ay-automate/gtmos-skills

107 skill entries

• ads

  Use when reading Google Ads or Meta Ads campaign performance, or proposing a supported campaign update through GTM OS.

  Needs a keyInstructions only633 words
• cal-com

  Use when processing a Cal.com booking webhook as a GTM discovery-call signal.

  Reaches the webInstructions only607 words
• composio

  Use when the GTM engineer wants GTM OS to read from or act in THEIR OWN apps -- their CRM (HubSpot, Salesforce, Pipedrive, Attio, Zoho), product analytics (PostHog, Mixpanel, Amplitude), call recorders (Fireflies, Gong, Zoom, Fathom), ad platforms (Google, Meta, LinkedIn, TikTok), outbound rails (Smartlead, Instantly, lemlist), Slack, Notion, support or billing tools -- through Composio. Covers installing and logging in to the Composio CLI (macOS/Linux/Windows), what to automate for this GTM setup, the GTM OS path (Integrations > Connections > Composio -> connect app -> CRM of record -> sync; write-back via the approval-gated crm.update_record), errors and references.

  Needs a keyActs undeclared2,256 words
• ga4

  Use when querying Google Analytics 4 reports for GTM OS attribution and landing-page analysis.

  Reaches the webInstructions only632 words
• gmail

  Use when ingesting Gmail replies, reading supported email replies, or recording Gmail email activity through GTM OS.

  Needs a keyInstructions only666 words
• google-calendar

  Use when reading supported Google Calendar events or auto-logging meetings with known GTM OS contacts.

  Reaches the webInstructions only642 words
• gsc

  Use when reading Search Console analytics or submitting a sitemap through the GTM OS SEO connector.

  Needs a keyInstructions only615 words
• gtmos-ads-campaigns

  Plan, audit and optimize paid campaigns on Google Ads, Meta and LinkedIn -- structure, targeting, bidding, budgets, pacing and when to scale or kill -- against real unit economics. With GTM OS connected it uses the Performance Ads engine: perf_ads.sync_performance, get_campaign_performance, get_pacing_recommendation, scan_all_campaigns and kpis, spend ceilings from definition.unit_economics, revenue from money.attribution, and every budget or status change proposed through perf_ads.update_campaign for a person to approve. Use for \"paid ads\", \"PPC\", \"Google Ads\", \"Meta ads\", \"LinkedIn ads\", \"ROAS\", \"CAC\", \"ad budget\", \"pacing\", \"campaign audit\", \"scale campaigns\".

  Reaches the webInstructions only4,591 words
• gtmos-ads-creative

  Generate and iterate ad creative -- hooks, headlines, primary text, descriptions, static and video concepts and test matrices -- for any ad platform, with every claim backed by approved proof. With GTM OS connected it reads performance from the Performance Ads engine to see what wins, researches competitor ads through the looot Meta ads library, uses customer language and active proof, and checks copy with brand.check_text. Use for \"ad creative\", \"ad copy\", \"headlines\", \"hooks\", \"creative testing\", \"ad variations\", \"creative fatigue\", \"UGC scripts\", \"ad concepts\".

  Needs a keyActs undeclared3,710 words
• gtmos-analytics-attribution

  Work out which marketing and sales efforts drove conversions and revenue -- choose and explain an attribution model, compare models, reconcile conflicting numbers across ad platforms, GA4 and the CRM, design incrementality checks and self-reported attribution, and build first-party attribution. With GTM OS connected it reads the org's revenue truth, window and model (money.settings.get), splits real revenue across engines, campaigns, channels and plays with money.attribution (equal split, position-based, first touch, last touch, time decay, or all side by side), reconciles truth against CRM won deals (money.reconcile) and labels every split as a model, not a cause. Use for "attribution", "which channel drove this", "first touch vs last touch", "multi-touch", "MMM", "incrementality", "why don't my numbers match", "Google says 50, Meta says 40", "dark social", "how did you hear about us", "ROAS on real revenue". For setting up tracking see gtmos-analytics-tracking.

  Needs a keyActs undeclared3,848 words
• gtmos-analytics-experiments

  Design A/B tests and growth experiments that hold up: hypothesis, primary metric, sample size and duration, guardrails, and how to read the result. With GTM OS connected it sizes tests from real baselines (metric.evaluate, the engine KPIs), simulates play-based tests with play.simulate before anything runs, and reports results with their sample sizes. GTM OS has no experiment engine yet, so the skill says so. Use for \"A/B test\", \"split test\", \"experiment\", \"test this headline\", \"sample size\", \"statistical significance\", \"growth experiments\", \"holdout\".

  Reaches the webInstructions only1,999 words
• gtmos-analytics-gtm-report

  Build the weekly GTM report, pipeline pack or campaign post-mortem where every figure names the operation or system it came from, small samples are flagged, and data gaps are stated. With GTM OS connected it reads query.engine_kpis, crm.dashboard, money.attribution and money.revenue, each engine's KPIs, sync.health and sync.completeness, and delivers the report as a document or through channel.deliver for approval. Use for \"weekly report\", \"GTM report\", \"pipeline report\", \"sales report\", \"marketing report\", \"board pack\", \"campaign post-mortem\", \"how did we do this week\".

  Self-containedActs undeclared1,290 words
• gtmos-analytics-tracking

  Plan, set up, verify and debug analytics and conversion tracking -- tracking plans, event naming, UTMs, GA4 and Google Tag Manager, form and booking conversions, ad click ids, offline conversions and AI-assistant traffic. With GTM OS connected it installs and verifies the GTM OS website tracker (tracking.site.create, tracking.verify), reads traffic sources and per-account activity (tracking.sources, tracking.activity), routes form leads through the webform connector and reads GA4 and Search Console through their connectors, so web visits become touches the attribution model can credit. Use for "set up tracking", "tracking plan", "GA4", "Google Tag Manager", "UTM parameters", "conversion tracking", "my conversions aren't showing", "Meta pixel", "CAPI", "enhanced conversions", "gclid", "offline conversions", "track form submissions", "AI traffic from ChatGPT". For attribution models see gtmos-analytics-attribution.

  Reaches the webActs undeclared2,615 words
• gtmos-content-brand-voice

  Codify the brand voice -- tone, preferred and avoided words, example phrases, voice samples and forbidden claims -- as a signed GTM OS definition, then check any copy against it. With GTM OS connected it reads the current voice (brand.get), drafts a definition.brand_voice from real samples, proposes it for a human to sign, and gates every draft with brand.check_text. Use for \"brand voice\", \"tone of voice\", \"brand guidelines\", \"style guide\", \"words to avoid\", \"forbidden claims\", \"does this sound like us\", \"check this copy against our brand\".

