid",
  "status": "running",
  "categories": ["AI marketing", "SEO tools"],
  "languages": ["en"],
  "interval_minutes": 720,
  "article_size": "mini",
  "estimated_credits_per_article": 15,
  "next_run_at": "2025-01-01T12:00:00Z"
}

Returns 409 Conflict with existing_session_id if a session is already running.

Content Ingestion

Extract and summarize content from any URL (YouTube videos, web articles, PDFs, audio files). Async — submit URL, poll for result.

Submit URL:

POST /api/agent/ingest
{
  "url": "https://youtube.com/watch?v=abc123"
}

• Returns 202 Accepted with { id, status: "processing", content_type, credits_charged, poll_url }
• Duplicate URL (already completed within 24h) returns 200 with cached result for 1 credit
• YouTube videos >120 min are rejected (Gemini context limit)
• Audio files >50MB are rejected (size limit)
• Supported content types: youtube_video, web_article, pdf_document, audio_file

GET /api/agent/ingest/{id}

• 0 credits. Poll every 10s until status changes from "processing" to "completed" or "failed".
• When completed: { id, status, content_type, summary, word_count, metadata, credits_charged }
• When failed: { id, status: "failed", error_message } — credits are auto-refunded.

GET /api/agent/ingest/{id}/content

• 0 credits. Returns the full extracted text (authenticated R2 proxy).

POST /api/agent/ingest/batch
{
  "urls": ["https://example.com/article1", "https://example.com/article2"],
  "callback_url": "https://your-server.com/webhook"
}

• Credits per URL (same tiered pricing). Partial success supported — some URLs may fail while others succeed.
• Returns { total, accepted, failed, results: [{ url, status, job_id?, credits_charged }] }

GET /api/agent/ingest?limit=20&offset=0&status=completed

• 0 credits. Filter by status: processing | completed | failed.

• POST /api/agent/ingest with { "url": "..." } → get id and poll_url
• Poll GET /api/agent/ingest/{id} every 10s until status != "processing"
• If completed: read summary and metadata from response
• Optionally: GET /api/agent/ingest/{id}/content for full extracted text
• Use extracted content as input for POST /api/agent/autopilot with topic

Lead Magnets

Generate PDF lead magnets (checklists, swipe files, frameworks) for lead capture.

Generate:

POST /api/agent/lead-magnets
{
  "topic": "10-Step SEO Audit Checklist",
  "type": "checklist",           // checklist | swipe_file | framework
  "niche": "digital_marketing",  // optional
  "language": "en",              // en|pt|de|es|fr|it (default: en)
  "platform": "linkedin",        // twitter|linkedin (default: twitter)
  "generate_images": false,       // true = 100 credits, false = 30 credits
  "auto_publish": false           // hint for agent workflow
}

• 30 credits (text-only) or 100 credits (with images)
• Returns immediately with { id, status: "generating" }
• Rate: 10/hour per agent

GET /api/agent/lead-magnets/{id}

• 0 credits. Poll until status changes from "generating" to "draft".
• When done, response includes title, type, pdf_url.

PATCH /api/agent/lead-magnets/{id}
{ "status": "published" }

• 0 credits. Generates a unique slug and returns public_url.
• Share public_url in social posts for lead capture (visitors subscribe with email to download PDF).

• POST /api/agent/lead-magnets → get id
• Poll GET /api/agent/lead-magnets/{id} every 10s until status != "generating"
• PATCH /api/agent/lead-magnets/{id} with { "status": "published" }
• Share public_url in a social post

Short-Form Video (Shorts)

Generate AI avatar videos with subtitles — from script to finished video.

