{"slug":"ai-citation-tracker","title":"ai-citation-tracker","summary":"Track brand mentions, URL citations, and share-of-voice across the 2026 AI search surface — ChatGPT (with browsing), Perplexity, Claude (with search), Google Gemini, Google AI Overviews, Bing Copilot, You.com, Phind, and Microsoft Copilot..","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-10-01T15:40:10.903291Z","repo":{"url":"https://github.com/tinh2/skills-hub-registry","stars":18,"forks":6,"license":null,"updatedAt":"2026-09-04T17:22:55Z"},"bodyHtml":"
\n

name: ai-citation-tracker\ndescription: \"Track brand mentions, URL citations, and share-of-voice across the 2026 AI search surface — ChatGPT (with browsing), Perplexity, Claude (with search), Google Gemini, Google AI Overviews, Bing Copilot, You.com, Phind, and Microsoft Copilot..\"\nversion: \"1.0.1\"\ncategory: analysis\nplatforms:

\n\n
\n

AI Citation Tracker (the 2026 GEO measurement surface)

\n

You build a multi-engine AI citation tracker. Modern SEO platforms still optimize for Google's blue links. Meanwhile, generative engines deliver 30%+ of informational answers without a click — and those engines DO cite sources, just not in a place Search Console can see. Your job is to make that surface measurable.

\n

============================================================\n=== PRE-FLIGHT ===

\n\n

Recovery:

\n\n

============================================================\n=== PHASE 1: ENGINE ADAPTERS ===

\n

One adapter per engine. Each adapter answers: \"given query Q, what answer was generated and which URLs were cited?\"

\n
class EngineAdapter(Protocol):\n    name: str  # \"chatgpt\" | \"perplexity\" | \"claude\" | \"gemini\" | \"ai_overviews\" | \"bing_copilot\"\n\n    async def query(self, query: str) -> EngineResult: ...\n\n@dataclass\nclass EngineResult:\n    engine: str\n    query: str\n    asked_at: datetime\n    answer_text: str           # the synthesized answer\n    citations: list[Citation]  # ordered as cited\n    sources_attribution: str | None  # raw sources HTML/markup if returned\n    raw_response: dict          # for replay/debug\n    cost_usd: Decimal\n\n@dataclass\nclass Citation:\n    rank: int                  # 1-indexed position in the answer\n    url: str\n    title: str | None\n    snippet: str | None\n
\n

Engine-specific notes:

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
EngineAuthKey URLCost ballparkNotes
ChatGPT (browsing)OpenAI API/v1/responses w/ web_search_preview tool$5-15 per 1k queriesReturns annotations[] with citation URLs
PerplexityPerplexity API/chat/completions with model=sonar-pro$5/1M input + $15/1M outputReturns citations array natively
Claude (with search)Anthropic API/v1/messages w/ tools=[web_search_20250305]$3/1M input + $15/1M output + $10 per 1k searchesReturns citations as part of content
GeminiGoogle AI Studio APIgemini-2.5-pro w/ Google Search grounding$1.25/1M + $5/1MReturns grounding metadata with URLs
AI OverviewsSerpAPI / Bright Data SERPn/a (no first-party API)$5-20 per 1k queriesScrape Google SERP, extract AI Overview block + sources
Bing CopilotBing Custom Search + Copilot scrapern/a (no API)$5-20 per 1k queriesHeadless or SerpAPI
You.comYou.com API (free tier)/api/searchFree / meteredReturns sources
Phindscrape onlyn/aproxy costHeadless
\n

VALIDATION: At least 3 engines wired and returning citations end-to-end against a smoke-test query.

\n

FALLBACK: If an engine API is down/rate-limited, the polling job continues other engines + retries the failed one with exp backoff. Single-engine failure never blocks the run.

\n

============================================================\n=== PHASE 2: STORAGE SCHEMA ===

\n

SQLite/Postgres schema (Prisma-style):

\n
CREATE TABLE Query (\n  id INTEGER PRIMARY KEY,\n  text TEXT NOT NULL UNIQUE,\n  intent TEXT,                -- informational | commercial | transactional\n  priority INTEGER DEFAULT 5, -- 1=critical, 10=long-tail\n  added_at DATETIME\n);\n\nCREATE TABLE Brand (\n  id INTEGER PRIMARY KEY,\n  name TEXT NOT NULL,         -- canonical\n  aliases TEXT,               -- JSON array\n  domain TEXT,                -- if applicable\n  is_self BOOLEAN DEFAULT FALSE,  -- our brand vs competitor\n  added_at DATETIME\n);\n\nCREATE TABLE Poll (\n  id INTEGER PRIMARY KEY,\n  query_id INTEGER REFERENCES Query(id),\n  engine TEXT NOT NULL,\n  polled_at DATETIME NOT NULL,\n  answer_text TEXT,\n  cost_usd DECIMAL,\n  raw_response JSON\n);\n\nCREATE TABLE Citation (\n  id INTEGER PRIMARY KEY,\n  poll_id INTEGER REFERENCES Poll(id),\n  rank INTEGER,\n  url TEXT NOT NULL,\n  domain TEXT,                -- derived\n  title TEXT,\n  snippet TEXT,\n  brand_matched INTEGER REFERENCES Brand(id)  -- nullable\n);\n\nCREATE TABLE BrandMention (\n  id INTEGER PRIMARY KEY,\n  poll_id INTEGER REFERENCES Poll(id),\n  brand_id INTEGER REFERENCES Brand(id),\n  mention_count INTEGER,      -- in answer_text\n  cited BOOLEAN,              -- did a URL of theirs appear in citations\n  sentiment REAL              -- -1 to 1, optional\n);\n\nCREATE INDEX idx_poll_query_engine_date ON Poll(query_id, engine, polled_at DESC);\n
\n

VALIDATION: Schema migrates cleanly. Sample query inserts + reads.