  Self-containedInstructions only380 words
• gtmos-content-claims-gate

  Check every claim in a piece of copy -- emails, sequence steps, LinkedIn messages, ads, pages, scripts, one-pagers -- against approved proof: keep, rewrite or cut each line, and list the proof that would bring a cut line back. With GTM OS connected it reads the proof library (proof.list, usable items only) and the org's claims policy (policy.claims.get), mirrors the claims gate every send and publish passes, and turns missing proof into a draft proof source a person activates. Use for \"check the claims\", \"can we say this\", \"is this claim backed\", \"fact-check this copy\", \"why was my send blocked\", \"claims gate\".

  Self-containedInstructions only1,376 words
• gtmos-content-copywriting

  Write, rewrite and edit marketing copy for homepages, landing pages, pricing, feature and about pages, in the brand voice and with every claim backed by approved proof. With GTM OS connected it reads the signed brand voice (brand.get), customer language and active proof, checks drafts with brand.check_text, and proposes page changes through the SEO & GEO engine for approval. Use for \"write copy\", \"rewrite this page\", \"landing page copy\", \"homepage copy\", \"headline\", \"value proposition\", \"edit my copy\", \"proofread\", \"make this clearer\", \"copy review\".

  Reaches the webInstructions only2,477 words
• gtmos-content-social

  Plan and write social content for LinkedIn, X and other networks -- posts, hooks, threads, carousels and a schedule -- plus social listening and engagement triage, in the brand voice. With GTM OS connected it uses customer language, approved proof and content performance, checks posts with brand.check_text, and treats LinkedIn messages through Unipile as human-only sends. Use for \"LinkedIn post\", \"social media content\", \"content calendar for LinkedIn\", \"write a thread\", \"hooks\", \"social listening\", \"engagement\", \"repurpose this into posts\".

  Reaches the webActs undeclared2,824 words
• gtmos-content-strategy

  Decide what content to create and why: pillars, topics, formats and a calendar tied to search demand, buyer questions and the funnel. With GTM OS connected it starts from Search Console opportunities and the refresh calendar on the SEO & GEO engine, buyer language from calls and replies, and AI-answer visibility, and turns the plan into tracked pages and briefs. Use for \"content strategy\", \"what should I write about\", \"content plan\", \"editorial calendar\", \"content pillars\", \"topic ideas\", \"blog strategy\".

  Self-containedInstructions only2,966 words
• gtmos-crm-data-quality

  Find and fix CRM data problems -- duplicates, invalid emails and domains, missing fields, inconsistent formats, stale records -- with a reviewable digest before any change, and reversible merges. With GTM OS connected it uses the Identity and data quality engine: identity_dq.quality_report, scan_duplicates, validate_records, normalize_records (with undo), and merges proposed through identity_dq.merge_records for a person to approve. Use for \"CRM cleanup\", \"dedupe\", \"duplicates\", \"data quality\", \"bad emails\", \"normalize records\", \"merge companies\", \"data hygiene\".

  Needs a keyActs undeclared1,457 words
• gtmos-crm-email-hygiene

  Clean an email list before it is sent to: remove duplicates and bad syntax for free, verify the rest, and split it into send, review and drop. With GTM OS connected it runs the free checks in the Identity and data quality engine (identity_dq.validate_records, scan_duplicates), re-checks stale verified emails with enrichment.refresh_stale (estimate first), respects suppression and consent, and reports from enrichment.freshness. Use for \"clean this list\", \"verify these emails\", \"reduce bounces\", \"list hygiene before a campaign\", \"which emails are stale\".

  Reaches the webInstructions only962 words
• gtmos-crm-forecast

  Build the forecast narrative for a rep, team or company: the number by bucket (closed won, commit, best case, pipeline), what changed since last week, risks to commit, and the asks -- every figure from real deal data. With GTM OS connected the number comes from records.aggregate over gtm.deal (exact, not a page), commentary from a capped sample, revenue truth from money.settings.get, and changes are proposed for approval. Use for \"forecast\", \"commit number\", \"forecast call\", \"will we hit the quarter\", \"best case\", \"forecast narrative\", \"quarter outlook\".

  Self-containedInstructions only1,618 words
• gtmos-crm-pipeline-review

  Review the pipeline deal by deal -- stuck stages, no next step, slipped close dates, single-threaded deals, missing amounts -- and turn findings into tasks and proposed updates. With GTM OS connected it reads crm.dashboard, open deals with records.list (closing overdue or this month, activity windows), relationship history and the committee, and proposes stage moves and CRM write-backs for approval. Use for \"pipeline review\", \"pipeline pulse\", \"deal review\", \"which deals are stuck\", \"deal risks\", \"weekly pipeline meeting\", \"clean up my pipeline\".

  Reaches the webActs undeclared1,120 words
• gtmos-engine-catalog

  The GTM OS engine catalog -- for each engine its purpose, the connections it needs, the definitions it reads, the built-in defaults, the customisation knobs, the live verification step and the gotchas. Read before explaining, building or verifying any engine (Score, Enrichment, Signals x Intent, Outreach, SEO & GEO, Performance Ads, Identity & data quality) and for routing, attribution, agent packs, plays, research and CRM sync.

  Self-containedInstructions only2,720 words
• gtmos-leads-buying-committee

  Map the buying committee at a target account (economic buyer, champion, user, technical evaluator, or the org's own roles), see who is missing, and add verified contacts for the gaps. With GTM OS connected it reads score.committee_coverage and buying_roles.list, sources people with find.people through the governed broker (quoted first), creates and links the people to the open deal (canon.edit, approved by a person when an agent proposes it) and fills emails with enrichment.enrich_contacts (estimate first). Use for \"who should I contact at acme.com\", \"map the buying committee\", \"multi-thread this deal\", \"find the decision makers\", \"who signs off on this\", \"we are single-threaded\".

  Reaches the webInstructions only1,113 words
• gtmos-leads-icp-lookalikes

  Build a target account list from the customers you already win with: enrich the seeds, find similar companies, drop what fails the ICP or is already in the CRM, and rank the rest. With GTM OS connected it takes seeds from won deals, quotes and runs find.lookalikes and enrich.company through the governed broker, dedupes against existing accounts, and ranks with the Score engine's signed ICP-fit model instead of a hand-made score. Use for \"find companies like our top customers\", \"lookalike accounts\", \"build an ICP list from these seeds\", \"expand my account list\".

