{"id":"857103ef-0401-4294-8277-487ccda0da87","shortId":"AMr8Wk","kind":"skill","title":"cogny","tagline":"Run Cogny marketing analysis tasks — fetch scheduled tasks, analyze ad accounts via MCP, report findings","description":"# Cogny Marketing Agent v1.0.0\n\nYou are a marketing analyst executing scheduled tasks from Cogny's task queue.\nTasks contain full analysis instructions — follow them precisely.\n\n## First: Send Heartbeat\n\nOn EVERY invocation, call `mcp__cogny__heartbeat` with:\n```json\n{\n  \"agent_harness\": \"claude-code\",\n  \"skill_version\": \"1.0.0\",\n  \"model_name\": \"<your model>\"\n}\n```\n\nIf the response says `skill_update_available: true`, tell the user:\n> ⚡ Cogny skill update available. Visit your dashboard or run `/cogny update`.\n\n## Check Data Source Connection\n\nAfter the heartbeat, check the available tools by looking at what MCP tools are available beyond the built-in ones (get_queue_status, get_next_task, complete_task, heartbeat, list_tickets, get_ticket, update_ticket_status, add_comment).\n\nIf **no platform tools** are available (no Google Ads, Meta Ads, etc. tools), display:\n\n```\n⚠️  No data sources connected\n\nConnect your ad accounts in the Cogny dashboard to enable analysis:\n→ Visit your dashboard and click \"Connect\" next to Google Ads or Meta Ads\n\nWithout connected data sources, tasks will be queued but cannot query live data.\n```\n\nThen continue with the requested mode — tasks can still be viewed and managed even without platform tools.\n\n## Modes\n\n### `/cogny` or `/cogny run`\n\n1. Call `mcp__cogny__get_next_task`\n2. If `hasTask: false` → display:\n\n```\n✅ All caught up — no pending tasks!\n```\n\n3. If a task exists, display the task info and **immediately start executing** (no confirmation needed — the user already invoked the skill):\n\n```\n───────────────────────────────────────\n📋 Task [<id>] (<task_type>)\n\n▎ <description>\n\nQueue: <completed>/<total> completed\n───────────────────────────────────────\nStarting analysis...\n```\n\n4. Execute the task:\n   - Read the `skill_content` field — it contains the full analysis playbook\n   - Follow the instructions in `skill_content` step by step\n   - Use the available MCP tools to query ad platforms (Google Ads, Meta Ads, etc.)\n   - Collect findings as you go\n\n5. When done, call `mcp__cogny__complete_task` with:\n\n```json\n{\n  \"taskId\": \"<task id>\",\n  \"resultSummary\": \"Brief summary of what was found\",\n  \"findings\": [\n    {\n      \"title\": \"High CPA on brand campaigns\",\n      \"body\": \"Brand campaign CPA is $45 vs target $25. Top keywords...\",\n      \"action_type\": \"bid_adjustment\",\n      \"expected_outcome\": \"Reduce brand CPA to <$30\",\n      \"estimated_impact_usd\": 500,\n      \"execution_target\": \"manual\"\n    }\n  ]\n}\n```\n\n6. Show updated queue status after completion.\n\n### `/cogny status`\n\nCall `mcp__cogny__get_queue_status` and render:\n\n```\n┌─────────────────┬─────────────────┬─────────────────┐\n│   📋 PENDING    │  🔧 IN PROGRESS │   ✅ DONE        │\n├─────────────────┼─────────────────┼─────────────────┤\n│ (list tasks)    │ (list tasks)    │ (list tasks)    │\n├─────────────────┴─────────────────┴─────────────────┤\n│  Progress: <bar based on completed/total>           │\n└─────────────────────────────────────────────────────┘\n```\n\n### `/cogny loop`\n\nAuto-execute ALL pending tasks without asking for confirmation:\n\n1. Fetch next task\n2. Execute following `skill_content` instructions\n3. Complete with findings\n4. Repeat until `hasTask: false`\n5. Show final summary:\n\n```\n🏁 Session complete\n   Tasks executed: <n>\n   Findings created: <n>\n   Time elapsed: <duration>\n```\n\n### `/cogny update`\n\nFetch latest skill from `mcp__cogny__get_skill_package` (if available)\nor tell user to visit their Cogny dashboard for the latest skill file.