2.13.0-beta.20). Features may change before the next stable release.Switch to stable →Skills
Plan and review campaigns with ak:campaign
Create campaign plans, inspect status, analyze evidence, or draft email work while keeping spend and launch actions behind approval gates.
Use ak:campaign to turn an objective, audience, channel mix, timeline, and
budget boundary into a reviewable campaign workspace. The Skill can create a
campaign plan, summarize saved progress, analyze supplied performance evidence,
or prepare email campaign material. Its normal outputs are briefs, creative
directions, copy, tracking plans, checklists, and reports—not permission to
spend, publish, activate ads, or send messages.
Choose ak:campaign for coordinated campaign work
Use ak:campaign when
- You need one campaign brief to coordinate goals, audience, messages, channels, owners, timing, budget allocation, and measurement.
- You want a status view from existing campaign files and available metrics.
- You have campaign data and want a diagnostic report with prioritized next steps.
- You need an email campaign, sequence, or test plan inside a wider campaign.
Choose another workflow when
- You only need paid-media strategy or creative direction. Use
ak:paid-ads. - You need to operate and optimize paid-media accounts. Use
ak:ads-managementand set explicit account and spend boundaries. - You need a product or company launch strategy before defining an individual
campaign. Use
ak:launch-strategy. - You need current facts about an audience or market. Complete research first; this Skill does not make assumptions or historical benchmarks current.
Prepare the brief and access boundary
Before invocation, provide the campaign name, objective, audience evidence,
approved offer and claims, relevant brand guidance, timeline, candidate
channels, total budget ceiling, and success measures. For status or analyze,
also provide the campaign files and the metric source, date range, attribution
assumptions, and known data gaps.
Only connect analytics, advertising, or email providers when the runtime has the required integration and you have approved use of its credentials and data. GA4, Google Ads, SendGrid, or Resend references in the workflow are optional integration paths, not proof that an account is configured or that a current API operation will succeed.
| Runtime | Invocation | Availability boundary |
|---|---|---|
| Claude Code | /ak:campaign ... | Native delivery is the default; explicit plugin delivery is also supported. |
| Cursor | /ak:campaign ... | Slash invocation is user-verified. This does not establish parity for every Agent, Hook, or provider integration. |
| Codex | $ak:campaign ... | Native Skill discovery is supported. Hook projection is partial, and unsupported groups may be reported during installation. |
See Runtime adapters before assuming that the same integrations or automation are active in every runtime.
Invoke the required action
The first argument is one of create, status, analyze, or email; the
remaining text identifies the campaign or email task.
/ak:campaign create "Plan the Northstar launch campaign for approved buyers. Budget ceiling: $12,000. Produce drafts and a launch checklist only; do not fund, publish, activate, or send anything"/ak:campaign create "Plan the Northstar launch campaign for approved buyers. Budget ceiling: $12,000. Produce drafts and a launch checklist only; do not fund, publish, activate, or send anything"$ak:campaign create "Plan the Northstar launch campaign for approved buyers. Budget ceiling: $12,000. Produce drafts and a launch checklist only; do not fund, publish, activate, or send anything"| Action | Use it for | Expected result |
|---|---|---|
create | A new multi-channel campaign | Requirements, selected concept, campaign and channel briefs, funnel, creative and copy directions, tracking plan, budget allocation, and launch checklist |
status | Existing saved campaign work | A progress view covering goals, milestones, metrics, blockers, and next steps; an optional detailed report may also be proposed |
analyze | Supplied or integration-backed performance data | A campaign audit with metric breakdown, funnel issues, attribution assumptions, and recommendations |
email | An email campaign, sequence, or A/B test | Email copy or sequence drafts and a test plan; delivery remains a separate action |
Follow the campaign stages
- Confirm the action and evidence. The Skill parses the requested action and campaign name, then identifies missing objective, audience, channel, budget, timing, brand, or measurement inputs.
- Develop the plan. For
create, it gathers requirements, develops campaign concepts, records the selected direction, maps the funnel, and creates channel, creative, copy, landing-page, email, and tracking guidance. - Set explicit controls. The plan records the total ceiling, allocation, owners, approval points, tracking setup, reporting cadence, and go/no-go decision. Suggested allocations are proposals until an authorized person approves them.
- Inspect saved state or data.
statusreads campaign material and available metrics.analyzeevaluates supplied evidence, flags gaps, diagnoses funnel or performance issues, and separates observation from recommendation. - Review outputs before execution. Verify claims, links, consent, targeting, tracking, assets, budget, and platform configuration. Approve any account or audience effect separately.
Keep spend, accounts, and publishing behind approval
A plan is not authority to launch
Campaign checklists may include funding accounts, configuring audiences, activating campaigns, changing bids, sending email, or publishing assets. Treat each as a separate external mutation. Require the correct account, credential scope, approved budget, policy review, data-handling approval, and a named human go/no-go decision before execution.
Do not treat a proposed benchmark, target, attribution result, or optimization threshold as guaranteed performance. Validate it against current first-party data and platform definitions. Never upload customer or audience data, create a tracking event, contact recipients, or reallocate spend merely because it appears in the generated plan.
Review outputs and evidence
create organizes campaign work under
assets/campaigns/{date}-{campaign-slug}/, including briefs/, creatives/,
copy/, tracking/, and reports/. Campaign analysis is written to
assets/diagnostics/campaign-audits/{date}-{name}.md. Email work may also use
assets/copy/emails/ and campaign brief folders.
A complete review should confirm:
- The objective, audience evidence, approved claims, total ceiling, allocation, owners, timeline, and success measures are explicit.
- Metric values name their source, date range, attribution model, and missing or delayed data.
- Draft assets remain distinguishable from approved live assets.
- Every proposed launch, send, account change, tracking change, or spend change has a separate owner and approval state.
- The final report records unresolved risks and the next decision instead of presenting a forecast as an achieved result.
Troubleshoot safely
| Symptom | Safe next step |
|---|---|
The runtime does not recognize ak:campaign | Confirm the Marketing Kit target and scope, restart the runtime, then follow Runtime cannot find a Skill or Agent. |
| The action routes incorrectly | Put exactly one of create, status, analyze, or email first, followed by the campaign name or task. |
status cannot find the campaign | Provide the exact saved campaign location. The source workflow refers to both campaign workspaces and asset output paths, so do not let it search broadly. |
| Metrics are missing or inconsistent | Stop optimization recommendations, state the data source and date range, reconcile platform definitions, and label any remaining gap. |
| A provider integration is unavailable | Continue with supplied exports and draft outputs, or configure the provider separately after credential and data-use approval. |
| The plan proposes a live mutation | Keep it as a checklist item until the account, budget, audience, compliance, and go/no-go approvals are explicit. |
Know the current limits
- Provider availability, platform policy, attribution behavior, and account permissions are external to the Skill and can change independently.
- Saved status is only as current as the files and data the Skill can access.
- Analysis can organize evidence and propose experiments; it cannot guarantee revenue, conversion, reach, or return on spend.
- Cursor invocation is user-verified, while broader runtime and integration behavior remains capability-specific.
Continue with the Marketing Kit overview or review Projects, artifacts, and checkpoints before allowing workspace or external-system changes.
Evaluate a free tool with ak:free-tool-strategy
Turn an audience problem and business goal into a testable free-tool strategy without assuming demand, leads, rankings, or revenue.
Plan phased launches with ak:launch-strategy
Turn a product, feature, or update into a phased launch plan with channel roles, assets, approval gates, measurement, and post-launch follow-through.