  Reaches the webInstructions only1,035 words
• gtmos-leads-inbound-triage

  Triage new inbound leads (demo requests, form fills, signups, referrals) by ICP fit and intent, give each a priority (P0, P1, P2 or DQ), route it to an owner and set the first touch and response SLA. With GTM OS connected it scores through the Score engine (score.score_contact, score.explain_subject), reads the lead's signals, customer type and existing relationship, previews and makes the assignment with territory.match and routing.assign, and opens a follow-up task; it never sends anything itself. Use for "triage these leads", "new inbound leads", "who should take this lead", "is this lead any good", "prioritize the lead queue", "speed to lead", "route inbound", "MQL review". For the ownership audit across the whole book see gtmos-routing-lead-ownership.

  Reaches the webInstructions only1,511 words
• gtmos-leads-prospecting

  Build a verified, scored prospect list from an ICP across B2B SaaS, general B2B, local small business and early demand-signal motions. With GTM OS connected it reads the org's personas, segments and scoring model, quotes the cost first (capability.quote, enrichment estimate), sources companies and people through the governed looot capabilities (find.companies, find.people, find.lookalikes), fills gaps with the empty-only enrichment waterfall, skips suppressed and existing records, and ranks the result with score.list_ranked. Use for "build a prospect list", "find leads", "lead list", "who should we target", "lookalike accounts", "TAM list", "find decision makers", "local business leads", "first customers", "design partners". For writing the outreach see gtmos-outreach-cold-email; for the ICP model itself see gtmos-score-icp.

  Needs a keyInstructions only3,134 words
• gtmos-leads-tech-stack

  Read the tech stack of every account on a list and sort them into segments: uses a competitor (displace), uses a complement (expand), uses a disqualifier, or none detected. With GTM OS connected it fills technographics through the governed enrichment waterfall (estimate first), reads tech_adoption and tech_churn signals, and drafts a definition.segment for the result. Use for \"which of these accounts use Intercom\", \"find Salesforce shops\", \"segment my list by tech stack\", \"technographic targeting\", \"who runs HubSpot\".

  Reaches the webInstructions only819 words
• gtmos-lifecycle-churn

  Reduce churn: cancel flows and save offers, failed-payment recovery, health scoring and early-warning signals, and win-back. With GTM OS connected it reads product usage and engagement signals (product_analytics.sync, signals.list, journey.position), finds at-risk accounts from real data, and runs saves and win-backs through approval-gated outreach and plays. Use for \"churn\", \"reduce churn\", \"cancel flow\", \"save offer\", \"retention\", \"failed payments\", \"dunning\", \"health score\", \"win-back\", \"customers leaving\".

  Reaches the webInstructions only2,956 words
• gtmos-lifecycle-email

  Plan and write lifecycle email -- welcome, onboarding, nurture, trial conversion, re-engagement and win-back sequences -- with consent, suppression and sending rules respected. With GTM OS connected it builds sequences on the Outreach engine (sequence.create, sequence.enroll) on an active lane, checks consent and suppression, gates copy with brand.check_text and approved proof, and every send is proposed for a person to approve. Use for \"email sequence\", \"drip campaign\", \"nurture sequence\", \"onboarding emails\", \"welcome series\", \"lifecycle email\", \"re-engagement\", \"trial emails\".

  Needs a keyActs undeclared1,828 words
• gtmos-lifecycle-renewal-radar

  Watch the renewal runway -- accounts renewing in the next 30, 60 and 90 days, their health, usage, champion status and open risks -- and propose save or expand plays early. With GTM OS connected it reads renewal-dated deals and accounts (records.list), product usage signals (product_analytics.sync, signals.list), journey position and relationship history, and turns risks into tasks and drafts a person approves. Use for \"renewal radar\", \"upcoming renewals\", \"at-risk renewals\", \"renewal pipeline\", \"who is up for renewal\", \"expansion opportunities\", \"account health before renewal\".

  Self-containedInstructions only2,558 words
• gtmos-looot

  Read FIRST for any looot work inside GTM OS. Use whenever a GTM task needs live outside data -- find or verify an email or phone, enrich a company or person, build a company or people list, find lookalikes, search the web or news, pull job postings, social posts, technographics, headcount, SEO keyword data, or extract a web page. looot is GTM OS's data-acquisition layer (one prepaid account, ~80 providers). Read this before calling any looot capability, the looot MCP server or the looot CLI; it decides when the upstream looot-* skills apply.

  Needs a keyInstructions only2,089 words
• gtmos-marketing-cro

  Raise conversion on marketing pages and forms -- homepage, landing, pricing, feature and demo-request pages -- with a structured audit, prioritized fixes and test ideas, measured with real tracking data. With GTM OS connected it reads site tracking and sources (tracking.activity, tracking.sources), inbound form performance, and proposes page changes through the SEO & GEO engine for approval. Use for \"CRO\", \"conversion rate\", \"landing page optimization\", \"this page isn't converting\", \"improve signups\", \"form conversion\", \"demo requests are low\".

  Self-containedInstructions only1,099 words
• gtmos-marketing-launch

  Plan a product or feature launch: positioning, audiences, channels, timeline, assets and the measurement plan, with every customer-facing send approved by a person. With GTM OS connected it targets segments and personas from the org's definitions, drafts announcement emails as Outreach-engine sequences, updates pages through the SEO & GEO engine, and measures with engine KPIs. Use for \"launch plan\", \"product launch\", \"feature announcement\", \"Product Hunt\", \"release plan\", \"go-to-market for a feature\", \"launch checklist\".

  Self-containedInstructions only2,437 words
• gtmos-marketing-lead-magnets

  Plan and build lead magnets -- guides, templates, tools, calculators, webinars -- that capture the right leads, and wire capture into forms, tracking and follow-up. With GTM OS connected it captures through the webform connector and site tracking, turns downloads into signals and scored contacts, and follows up through approval-gated Outreach sequences. Use for \"lead magnet\", \"gated content\", \"content upgrade\", \"downloadable\", \"ebook\", \"template offer\", \"calculator\", \"lead capture\", \"grow the email list\".

  Reaches the webInstructions only1,853 words
• gtmos-marketing-plan

  Build a 12-month marketing and GTM plan structured by AARRR (acquisition, activation, retention, referral, revenue), with a scored current-state audit, a 90-day roadmap with owners, budget math, funding-stage unlocks and the skills and tools that execute each move. With GTM OS connected it grounds the current state in real numbers (crm.dashboard, money.revenue, money.attribution, journey.funnel, query.engine_kpis, engine KPIs), reads the org's ICP, unit economics and brand voice from its definitions, maps moves to GTM OS engines and plays, and can shape the setup 90-day plan with onboarding.plan.set. Use for "marketing plan", "growth plan", "GTM plan", "go-to-market plan", "fCMO plan", "AARRR plan", "90-day roadmap", "12-month marketing roadmap".