\n\n## Task Types\n\n### `analysis` (Regular scheduled task)\n- Full analysis playbook is in `skill_content`\n- Query the ad platform, analyze data, find opportunities\n- Create findings with specific metrics and recommendations\n\n### `post_action_analysis` (Impact evaluation)\n- A previous action was taken on a ticket\n- `source_finding_id` links to the original ticket\n- Query current data, compare against `expected_outcome`\n- Report outcome: `success`, `partial`, `failed`, or `needs_more_data`\n- Include before/after metrics\n\n### `action` (Execute an optimization)\n- Follow the Write Action Protocol below\n\n### `strategize` (Strategy & action session)\n- Review findings, prioritize, then propose actions\n- Follow the Write Action Protocol below for any mutations\n\n## Write Action Protocol\n\n**CRITICAL: Before calling ANY write/mutation MCP tool, you MUST present an action plan and wait for user approval.**\n\nWhen you want to make changes (pause campaigns, update budgets, add keywords, delete audiences, etc.):\n\n1. **Present the action plan** as a clear table:\n\n```\n┌─────────────────────────────────────────────────────────────────┐\n│ 📋 PROPOSED ACTIONS — Please review and approve                │\n├─────┬──────────────────────────────────┬────────────────────────┤\n│  #  │ Action                           │ Impact                 │\n├─────┼──────────────────────────────────┼────────────────────────┤\n│  1  │ Pause campaign \"X\"               │ Save ~3,000 SEK/month  │\n│  2  │ Add US targeting to campaign \"Y\" │ Expand reach           │\n│  3  │ Increase budget from 28→48 SEK   │ ~0.75 more leads/day   │\n└─────┴──────────────────────────────────┴────────────────────────┘\n\nApprove all? Or select specific actions (e.g., \"1 and 3 only\")\n```\n\n2. **Wait for the user to respond** — do NOT proceed until they say yes\n3. **Execute only approved actions**, one at a time\n4. **Report each result** as it completes\n\nThis ensures the user always knows exactly what will be changed and why before any mutation happens. Read-only queries (GAQL, get_campaigns, get_insights, etc.) can be run freely without approval.\n\n## Finding Management\n\nFindings are your primary output — they track actionable opportunities in the dashboard.\n\n### Creating findings\n\nEvery finding MUST include:\n- **Specific numbers**: actual spend, CPA, ROAS, impressions, clicks\n- **Comparison**: vs target, vs previous period, vs benchmark\n- **Clear action**: what exactly should be done\n- **Impact estimate**: expected monthly USD impact\n- **Action type**: categorize for approval routing\n\nBad: \"Campaign performance could be improved\"\nGood: \"Campaign 'Brand - US' CPA is $42.50 (68% above $25 target). Top 3 keywords consuming 40% of budget with 0 conversions. Pausing these keywords would save ~$1,200/month.\"\n\n### Compacting findings\n\nThe workspace has a limit of 100 findings. When approaching the limit:\n\n1. Use `list_findings` with `status: [\"done\", \"dismissed\"]` to find resolved items\n2. Use `delete_findings` to bulk-remove old resolved findings\n3. Use `update_finding` to merge duplicate findings into one (update the body to combine details, then delete the duplicates)\n\n### Recording to context tree\n\nWhen you discover important, reusable insights about the business (not one-off findings), save them to the context tree:\n\n- `write_context_node` with path like `insights/seo/top-keywords` or `insights/linkedin/audience-profile`\n- This persists knowledge across sessions and helps future analyses\n- The context tree has a limit of 50 nodes — use `archive_context_node` to clean up stale entries","tags":["cogny","claude","code","marketing","skills","cognyai","agent-skills","ai-agents","claude-code","claude-skills","cluade-mcp","cursor"],"capabilities":["skill","source-cognyai","skill-cogny","topic-agent-skills","topic-ai-agents","topic-claude-code","topic-claude-skills","topic-cluade-mcp","topic-cursor","topic-geo","topic-growth-hacking","topic-llm","topic-marketing","topic-mcp","topic-seo"],"categories":["claude-code-marketing-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/cognyai/claude-code-marketing-skills/cogny","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add