agentleFS
awaikawaik/direct-mcp-ai-project/.agents/skills/vk-ads-campaign-builder/SKILL.md

vk-ads-campaign-builder

Skill20 starsChanged 4 days agoMITView on GitHub

Создавать, аудитить и оптимизировать кампании VK Ads через LidFly MCP v3: connection_id/client_id, группы, объявления, лид-формы, аудитории, статистика и бюджеты. Использовать для работы с VK Рекламой с безопасным scope и write workflow.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 5 are distinctive, and 8 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 8

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use for VK Ads campaigns, ad groups, banners, lead forms, statistics, audiences, contextual phrases, budgets, statuses, and creative checks.

    line 8
  2. Very specific:

    Read schemas with get_tool_schema.

    line 13
  3. Very specific:

    If account/client is unclear, call get_provider_context({ provider: "vk", query?

    line 14
  4. Very specific:

    If campaign is named, call resolve_campaign_scope({ provider: "vk", query, workspace_project_id?

    line 15
  5. Very specific:

    Use only returned connection_id, client_id, workspace_project_id, and scope_arguments.

    line 16
  6. Very specific:

    Read via call_tool; write via call_write_tool.

    line 17
  7. Very specific:

    Before any campaign or ad-group write, read VK goal mode and write safety.

    line 29
  8. Distinctive:

    Before evaluating statistics, CPA, reach, budget distribution, missing impressions or a reach forecast, read VK Ads analytics.

    line 39
  9. Very specific:

    Read banner_requirements in vk_prepare_campaign / vk_get_package before requesting assets.

    line 49
  10. Distinctive:

    Keep the approved copy in a manual kit.

    line 53
  11. Distinctive:

    Avoid unsupported symbols in text.

    line 57
  12. Distinctive:

    Do not replace complete banner sections unless the schema requires it and current content has been reread.

    line 59
  13. Very specific:

    When the user asks to export a VK Ads report to Google Sheets or Google Docs, keep this skill for account/campaign scope and report reads, …

    line 67

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

VK Ads Campaign Builder

Use for VK Ads campaigns, ad groups, banners, lead forms, statistics, audiences, contextual phrases, budgets, statuses, and creative checks.

v3 Scope First

  1. Search with search_tools({ provider: "vk", query }).
  2. Read schemas with get_tool_schema.
  3. If account/client is unclear, call get_provider_context({ provider: "vk", query? }).
  4. If campaign is named, call resolve_campaign_scope({ provider: "vk", query, workspace_project_id? }).
  5. Use only returned connection_id, client_id, workspace_project_id, and scope_arguments.
  6. Read via call_tool; write via call_write_tool.

VK Account Rules

  • connection_id selects VK OAuth connection.
  • client_id selects agency/manager/client account.
  • Manual VK user-filter is allowed only when it appears in provider context; do not pass arbitrary VK user ids.
  • In manual user-filter mode, creation/upload/single update may be unavailable; prefer allowed mass actions and reread state.
  • For max_goals bidding, use max_price when updating limits; reread the group after write.

Write Safety

Before any campaign or ad-group write, read VK goal mode and write safety. It is the source of truth for checked_packages.goal_mode, priced_goal, bootstrap groups and ambiguous outcomes.

The compact legacy provider projection is methodology.

MAX Messenger

For «реклама в MAX», «канал MAX» or MAX messenger ads, read MAX placements and API limits. Distinguish ads inside MAX from a MAX channel promoted in the VK feed.

Analytics

Before evaluating statistics, CPA, reach, budget distribution, missing impressions or a reach forecast, read VK Ads analytics. Tool descriptions carry no evaluation thresholds; judgment comes from that reference and the client's goals in project memory.

Phrase Expansion

Offer search_phrase_vectorization only when the user wants broader reach from contextual phrases; do not enable it as a default optimization. Explain the audience change before confirmation, use a phrase segment in targetings.segments, and reread the field after creation. Manual user-filter does not support this setting. An unconfirmed write must not be repeated. Source: VK AdGroup API, local documentation checked 2026-09-26; live reads on the own account accepted fields but omitted this field, so absence is not proof of false.

Creative And Text

For a lead form or multiple image proportions in one ad, read Lead forms and image formats.

  • Read banner_requirements in vk_prepare_campaign / vk_get_package before requesting assets. Check draft text immediately with vk_prepare_banner(package_id=..., targetings=...), even before group creation or image upload. Missing material keeps ready=false; never present that as a completed launch.
  • In built-in chat import attached originals with vk_upload_image(chat_attachment_id=..., role=...). The server checks ownership, geometry and the original hash before approval. A connected ChangeSet may bind its content_id to content.<role>.id and a campaign's ad_group_id to the banner in the same exact VK scope.
  • list_media is the LidFly transit library, not the VK cabinet library; empty results do not disprove the user's assets. request_upload → curl is for external MCP clients only. A filename/path without bytes is not an uploaded image.
  • For a manual kit, use the checked format limits, identify any missing slots and preserve the chosen texts and budget in existing conversation decisions. Do not return to unavailable import steps.
  • Keep the approved copy in a manual kit. Label proposed rewrites as drafts; never add prices, guarantees, service details or delivery promises absent from the client's confirmed brief. Unknown fields stay explicitly unfilled.
  • Keep distinct customer offers separate: for example, family and children's formats retain their own conditions. A newer explicit customer decision replaces the older condition for that same offer; it does not silently change another offer. Label audience/cost estimates as hypotheses with inputs, period and uncertainty unless supported by observed statistics.
  • After a definitive provider refusal, deliver the useful work already prepared: all text variants, checked format limits, missing assets and exact blocking step. Distinguish supplied text, API readiness and actual creation; missing group IDs or images keep the banner unready. Do not invent IDs, repeat the rejected POST or substitute a package/account. Explain the known failure from its receipt; an unknown provider cause remains unknown. Offer further read-only diagnosis or an actionable manual kit without requiring the client to ask for the existing texts again.
  • Validate package/banner pattern before creating banners.
  • Avoid unsupported symbols in text; keep copy within VK field limits.
  • Upload images/videos only after source asset is final.
  • Do not replace complete banner sections unless the schema requires it and current content has been reread.

Workspace

Save decisions, campaign snapshots, analytics summaries, and follow-up scheduled tasks only after resolving workspace_project_id.

Google Export

When the user asks to export a VK Ads report to Google Sheets or Google Docs, keep this skill for account/campaign scope and report reads, then hand the verified Google write and reread to $export-ad-reports.

The export handoff applies only when this host exposes the skill and a Google write connector. If either is unavailable, state the exact missing capability and return the requested report as a draft; do not claim a saved Google file or silently switch to Workspace.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.