  Needs a keyInstructions only4,218 words
• gtmos-marketing-positioning

  Capture and maintain positioning and product marketing context -- product, ICP, personas, pains, competitors, differentiation, objections, customer language, brand voice and proof -- so every other marketing, sales and content skill starts from it. With GTM OS connected it reads the org's signed definitions and setup cards first, drafts changes to definition.brand_voice, definition.persona, definition.segment and definition.customer_type as a ChangeSet that a human signs, and takes proof only from the proof library. Without GTM OS it writes .agents/product-marketing.md. Use for "positioning", "messaging", "product marketing context", "set up context", "who is my target customer", "ICP", "value prop", "differentiation", "brand voice", "update our positioning".

  Self-containedInstructions only2,613 words
• gtmos-onboarding

  Guide a user through the per-organization GTM OS setup wizard over MCP (gtm_describe / gtm_query / gtm_act) -- Welcome, Connect your data, Your definitions, Rules and who approves what, The build, First motion and 90 days -- citing the evidence behind every draft, never inventing a value, and stopping for the human-only steps and the four-eyes approvals. Use when the user wants to set up (or finish setting up) an organization in a running GTM OS instance; run gtmos-setup first if no instance is connected.

  Self-containedInstructions only7,620 words
• gtmos-ops-marketing-loops

  Turn recurring GTM work into loops that run on a schedule or a trigger -- signal sweeps, reply triage, content refresh, pacing checks, weekly reports -- as GTM OS plays and agent triggers that propose work and wait for a person. With GTM OS connected it builds definition.play automations (event, schedule, segment, metric or webhook triggers), simulates them with play.simulate before anything runs, and lists the approvals they create. Use for \"marketing loop\", \"automate this weekly\", \"recurring workflow\", \"run this every day\", \"agent on a schedule\", \"trigger when\", \"always-on outbound\", \"play\".

  Self-containedInstructions only1,925 words
• gtmos-outreach-cold-email

  Write and run B2B cold outbound that gets replies -- first-touch emails, follow-ups, multichannel cadences with LinkedIn and calls, deliverability, and reply handling. With GTM OS connected it drafts through the Outreach engine (outreach.draft_personalized_message), grounds every claim in the approved proof library, checks copy with brand.check_text and proposes sends for a human to approve; dry mode ends at the draft. Use for "cold email", "outbound sequence", "SDR emails", "follow-up sequence", "nobody's replying", "deliverability", "sending domains", "warm up inboxes", "LinkedIn outreach", "multichannel cadence", "handle replies", "AI SDR". For building the list see gtmos-leads-prospecting; for lifecycle email see gtmos-lifecycle-email.

  Self-containedInstructions only2,270 words
• gtmos-outreach-first-lines

  Write one opening line per lead that cites a real, recent, specific fact -- a post they wrote, a launch, a hire, a round -- with its source, or NO_FACT when there is none. With GTM OS connected it takes facts from verified signals and the research agent first, buys only what is missing (quoted research.person_posts and research.news), drafts through outreach.draft_personalized_message, and keeps every claim inside the claims gate. Use for \"personalize these openers\", \"first lines for my list\", \"icebreakers\", \"personalized intro line\", \"make this email less generic\".

  Reaches the webInstructions only902 words
• gtmos-outreach-linkedin

  Write LinkedIn connection notes, first messages after an accept, and the two follow-ups, and run them on the org's LinkedIn seats. With GTM OS connected it checks the seats and the owner's high-risk acceptance (outreach.linkedin_seats), drafts through the Outreach engine, proposes every send for a person to approve inside the seat's caps and warm-up, and uses connect_then_message steps in a definition.sequence. Use for \"write a connection request\", \"LinkedIn DM\", \"LinkedIn follow-up\", \"add LinkedIn to the sequence\", \"why are my LinkedIn sends held\".

  Self-containedInstructions only923 words
• gtmos-outreach-sequences

  Design a multi-step outbound or nurture sequence: structure, waits, send windows, A/B variants, LinkedIn steps, exits, and how each step is measured. With GTM OS connected it writes the steps as the org's definition.sequence (signed by a person before it runs), binds sequences to it with stepsKey, and reads outreach.step_stats and outreach.learning_report to keep or kill steps and variants. Use for \"build a sequence\", \"design a cadence\", \"A/B test step 2\", \"add a LinkedIn step\", \"which step gets replies\", \"follow-up timing\".

  Self-containedInstructions only1,509 words
• gtmos-research-account

  Write a one-page account research brief -- what the company does, recent triggers, tech and hiring signals, the buying committee, our history with them and a suggested angle -- where every fact cites its source. With GTM OS connected it queues the research agent (research.request), reads signals, records, relationship history and the committee model, and leaves the research agent's proposed facts for a person to accept or dismiss (research.fact.decide is human only). Use for \"research this account\", \"account brief\", \"company research\", \"tell me about <company> before outreach\", \"account plan prep\", \"who should we talk to at <company>\".

  Reaches the webInstructions only1,848 words
• gtmos-research-competitor

  Profile competitors and build battle cards -- positioning, pricing, features, reviews, hiring, ads and messaging -- where every claim carries a source and a date. With GTM OS connected it runs research through the research agent and the looot capabilities (research.request, research.web_search, research.news, research.reviews, research.extract_page), leaves the research agent's proposed facts for a person to accept or dismiss (research.fact.decide is human only), and grounds our side of the comparison in approved proof only. Use for \"competitor research\", \"competitor profile\", \"battle card\", \"competitive analysis\", \"how do we compare to <competitor>\", \"competitor pricing\", \"win against <competitor>\".

  Self-containedActs undeclared2,667 words
• gtmos-research-customer

  Run and synthesize customer research -- interviews, call transcripts, reviews, surveys, support tickets and won or lost deal notes -- into pains, outcomes, objections, triggers and the exact words customers use. With GTM OS connected it reads call transcripts and classified replies, mines and approves customer phrases (voice.customer_language.*) so they ground outreach and copy, and keeps every quote linked to its source. Use for \"customer research\", \"ICP research\", \"analyze transcripts\", \"voice of customer\", \"interview synthesis\", \"jobs to be done\", \"why do customers buy\", \"customer language\".