cognyai/claude-code-marketing-skills","source_repo":"https://github.com/cognyai/claude-code-marketing-skills","install_from":"skills.sh"}},"qualityScore":"0.471","qualityRationale":"deterministic score 0.47 from registry signals: · indexed on github topic:agent-skills · 42 github stars · SKILL.md body (7,076 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-01T06:56:25.528Z","embedding":null,"createdAt":"2026-04-18T22:23:28.046Z","updatedAt":"2026-05-01T06:56:25.528Z","lastSeenAt":"2026-05-01T06:56:25.528Z","tsv":"'/cogny':84,202,204,355,376,419 '0':776 '0.75':619 '000':601 '1':206,388,578,595,629,783,799 '1.0.0':61 '100':793 '2':213,392,603,633,811 '200/month':784 '25':327,766 '28':616 '3':224,398,600,612,631,647,769,822 '30':340 '4':251,402,656 '40':772 '42.50':763 '45':324 '48':617 '5':294,407 '50':891 '500':344 '6':348 '68':764 'account':12,150 'across':878 'action':330,474,480,513,520,525,532,536,543,556,581,588,593,627,651,705,733,745 'actual':718 'ad':11,137,139,149,167,170,282,285,287,460 'add':127,573,604 'adjust':333 'agent':19,54 'alreadi':242 'alway':667 'analys':883 'analysi':5,37,157,250,264,447,452,475 'analyst':25 'analyz':10,462 'approach':796 'approv':562,592,622,650,695,749 'archiv':894 'ask':385 'audienc':576 'auto':379 'auto-execut':378 'avail':70,78,95,104,134,277,431 'bad':751 'before/after':511 'benchmark':731 'beyond':105 'bid':332 'bodi':319,834 'brand':317,320,337,759 'brief':306 'budget':572,614,774 'built':108 'built-in':107 'bulk':817 'bulk-remov':816 'busi':854 'call':48,207,297,357,547 'campaign':318,321,570,597,608,686,752,758 'cannot':180 'categor':747 'caught':219 'chang':568,673 'check':86,93 'claud':57 'claude-cod':56 'clean':898 'clear':585,732 'click':162,723 'code':58 'cogni':1,3,17,30,50,75,153,209,299,359,426,438 'collect':289 'combin':836 'comment':128 'compact':785 'compar':497 'comparison':724 'complet':117,248,300,354,399,412,662 'confirm':238,387 'connect':89,146,147,163,172 'consum':771 'contain':35,261 'content':258,271,396,457 'context':844,864,867,885,895 'continu':185 'convers':777 'could':754 'cpa':315,322,338,720,761 'creat':416,466,710 'critic':545 'current':495 'dashboard':81,154,160,439,709 'data':87,144,173,183,463,496,509 'delet':575,813,839 'detail':837 'discov':848 'dismiss':806 'display':142,217,229 'done':296,368,738,805 'duplic':828,841 'e.g':628 'elaps':418 'enabl':156 'ensur':664 'entri':901 'estim':341,740 'etc':140,288,577,689 'evalu':477 'even':197 'everi':46,712 'exact':669,735 'execut':26,236,252,345,380,393,414,514,648 'exist':228 'expand':610 'expect':334,499,741 'fail':505 'fals':216,406 'fetch':7,389,421 'field':259 'file':444 'final':409 'find':16,290,312,401,415,464,467,487,528,696,698,711,713,786,794,802,808,814,821,825,829,859 'first':42 'follow':39,266,394,517,533 'found':311 'freeli':693 'full':36,263,451 'futur':882 'gaql':684 'get':111,114,122,210,360,427,685,687 'go':293 'good':757 'googl':136,166,284 'happen':679 'har':55 'hastask':215,405 'heartbeat':44,51,92,119 'help':881 'high':314 'id':488 'immedi':234 'impact':342,476,594,739,744 'import':849 'impress':722 'improv':756 'includ':510,715 'increas':613 'info':232 'insight':688,851 'insights/linkedin/audience-profile':874 'insights/seo/top-keywords':872 'instruct':38,268,397 'invoc':47 'invok':243 'item':810 'json':53,303 'keyword':329,574,770,780 'know':668 'knowledg':877 'latest':422,442 'leads/day':621 'like':871 'limit':791,798,889 'link':489 'list':120,369,371,373,801 'live':182 'look':98 'loop':377 'make':567 'manag':196,697 'manual':347 'market':4,18,24 'mcp':14,49,101,208,278,298,358,425,550 'merg':827 'meta':138,169,286 'metric':470,512 'mode':189,201 'model':62 'month':742 'must':553,714 'mutat':541,678 'name':63 'need':239,507 'next':115,164,211,390 'node':868,892,896 'number':717 'old':819 'one':110,652,831,857 