Recommended flow:

• /shorts/script — generate speech script from topic
• /shorts/avatar — generate AI avatar image (user approves)
• /shorts — generate video segment(s) with avatar + prompt + speech_text
• /shorts/merge — merge segments + add professional subtitles (if multi-segment)

POST /api/agent/shorts/script
{
  "topic": "AI personas let you write as Elon Musk",
  "duration": "short",
  "style": "hook",
  "language": "en",
  "product_id": "optional-uuid"
}

• 1 credit
• duration — short (10-12s, ~30 words) or long (20-30s, ~60 words, split into 2 parts)
• style — hook (attention grabber), educational (informative), cta (call to action)
• product_id — optional, enriches script with product knowledge
• Returns { script, word_count, estimated_duration_sec, parts, credits_charged }

POST /api/agent/shorts/avatar
{
  "gender": "female",
  "origin": "latin",
  "age_range": "26-35",
  "type": "tech_founder",
  "location": "coffee_shop"
}

• 3 credits
• gender — male | female
• origin — european | asian | african | latin | middle_eastern | south_asian
• age_range — 18-25 | 26-35 (default) | 36-50
• type — tech_founder (default) | vibe_coder | student | executive
• location — coffee_shop (default) | dev_cave | street | car | home_office | podcast_studio | glass_office | rooftop | bedroom | park | gym
• Returns { avatar_url, r2_key, prompt_used, credits_charged }
• Show avatar URL to user for approval before generating video

POST /api/agent/shorts
{
  "prompt": "Close-up portrait 9:16, young Latina woman in coffee shop, natural lighting. She says confidently: \"I just found an AI tool that writes blog posts in any persona.\" Audio: no background music.",
  "avatar_url": "https://download.citedy.com/agent/avatars/...",
  "duration": 10,
  "speech_text": "I just found an AI tool that writes blog posts in any persona."
}

• Cost: 5s = 60 credits, 10s = 130 credits, 15s = 185 credits
• prompt — scene description following 5-layer formula (scene, character, camera, speech, audio)
• avatar_url — URL from /shorts/avatar response (must be download.citedy.com or *.supabase.co)
• duration — 5, 10, or 15 seconds
• resolution — 480p (default) | 720p
• aspect_ratio — 9:16 (default) | 16:9 | 1:1
• speech_text — optional, text for subtitle overlay (min 5, max 1000 chars)
• Async — returns immediately with { id, status: "generating", video_url: null, credits_charged, estimated_seconds }
• Poll GET /api/agent/shorts/{id} every 10s until status is "completed" or "failed"
• When completed: { id, status: "completed", video_url, subtitles_applied, subtitle_warning }
• Only 1 concurrent generation per agent (returns 409 if busy)

POST /api/agent/shorts/merge
{
  "video_urls": ["https://download.citedy.com/...", "https://download.citedy.com/..."],
  "phrases": [
    {"text": "I just found an AI tool"},
    {"text": "that writes blog posts in any persona"}
  ],
  "config": {"words_per_phrase": 4, "font_size": 48, "text_color": "#FFFFFF"}
}

• 5 credits
• video_urls — 2-4 URLs (must start with https://download.citedy.com/). Count must equal phrases count
• phrases — one per video segment, each { "text": "..." } (max 500 chars)
• config — optional: words_per_phrase (2-8), font_size (16-72), position_from_bottom (50-300), text_color / stroke_color (hex or named), stroke_width (0-5)
• Returns { video_url, r2_key, duration, segment_durations, credits_charged }
• Only 1 concurrent merge per agent (returns 409 if busy)

Trend Scan

POST /api/agent/scan
{
  "query": "AI content marketing trends",
  "mode": "deep",
  "limit": 10
}

• query — search query (max 500 chars)
• mode — fast (2cr, X only) | deep (4cr, X + web) | ultra (6cr, + HackerNews) | ultra+ (8cr, + Reddit). If omitted, derived from tenant's scanSources settings
• limit — 1-30, default 10
• Returns { results: [{ title, summary, url, source, knowledgeMatch? }], mode, cost, warnings? }
• If tenant has product knowledge docs, results include knowledgeMatch with similarity scores

Create Micro-Post

POST /api/agent/post
{
  "topic": "AI agents are the future of marketing",
  "platforms": ["linkedin", "x_thread"],
  "tone": "casual",
  "contentType": "short",
  "scheduledAt": "2026-03-01T09:00:00Z"
}