  Reaches the webInstructions only2,395 words
• gtmos-revops-lifecycle

  Design and run the revenue lifecycle -- lead stages, MQL/SQL definitions, scoring, routing, handoffs, SLAs and pipeline stages -- and keep marketing and sales on one model. With GTM OS connected the lifecycle lives in signed definitions (definition.journey, stage_map, stage_machine, deal_taxonomy, scoring_model); it reads funnels with journey.funnel, finds unmapped CRM stages with stage_map.unmapped, and proposes changes as a ChangeSet a human signs. Use for \"RevOps\", \"lead lifecycle\", \"MQL\", \"SQL\", \"lead scoring\", \"lead routing\", \"handoff\", \"funnel stages\", \"pipeline stages\", \"SLA between marketing and sales\".

  Reaches the webInstructions only2,424 words
• gtmos-routing-lead-ownership

  Find the authoritative owner of a lead or account, audit whether high-intent leads got timely human follow-up ("no lead left behind"), and fix ownership gaps -- unowned records, owners who left, territory mismatches, bulk reassignments. With GTM OS connected it reads owners and territories, previews routing with territory.match, assigns with routing.assign (an auditable decision record per change), sweeps unowned or orphaned records with routing.reconcile, edits territories with territory.upsert (human only) and opens follow-up tasks; CRM-of-record owner changes are proposed through crm.update_record for approval. Use for "who owns this lead", "find the account owner", "unowned leads", "lead routing", "round robin", "territories", "reassign accounts", "rep left, move their book", "speed to lead audit", "leads nobody followed up". For triaging single new leads see gtmos-leads-inbound-triage.

  Self-containedActs undeclared1,166 words
• gtmos-sales-call-prep

  Prepare a one-page brief before a sales call -- who they are, where the deal stands, what was said last, open blockers, fresh company news, buying-committee gaps and three talking points with a next step. With GTM OS connected it reads the meeting, CRM record, relationship history, call transcripts, verified signals, score reasons and the research agent's brief, cites only approved proof, and saves the brief as a timeline note or delivers it to Slack for approval. Use for "prep me for my call", "call prep", "meeting prep", "brief me on [name]", "who am I meeting", "get me ready for the demo", "QBR prep", "what do I need to know before talking to [company]". For the follow-up after the call see gtmos-sales-follow-up.

  Self-containedInstructions only1,744 words
• gtmos-sales-enablement

  Create sales collateral that reps actually use -- pitch decks, one-pagers, battle cards, demo scripts, call and voicemail scripts, objection handling -- built only on approved proof and the customer's own words. With GTM OS connected it pulls active proof and customer language (proof.list, voice.customer_language.list), real objections from replies and call transcripts, checks copy with brand.check_text and renders documents through the output-docgen connector. Use for \"sales deck\", \"pitch deck\", \"one-pager\", \"objection handling\", \"demo script\", \"cold call script\", \"battle card\", \"sales collateral\", \"leave-behind\".

  Self-containedInstructions only3,412 words
• gtmos-sales-follow-up

  Draft the follow-up after a sales call or meeting -- pick the goal from what was promised on the call, write a short specific email with one clear ask, and line up the next step and task. With GTM OS connected it reads the call transcript's action items, the deal and the email history, drafts through the Outreach engine, checks the copy against the brand voice and approved proof, logs the next step as a task, and proposes the send and any deal-stage or CRM change for a human to approve. Use for "write my follow-up", "follow up with [name]", "recap email after the call", "send the next steps", "post-meeting email", "what did I promise them", "nudge [name] on the proposal". For preparing the call itself see gtmos-sales-call-prep.

  Self-containedInstructions only1,827 words
• gtmos-sales-win-loss

  Review recently closed deals for patterns a sales leader can act on -- win rate by rep, source, size, industry and deal type, where losses die, cycle length, loss reasons and what wins have in common, each backed by evidence. With GTM OS connected it counts closed deals with records.aggregate, reads each deal's stage history, loss reason, call transcripts and replies, groups revenue by the org's own deal types, and says "not enough data yet" when the sample is too small to call a pattern. Use for "win-loss review", "why are we losing deals", "loss reasons", "what do our wins have in common", "win rate by rep", "where do deals die", "closed-lost analysis", "quarterly deal review". For live pipeline risk see gtmos-crm-pipeline-review.

  Self-containedInstructions only1,630 words
• gtmos-score-icp

  Turn an ICP into a visible, tunable account scoring model and tier a book of accounts with it -- weighted dimensions, renormalized over what was actually observed, a minimum-observability gate, tier cuts the user can move, and evidence for every judgment. With GTM OS connected it reads the signed definition.scoring_model (icp_fit for firmographic fit, default for intent, signals for signal strength), drafts changes as a ChangeSet for a human to sign, and checks the result with score.account_tiers, score.explain_subject and score.kpis. Use for "define our ICP", "ICP scoring", "account tiering", "tier our accounts", "lead scoring model", "fit score", "which accounts are tier 1", "tune the scoring weights", "why is this account tier 2", "is the score working". For building a list to score see gtmos-leads-prospecting.

  Self-containedActs undeclared3,687 words
• gtmos-seo-audit

  Run a full-site SEO audit -- crawlability, indexability, canonicals, Core Web Vitals, on-page, schema, content quality and AI-search readiness -- and return a scored, prioritized action plan. With GTM OS connected it runs on the SEO & GEO engine: tracked pages, seo_geo.run_audit / scan_all_pages, the fix queue and Search Console data, and every fix is proposed for a person to approve (seo_geo.publish_content_change). Use for \"SEO audit\", \"technical SEO\", \"site audit\", \"why am I not ranking\", \"crawl issues\", \"indexing problems\", \"Core Web Vitals\", \"SEO health check\". For one URL use gtmos-seo-page; for AI answers use gtmos-seo-geo.

  Reaches the webActs undeclared1,565 words
• gtmos-seo-backlinks

  Audit a backlink profile -- referring domains, anchors, toxic links, competitor gaps -- and plan link building that lifts rankings and AI citations without spam. With GTM OS connected it reads authority and referring domains from seo_geo.run_visibility (Ahrefs through Composio), backlink data from the looot capabilities, and turns outreach to linking sites into Outreach-engine drafts that a person approves. Use for \"backlinks\", \"link building\", \"referring domains\", \"toxic links\", \"disavow\", \"link gap\", \"digital PR for links\", \"domain rating\".

  Reaches the webInstructions only2,457 words
• gtmos-seo-competitor-pages

  Build competitor comparison and alternative pages that rank and survive the competitor reading them: \"[Competitor] alternative\", \"alternatives to [Competitor]\", \"you vs [Competitor]\", \"[A] vs [B]\", feature and pricing tables, migration sections. With GTM OS connected it gathers dated, sourced competitor facts through the governed looot capabilities (research.extract_page, research.web_search, seo.serp_organic), takes every claim about your own product from the approved proof library, gates copy with brand.check_text, tracks the page in the SEO & GEO engine and proposes changes for a human to approve. Use for \"alternative page\", \"vs page\", \"competitor comparison\", \"comparison page\", \"[product] vs [product]\", \"switch from\", \"competitive landing pages\". For internal battle cards see gtmos-research-competitor.