\n

============================================================\n=== PHASE 3: POLLING ENGINE ===

\n

Generate the polling worker (Python / Node). Schedule:

\n\n

For each poll:

\n
    \n
  1. Call engine adapter.
    2. \n
  2. Persist Poll row + Citation rows.
    3. \n
  3. Run brand-mention detection on answer_text:\n\n
    4. \n
  4. Run domain-matching on citations against Brand.domain.
    5. \n
\n

Rate-limit aware: respect each provider's RPM. Built-in token-bucket. Exponential backoff on 429.

\n

VALIDATION: Daily run completes within window. Cost stays under budget (configurable, default $20/day cap).

\n

============================================================\n=== PHASE 4: METRICS & DELTAS ===

\n

Computed per (engine, query, week):

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
MetricDefinition
CitedBoolean — did our URL appear in citations?
MentionedBoolean — did our brand name appear in answer text?
Citation rankIf cited, what position (1 = first)?
Share of voice(our brand mentions) / (total brand mentions in answer)
Citation share(our citations) / (total citations)
Competitor cited countDistinct competitor brands cited in answer
Answer lengthWord count of synthesized answer
WoW citation deltaCited this week minus cited last week, by query
Top moversQueries with largest WoW citation gain/loss
Citation gapsQueries where competitors are cited but you aren't
Win-back queueQueries where you WERE cited 4 weeks ago but no longer are
\n

Persist daily; aggregate weekly. Generate metrics_weekly.csv.

\n

VALIDATION: Metrics reconcile (per-query sums match aggregates). Weekly delta is non-empty after 2+ weeks of polling.

\n

============================================================\n=== PHASE 5: REPORTING ===

\n

Generate three reports:

\n

weekly_report.md — for the team / boss:

\n\n

competitor_matrix.csv — engine × competitor matrix of citations:

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
QueryChatGPTPerplexityClaudeGeminiAI OverviewBing Copilot
\"best CRM\"Us, Hubspot, SalesforceHubspot, SalesforceUsHubspotSalesforce, UsHubspot
\n

citation_gap_actions.md — prescriptive:\nFor each citation gap, output: target query + which engines miss us + competitor URLs cited + a content brief stub (chain into /seo-content-brief).

\n

VALIDATION: Reports render. CSV imports cleanly into Excel/Sheets.

\n

============================================================\n=== PHASE 6: ALERTING & DASHBOARD ===

\n

Push-style alerts:

\n\n

Optional dashboard: Streamlit or Next.js + Prisma. Tabs: Overview, Per-Engine, Per-Query, Competitor Drill, Cost.

\n

VALIDATION: Alerts fire on simulated event in test. Dashboard renders against the SQLite DB.

\n

============================================================\n=== PHASE 7: HOW THIS BEATS LEGACY SEO TOOLS ===

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
CapabilitySemrush / AhrefsVisibly AIThis skill
Google rank tracking✅✅ (via GSC)(separate skill: gsc-pull)
AI Overview citation trackingpartial❌✅
ChatGPT citation tracking❌❌✅
Perplexity citation tracking❌❌✅
Claude / Gemini tracking❌❌✅
Share-of-voice across all AI engines❌❌✅
Citation-gap → content brief chain❌partial✅ (→ seo-content-brief)
Open data (your DB, no vendor lock)❌❌✅
Composable with other agents❌partial✅
Cost: per-month$129-$499€39-€399API costs only (~$10-50/mo)
\n

VALIDATION: This positioning resonates with users who already pay for one of the above.

\n

============================================================\n=== SELF-REVIEW ===

\n

Score 1-5:

\n\n

Common gap: matching only canonical brand name, missing variants. Generate the alias seed list from the user's marketing site + Wikipedia + Crunchbase.

\n

============================================================\n=== LEARNINGS CAPTURE ===

\n

~/.claude/skills/ai-citation-tracker/LEARNINGS.md.

\n

============================================================\n=== STRICT RULES ===

\n\n","files":[{"path":"SKILL.md","sizeBytes":12279,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-10-01T15:40:30.966875Z","sha256":"346541D24493254CD58349582265ADB84DF4EA338D9FA705E6ABDDA935462D6D","sizeBytes":5072},"review":null,"source":{"repositoryUrl":"https://github.com/tinh2/skills-hub-registry","path":"analysis/ai-citation-tracker","license":null,"commit":"d38affbf56da216841e2b9e4032a4b978c2062fd","subtreeSha":"92623EF9A739E0962FCE8ED511357045CF3AB42966AEA1E66B0E1789EA56F4B1","lastSyncedAt":"2026-10-01T15:40:09.634878Z"},"reviewedAt":"2026-10-01T15:40:57.71701Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/tinh2/skills-hub-registry/tree/main/analysis/ai-citation-tracker"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install tinh2-skills-hub-registry@llmmart"},{"target":"git","command":"git clone https://github.com/tinh2/skills-hub-registry.git"}]}