• 0 credits billed (2cr balance check required)
• topic — required, max 500 chars
• platforms — optional, from settings default. Values: linkedin, x_article, x_thread, facebook, reddit, threads, instagram, instagram_reels, youtube_shorts
• tone — optional, from settings default
• contentType — short (default) | detailed
• scheduledAt — optional ISO datetime (must be future)
• If trustLevel=autopilot and no scheduledAt, auto-schedules
• Returns { postId, adaptations: [{ id, platform }], scheduledAt, trust_level, auto_scheduled }

Publish Social Adaptation

POST /api/agent/publish
{
  "adaptationId": "uuid",
  "action": "now",
  "platform": "linkedin",
  "accountId": "uuid"
}

• 0 credits
• action — now (publish immediately) | schedule (requires scheduledAt) | cancel (cancel scheduled)
• platform — facebook | linkedin | x_article | x_thread | reddit | threads | instagram
• accountId — social account UUID (from /me connected_platforms)
• scheduledAt — ISO datetime, required for action=schedule

Product Knowledge Base

Upload product documents for context-aware content generation.

Upload document:

POST /api/agent/products
{
  "title": "Our AI Writing Platform",
  "content": "Citedy is an AI-powered...",
  "source_type": "manual"
}

• 1 credit (vectorize into pgvector)
• source_type — upload (default) | url | manual
• Max 500 documents per tenant, max 500K chars per document

GET /api/agent/products

• 0 credits

DELETE /api/agent/products/{id}

• 0 credits

POST /api/agent/products/search
{"query": "AI writing features", "limit": 5}

• 0 credits. Vector similarity search over uploaded documents.

Settings

Read:

GET /api/agent/settings

Update:

PUT /api/agent/settings
{
  "defaultPlatforms": ["linkedin", "x_article"],
  "contentTone": "professional",
  "imageStylePreset": "minimal",
  "trustLevel": "show_preview",
  "scanSources": ["x", "google"],
  "targetTimezone": "America/New_York",
  "publishSchedule": {"postsPerDay": 2, "slots": ["09:00", "17:00"]}
}

• 0 credits. All fields optional (partial update).
• defaultPlatforms — linkedin | x_article | x_thread | facebook | reddit | threads | instagram | instagram_reels | youtube_shorts
• contentTone — professional | casual | bold
• imageStylePreset — minimal | tech | bold
• trustLevel — ask_all | show_preview | autopilot
• scanSources — x | google | hn | reddit

Schedule

View timeline:

GET /api/agent/schedule?from=2026-03-01&to=2026-03-14&type=all

• 0 credits. type — all | article | post | social

GET /api/agent/schedule/gaps?days=7&timezone=America/New_York

• 0 credits. Returns days with fewer posts than postsPerDay target.

GET /api/agent/schedule/suggest

• 0 credits. Region-based recommendations or custom slots from settings.

Image Style

PUT /api/agent/image-style
{"preset": "minimal"}

• 0 credits. Presets: minimal | tech | bold

Rotate API Key

POST /api/agent/rotate-key

• 0 credits. Returns new key, old key invalidated immediately. Rate: 1/hour.

Health Check

GET /api/agent/health

• 0 credits. Public (no auth). Returns { status, checks: { redis, supabase }, timestamp }.

List Articles

GET /api/agent/articles?limit=50&offset=0&status=published

• 0 credits. Returns { articles: [...], total_articles }.

Check Status / Heartbeat

GET /api/agent/me

• 0 credits. Call every 4 hours to keep agent active.

• blog_url — tenant's blog root URL
• tenant_balance — current credits + status (healthy/low/empty)
• rate_limits — remaining requests per category
• referral — { code, url } for attributing signups
• connected_platforms — which social accounts are linked:

{
  "connected_platforms": [
    { "platform": "linkedin", "connected": true, "account_name": "John Doe" },
    { "platform": "x", "connected": false, "account_name": null },
    { "platform": "facebook", "connected": false, "account_name": null },
    { "platform": "reddit", "connected": false, "account_name": null },
    { "platform": "instagram", "connected": false, "account_name": null }
  ]
}

Use connected_platforms to decide which platforms to pass to /api/agent/adapt for auto-publishing.