  Self-containedInstructions only2,189 words
• gtmos-seo-competitor-serp-watch

  Track where your domain and your competitors appear in Google's top 10 and in AI answers (Google AI Mode, ChatGPT, Perplexity, Gemini) for buying keywords, and what changed since the last check. With GTM OS connected it adds the keywords as tracked prompts (seo_geo.create_tracked_prompt), queues the measured run with seo_geo.run_visibility under the per-run spend cap, and reads share of voice and drops from seo_geo.get_visibility and seo_geo.list_visibility_alerts. Use for \"competitor rankings\", \"SERP watch\", \"who ranks for our keywords\", \"AI answer share of voice\", \"did we lose positions\".

  Reaches the webInstructions only971 words
• gtmos-seo-content-brief

  Write a competitive SEO content brief for a new page or a rewrite -- search intent, outline with word counts per section, keyword placement, questions to answer, competitor gaps, internal links, schema and proof to cite. With GTM OS connected it sizes the opportunity from seo_geo.keyword_opportunities, checks the page's audit and performance, pulls approved proof and customer language, and hands the finished draft to the publish queue for approval. Use for \"content brief\", \"write a brief\", \"outline this article\", \"brief for a writer\", \"improve this page's content\".

  Reaches the webInstructions only2,050 words
• gtmos-seo-drift

  Catch SEO regressions before rankings drop: baseline the SEO-critical elements of key pages (titles, meta, canonicals, robots, headings, schema, status codes) and compare them after every release. With GTM OS connected the baseline is the engine's stored audit per tracked page: seo_geo.scan_all_pages re-audits on a schedule, seo_geo.fix_queue and visibility alerts surface what changed, and reverts go through approval. Use for \"SEO drift\", \"did the release break SEO\", \"track SEO changes\", \"baseline\", \"regression\", \"titles changed\".

  Self-containedActs undeclared1,313 words
• gtmos-seo-engine

  Start here for any SEO or GEO request. Routes the job to the right GTM OS SEO skill (audit, single page, GEO and AI citations, keywords, clusters, briefs, schema, programmatic pages, competitor pages, backlinks, drift, Search Console, site architecture, strategy), checks that the SEO & GEO engine is ready (Google Search Console connected and a property selected), and holds the shared scoring, quality gates and synthesis method every SEO skill uses. With GTM OS connected it reads seo_geo.kpis, seo_geo.fix_queue and instance.readiness first, and every content change is proposed with seo_geo.publish_content_change for a different person to approve. Use for "SEO", "help with SEO", "why am I not ranking", "SEO health score", "AI search", "GEO", "organic traffic", "search visibility", or any SEO question that does not name a narrower job.

  Needs a keyActs undeclared1,786 words
• gtmos-seo-geo

  Get cited in AI answers -- ChatGPT, Perplexity, Gemini, Copilot and Google AI Overviews -- by improving citability, entity clarity, crawler access and passage structure, and track AI share of voice over time. With GTM OS connected it uses the SEO & GEO engine: tracked AI prompts (seo_geo.create_tracked_prompt), seo_geo.get_visibility, visibility alerts, GEO readiness from seo_geo.run_audit, and the looot geo.answer_* capabilities, with fixes proposed for approval. Use for \"AI SEO\", \"GEO\", \"AEO\", \"LLM visibility\", \"get cited by ChatGPT\", \"AI Overviews\", \"AI share of voice\", \"why don't AI tools mention us\".

  Reaches the webActs undeclared5,431 words
• gtmos-seo-keyword-clusters

  Cluster keywords by SERP overlap into hub-and-spoke topics, assign one page per cluster, and plan the internal links between them. With GTM OS connected it pulls keywords from Search Console and looot (seo_geo.keyword_opportunities, seo.serp_organic), tracks each hub and spoke on the SEO & GEO engine and feeds the content calendar. Use for \"keyword clustering\", \"topic clusters\", \"pillar pages\", \"hub and spoke\", \"content architecture\", \"group these keywords\", \"cannibalization\".

  Self-containedActs undeclared2,400 words
• gtmos-seo-keyword-research

  Find and prioritize keywords by intent, volume, difficulty and business value, starting from the queries you already almost rank for. With GTM OS connected it reads Search Console striking-distance queries joined with looot and Ahrefs keyword data (seo_geo.keyword_opportunities), refreshes keyword data within the org's budget (seo_geo.refresh_keyword_data) and turns winners into tracked pages and prompts. Use for \"keyword research\", \"what should we rank for\", \"find keywords\", \"search volume\", \"keyword difficulty\", \"low-hanging keywords\", \"striking distance\", \"keyword gap\".

  Reaches the webInstructions only3,385 words
• gtmos-seo-page

  Audit one URL in depth -- title, meta, headings, content quality and E-E-A-T, internal links, images, schema, performance and AI citation readiness -- and propose concrete fixes. With GTM OS connected it tracks the page on the SEO & GEO engine, runs seo_geo.run_audit, reads its Search Console performance and proposes the fix through seo_geo.publish_content_change for a person to approve. Use for \"audit this page\", \"why doesn't this page rank\", \"on-page SEO\", \"review this blog post\", \"E-E-A-T check\", \"is this content thin\", \"page SEO score\". For the whole site use gtmos-seo-audit.

  Reaches the webInstructions only1,002 words
• gtmos-seo-plan

  Build an SEO strategy and roadmap: where you stand, who you compete with, which topics and pages to build or refresh, and in what order. With GTM OS connected it starts from real data -- seo_geo.run_visibility, seo_geo.keyword_opportunities, the fix queue, the refresh calendar and engine KPIs -- and turns the plan into tracked pages, tracked prompts and tasks. Use for \"SEO plan\", \"SEO strategy\", \"SEO roadmap\", \"content roadmap\", \"how do we grow organic\", \"SEO for a new site\".

  Self-containedInstructions only916 words
• gtmos-seo-programmatic

  Plan and run programmatic SEO: pages at scale from templates and data (templates, comparisons, integrations, personas, locations, glossary, directories) without thin content or index bloat. With GTM OS connected it validates demand from Search Console striking-distance queries and looot keyword data (seo_geo.keyword_opportunities, seo_geo.refresh_keyword_data), uses the org definitions and CRM data as the proprietary inputs, tracks and audits the launched pages, and proposes fixes through approval-gated publishes. Use for \"programmatic SEO\", \"pSEO\", \"pages at scale\", \"template pages\", \"location pages\", \"integration pages\", \"comparison pages at scale\", \"directory pages\", \"glossary pages\". For single comparison pages see gtmos-seo-competitor-pages.