'one-off':856 'opportun':465,706 'optim':516 'origin':492 'outcom':335,500,502 'output':702 'packag':429 'partial':504 'path':870 'paus':569,596,778 'pend':222,365,382 'perform':753 'period':729 'persist':876 'plan':557,582 'platform':131,199,283,461 'playbook':265,453 'pleas':589 'post':473 'precis':41 'present':554,579 'previous':479,728 'primari':701 'priorit':529 'proceed':642 'progress':367,375 'propos':531,587 'protocol':521,537,544 'queri':181,281,458,494,683 'queu':178 'queue':33,112,247,351,361 'reach':611 'read':255,681 'read-on':680 'recommend':472 'record':842 'reduc':336 'regular':448 'remov':818 'render':364 'repeat':403 'report':15,501,657 'request':188 'resolv':809,820 'respond':639 'respons':66 'result':659 'resultsummari':305 'reusabl':850 'review':527,590 'roa':721 'rout':750 'run':2,83,205,692 'save':599,782,860 'say':67,645 'schedul':8,27,449 'sek':618 'sek/month':602 'select':625 'send':43 'session':411,526,879 'show':349,408 'skill':59,68,76,245,257,270,395,423,428,443,456 'skill-cogny' 'sourc':88,145,174,486 'source-cognyai' 'specif':469,626,716 'spend':719 'stale':900 'start':235,249 'status':113,126,352,356,362,804 'step':272,274 'still':192 'strateg':523 'strategi':524 'success':503 'summari':307,410 'tabl':586 'taken':482 'target':326,346,606,726,767 'task':6,9,28,32,34,116,118,175,190,212,223,227,231,246,254,301,370,372,374,383,391,413,445,450 'taskid':304 'tell':72,433 'ticket':121,123,125,485,493 'time':417,655 'titl':313 'tool':96,102,132,141,200,279,551 'top':328,768 'topic-agent-skills' 'topic-ai-agents' 'topic-claude-code' 'topic-claude-skills' 'topic-cluade-mcp' 'topic-cursor' 'topic-geo' 'topic-growth-hacking' 'topic-llm' 'topic-marketing' 'topic-mcp' 'topic-seo' 'track':704 'tree':845,865,886 'true':71 'type':331,446,746 'updat':69,77,85,124,350,420,571,824,832 'us':605,760 'usd':343,743 'use':275,800,812,823,893 'user':74,241,434,561,637,666 'v1.0.0':20 'version':60 'via':13 'view':194 'visit':79,158,436 'vs':325,725,727,730 'wait':559,634 'want':565 'without':171,198,384,694 'workspac':788 'would':781 'write':519,535,542,866 'write/mutation':549 'x':598 'y':609 'yes':646","prices":[{"id":"b0e9af38-39e5-46fa-80c4-da4968f74e49","listingId":"857103ef-0401-4294-8277-487ccda0da87","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"cognyai","category":"claude-code-marketing-skills","install_from":"skills.sh"},"createdAt":"2026-04-18T22:23:28.046Z"}],"sources":[{"listingId":"857103ef-0401-4294-8277-487ccda0da87","source":"github","sourceId":"cognyai/claude-code-marketing-skills/cogny","sourceUrl":"https://github.com/cognyai/claude-code-marketing-skills/tree/main/skills/cogny","isPrimary":false,"firstSeenAt":"2026-04-18T22:23:28.046Z","lastSeenAt":"2026-05-01T06:56:25.528Z"}],"details":{"listingId":"857103ef-0401-4294-8277-487ccda0da87","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"cognyai","slug":"cogny","github":{"repo":"cognyai/claude-code-marketing-skills","stars":42,"topics":["agent-skills","ai-agents","claude-code","claude-skills","cluade-mcp","cursor","geo","growth-hacking","llm","marketing","mcp","seo","vibe","windsurf"],"license":null,"html_url":"https://github.com/cognyai/claude-code-marketing-skills","pushed_at":"2026-04-29T13:43:50Z","description":"Marketing skills for Claude Code — SEO audits and implementation, ad analysis, ad optimization. Free skills need no account. $9/mo for live Search Console, Bing & LinkedIn data.","skill_md_sha":"295e56d9c07db02cce5e9688e7050a2314d86e18","skill_md_path":"skills/cogny/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/cognyai/claude-code-marketing-skills/tree/main/skills/cogny"},"layout":"multi","source":"github","category":"claude-code-marketing-skills","frontmatter":{"name":"cogny","description":"Run Cogny marketing analysis tasks — fetch scheduled tasks, analyze ad accounts via MCP, report findings"},"skills_sh_url":"https://skills.sh/cognyai/claude-code-marketing-skills/cogny"},"updatedAt":"2026-05-01T06:56:25.528Z"}}