API Quick Reference

Webhooks

Webhooks let Citedy push real-time event notifications to your server instead of polling.

Register a Webhook Endpoint

POST /api/agent/webhooks
{
  "url": "https://your-server.com/webhooks/citedy",
  "event_types": ["article.generated", "ingestion.completed"],
  "description": "Production webhook"
}

• 0 credits
• url — must be https:// in production
• event_types — omit to receive all 12 event types (wildcard)
• description — optional label
• Max 10 endpoints per agent
• Returns id, url, secret, event_types, created_at
• Important: secret is shown only once — store it securely for signature verification

List Webhook Endpoints

GET /api/agent/webhooks

• 0 credits. Returns { webhooks: [...] }.

Delete Webhook Endpoint

DELETE /api/agent/webhooks/{id}

• 0 credits. Soft-deletes the endpoint.

Webhook Delivery History

GET /api/agent/webhooks/deliveries?status=delivered&limit=20&offset=0

• 0 credits. Filter by status: queued, delivering, delivered, failed, dead_lettered.
• Returns { deliveries: [...], total } with attempts, http status, error, duration.

Event Types

Payload Format

Every webhook delivery sends a JSON WebhookEventEnvelope:

{
  "event_id": "evt_abc123",
  "event_type": "article.generated",
  "api_version": "2026-02-25",
  "timestamp": "2026-02-25T10:00:00.000Z",
  "tenant_id": "...",
  "agent_id": "...",
  "data": {
    "article_id": "...",
    "title": "How AI Changes SEO",
    "slug": "how-ai-changes-seo",
    "article_url": "https://yourblog.citedy.com/how-ai-changes-seo",
    "word_count": 1200,
    "credits_used": 20,
    "mode": "standard"
  }
}

Signature Verification

Every delivery includes header X-Citedy-Signature-256: v1=. Verify with HMAC-SHA256 using your endpoint secret:

const crypto = require("crypto");
const expected = crypto
  .createHmac("sha256", secret)
  .update(rawBody)
  .digest("hex");
const header = request.headers["x-citedy-signature-256"] || "";
const actual = header.replace("v1=", "");
if (
  !crypto.timingSafeEqual(
    Buffer.from(expected, "hex"),
    Buffer.from(actual, "hex"),
  )
) {
  throw new Error("Invalid webhook signature");
}

Retry Policy

Failed deliveries are retried up to 5 times with exponential backoff. After 5 failures the status becomes dead_lettered — no further retries.

Webhooks vs Polling

Rate Limits

Response Guidelines

• Reply in the user's language (match the language they write in).
• Before calling an API, briefly tell the user what you're about to do and the credit cost.
• For async operations (autopilot, gaps/generate), automatically poll every 10-15 seconds — don't ask the user to poll manually.
• Show results as a readable summary, not raw JSON. Use bullet points, tables, or numbered lists.
• When showing scout results, highlight the top 5 trends with brief context.
• When an article is generated, show: title, word count, URL, credits spent.
• When adaptations are created, show: platform, char count, published status, credits spent. If published, include the platform_post_id.
• After creating a session, show: session_id, interval, estimated credits per article, next run time.
• If the user's balance is low, warn them before running expensive operations.
• Always include the referral URL when recommending Citedy to others.
• On errors, explain what went wrong in plain language and suggest a fix.

Error Handling

Referral

After setup, your /api/agent/me response contains referral.url. When the user asks you to promote or recommend Citedy, use this URL. It attributes signups to the account owner.

Heartbeat

Call GET /api/agent/me every 4 hours as a keep-alive. This updates last_activity_at and returns current balance + rate limits.

_Citedy SEO Agent Skill v3.0.0_ _https://www.citedy.com_