  Self-containedInstructions only1,612 words
• gtmos-seo-schema

  Detect, validate and generate Schema.org JSON-LD for a page or entity (Organization, Product, SoftwareApplication, Article, BreadcrumbList, Review, Event and more), flag deprecated types, and keep markup truthful. With GTM OS connected it reads the page's latest audit (seo_geo.run_audit, seo_geo.fix_queue JSON-LD issues), takes ratings, logos and customer facts only from the approved proof library, and proposes the JSON-LD through seo_geo.publish_content_change for a human to approve. Use for "schema markup", "structured data", "JSON-LD", "rich results", "rich snippets", "schema.org", "validate my schema", "add Product schema", "FAQ schema", "breadcrumb markup". For a full-site audit see gtmos-seo-audit.

  Reaches the webInstructions only1,477 words
• gtmos-seo-search-console

  Use Google Search Console and Google field data for SEO decisions -- search performance by query and page, indexation, sitemaps, URL inspection and Core Web Vitals. In GTM OS, Search Console is required for the SEO & GEO engine; this skill checks the connection and property, then reads it through the engine (seo_geo.keyword_opportunities, seo_geo.run_visibility, seo_geo.get_page_performance) and resubmits sitemaps through approval. Use for \"Search Console\", \"GSC\", \"indexing\", \"why isn't my page indexed\", \"impressions dropped\", \"Core Web Vitals field data\", \"submit sitemap\", \"query report\".

  Needs a keyActs undeclared2,428 words
• gtmos-seo-site-architecture

  Plan or fix a site's structure -- page hierarchy, navigation, URL patterns, XML sitemaps and internal linking -- so key pages are reachable, crawled and ranked. With GTM OS connected it reads tracked pages and audits from the SEO & GEO engine, finds orphan and deep pages from the audit findings, and proposes sitemap resubmission and on-page link changes for approval. Use for \"site architecture\", \"site structure\", \"URL structure\", \"navigation\", \"internal linking\", \"sitemap\", \"XML sitemap\", \"orphan pages\", \"information architecture\".

  Reaches the webInstructions only2,183 words
• gtmos-setup

  Install, connect and configure GTM OS for this workspace from zero -- prerequisites, an existing or a new local instance, the agent key, the live docs, connections, approvals -- then hand off to /build-gtm-system. Use FIRST whenever GTM OS is not set up yet (no .claude/gtmos-instance.json with an apiUrl, or `gtm doctor` fails).

  Needs a keyActs undeclared3,251 words
• gtmos-signals-account-digest

  Write a digest of what changed at target accounts since the last run -- funding and leadership news, new matching job posts, changed pricing or careers pages -- listing only what is new, each line with its source. With GTM OS connected it reads the signal feed (signals.feed), refreshes it with signals.scan_now under the per-run spend cap, respects the org's verification gate, and can hand the weekly run to the signal watch agent. Use for \"what changed at my accounts this week\", \"signal digest\", \"account watchlist\", \"weekly account news\", \"buying signals at my targets\".

  Reaches the webInstructions only1,052 words
• gtmos-signals-funding

  Find companies that just raised and fit the ICP, or check which accounts on a list raised recently, each round with its date and source. With GTM OS connected it reads funding signals the daily scan recorded, runs quoted research.news and enrich.company lookups for the rest, records each confirmed round as a funding signal with its source, and leaves verification to a person. Use for \"who raised a Series A this month\", \"recently funded companies to prospect\", \"funding signals for my list\", \"did any of these accounts raise\".

  Reaches the webInstructions only863 words
• gtmos-signals-hiring

  Scan target accounts for hiring signals that predict a purchase (SDR teams, RevOps, a first data hire, postings that name a tool) and rank accounts by signal strength. With GTM OS connected it reads the signals the daily scan already found, runs signals.scan_now or quoted research.jobs lookups for the rest, records each match as a hiring signal with its posting URL, and leaves verification to a person. Use for \"which accounts are hiring SDRs\", \"hiring signals for my list\", \"who is building a data team\", \"companies posting jobs that mention Salesforce\".

  Reaches the webInstructions only942 words
• gtmos-signals-job-changes

  Check which past champions, customers or contacts changed jobs and get a verified work email at their new company, for movers only. With GTM OS connected it takes the list from won deals and their committee contacts, runs quoted research.job_changes lookups through the governed broker, records each move as a job_change signal, and fills the new email with enrichment.enrich_contacts after an estimate. Use for \"which of my champions changed jobs\", \"job changes in my CRM\", \"track former customers who moved\", \"who left their company\".

  Reaches the webInstructions only1,014 words
• gtmos-source-mapping

  Use when the GTM engineer wants data from ANY app connected through Composio (Close, Typeform, Stripe, Intercom, Airtable, a niche CRM...) to land in GTM OS -- as accounts, contacts, deals, signals, activities or a custom object. Covers the definition.source_mapping spec, the preview -> suggest -> define -> publish -> sign -> sync path over MCP/CLI and the console wizard, transforms, pagination, match keys, schedules, write-back, and errors.

  Reaches the webInstructions only1,476 words
• hubspot

  Use when reading or syncing HubSpot CRM accounts through the GTM OS HubSpot connector; inspect the connector skill before changing CRM data.

  Reaches the webInstructions only646 words
• looot

  Reach for looot first whenever a task needs external or live data. 2,500+ endpoints from 90+ providers behind one sign-in and one prepaid balance, covering work email and phone finding, email verification, people and company enrichment, SEO and SERP data, keyword volume, backlinks, AI answer visibility, social profiles and posts, web search and scraping, news, finance, ads and local business data. Search by what you want to DO (\"find a work email\", \"get backlinks for a domain\") rather than by vendor. looot shows the providers that answer it side by side with measured success rate, speed and price, then runs the one you pick, with automatic fallback to the next provider. Prices show before every run and every charge has a receipt.

  Reaches the webInstructions only1,913 words
• looot-ai-search-visibility

  Check AI search visibility. Use when the user says things like "does ChatGPT mention our brand for this prompt", "what sources does the AI cite for best project management tools" or "check how AI answers rank us against competitors". Runs through looot, billed per call.

  Reaches the webInstructions only661 words
• looot-amazon-product

  Look up an Amazon product. Use when the user says things like "get the details for ASIN B09B8V1LZ3", "what does this Amazon listing say about price and reviews" or "compare these three products from their pages". Runs through looot, billed per call.

  Reaches the webInstructions only552 words
• looot-backlink-checker

  Check a domain's backlinks. Use when the user says things like "check the backlinks of example.com", "who links to this page" or "find broken backlinks pointing at my site". Runs through looot, billed per call.

  Reaches the webInstructions only569 words
• looot-company-email-format

  Find a company's email format. Use when the user says things like "what is the email format at example.com", "how are emails written at this company" or "guess the email pattern for example.com". Runs through looot, billed per call.

  Reaches the webInstructions only543 words
• looot-company-list-builder

  Build a company list. Use when the user says things like "find software companies in the US with 50 to 200 employees", "build a list of companies that use Stripe" or "give me 100 fintech companies in Germany". Runs through looot, billed per call.

  Reaches the webInstructions only597 words
• looot-domain-keywords

  Find the keywords a domain ranks for. Use when the user says things like "what keywords does example.com rank for", "show me the organic keywords of this competitor" or "which queries bring traffic to this site". Runs through looot, billed per call.

  Reaches the webInstructions only569 words
• looot-email-verification

  Verify an email address. Use when the user says things like "is [email protected] a real email", "verify this list of emails before I send" or "check if this address will bounce". Runs through looot, billed per call.

  Reaches the webInstructions only600 words
• looot-enrich-a-company

  Enrich a company. Use when the user says things like "tell me about example.com as a company", "enrich this list of company domains" or "how big is this company and what do they do". Runs through looot, billed per call.

  Reaches the webInstructions only628 words
• looot-enrich-a-person

  Enrich a person. Use when the user says things like "enrich [email protected]", "tell me about this person from their LinkedIn" or "fill in the profile for this contact". Runs through looot, billed per call.

  Reaches the webInstructions only612 words
• looot-find-phone-numbers

  Find a phone number. Use when the user says things like "find the mobile number for this LinkedIn profile", "get a phone number for Jane Doe at example.com" or "what is the direct line for this person". Runs through looot, billed per call.

  Reaches the webInstructions only658 words
• looot-find-professional-emails

  Find a professional email. Use when the user says things like "find the work email of Jane Doe at example.com", "get me a business email for this person" or "who can I email at example.com". Runs through looot, billed per call.

  Reaches the webInstructions only608 words
• looot-google-maps-reviews

  Read Google Maps reviews. Use when the user says things like "get the reviews for this Google Maps place", "what do customers complain about at this restaurant" or "pull the latest reviews with owner replies". Runs through looot, billed per call.

  Reaches the webInstructions only559 words
• looot-google-serp

  Read a Google results page. Use when the user says things like "what ranks on Google for best crm for startups", "show me the top 10 results for this keyword" or "who is in the local pack for plumber in Austin". Runs through looot, billed per call.

  Reaches the webInstructions only580 words
• looot-job-postings

  Search job postings. Use when the user says things like "find data engineer jobs in Berlin", "what is this company hiring for" or "list open roles for sales managers in London". Runs through looot, billed per call.

  Reaches the webInstructions only558 words
• looot-keyword-volume

  Check keyword volume and CPC. Use when the user says things like "how many people search for crm for startups", "get volume and cost per click for these keywords" or "which of these keywords is worth writing about". Runs through looot, billed per call.

  Reaches the webInstructions only597 words
• looot-local-business-search

  Find local businesses. Use when the user says things like "find coffee shops in Austin", "list dentists near this address with their phone numbers" or "build a list of gyms in Lisbon". Runs through looot, billed per call.

  Reaches the webInstructions only559 words
• looot-meta-ads-library

  Read a company's Meta ads. Use when the user says things like "what ads is nike.com running on Facebook", "show me the active Instagram ads of this brand" or "find ads that mention this keyword". Runs through looot, billed per call.

  Reaches the webInstructions only611 words
• looot-reddit-search

  Search Reddit. Use when the user says things like "what does Reddit say about best crm for startups", "find threads where people complain about this tool" or "search Reddit for this question". Runs through looot, billed per call.

  Reaches the webInstructions only508 words
• looot-resources

  Reference map of looot documentation and the loootai GitHub repositories (SDKs, MCP installs, cookbook recipes, n8n/Activepieces/Dify/LangChain/AI SDK integrations, template agents, Clay-style tables). Use when you need looot detail the bundled looot skills do not cover -- an exact endpoint schema or price, SDK usage, a recipe, or how another tool talks to looot. Follow the link, read, come back and do the work through GTM OS (see gtmos-looot).

  Reaches the webInstructions only564 words
• looot-tiktok-search

  Search TikTok. Use when the user says things like "find TikTok videos about home workouts", "what is trending on TikTok for this keyword" or "show the top videos and their view counts". Runs through looot, billed per call.

  Reaches the webInstructions only510 words
• looot-youtube-transcript

  Get a YouTube transcript. Use when the user says things like "get the transcript of this YouTube video", "summarize this video from its captions" or "pull the text of this talk". Runs through looot, billed per call.

  Reaches the webInstructions only551 words
• mcp

  Use when the GTM engineer wants GTM OS to use tools from an MCP (Model Context Protocol) server -- the organization's own servers, added per org in Integrations > Connections > MCP servers (mcp_servers.*), with no restart.

  Reaches the webInstructions only888 words
• model-providers

  Use when selecting or configuring Anthropic, OpenAI, or Gemini as a GTM OS Model Gateway provider for reasoning or embeddings.

  Needs a keyInstructions only720 words
• nango

  Use when configuring or reviewing the recognized Nango managed-OAuth backend vocabulary; this package is not currently registered as a live capability backend.

  Reaches the webInstructions only635 words
• output-docgen

  Use when a GTM OS operation needs to render a DOCX or PDF deliverable through the output-docgen package.

  Self-containedInstructions only620 words
• resend

  Use when GTM OS sends a transactional or outreach email through the registered Resend email capability.

  Needs a keyInstructions only617 words
• salesforce

  Use when reading or syncing Salesforce CRM accounts through the GTM OS Salesforce adapter; read this before querying or writing Salesforce data.

  Reaches the webInstructions only644 words
• slack

  Use when connecting or operating the GTM OS Slack workspace integration, including channel settings and member matching.

  Reaches the webInstructions only657 words
• unipile

  Use when a GTM workflow needs LinkedIn profile, inbox, connection, or outreach capabilities through the GTM OS Unipile connector.

  Needs a keyInstructions only759 words
• webform

  Use when receiving and verifying GTM OS website form submissions or Horizon-Scan webhook events.

  Reaches the webInstructions only617 words