AGENTS.md Template for Marketing Campaign Orchestration

Overview

Direct answer: This AGENTS.md Template defines a project-level operating manual for a marketing campaign orchestration workflow using AI coding agents, enabling both single-agent and multi-agent coordination with clear roles, handoffs, memory, tool governance, and human review.

The AGENTS.md template documents the governance, decision boundaries, and operational constraints for a marketing campaign orchestration system. It covers agent rosters (planner, content researcher, copywriter, channel ops, data analyst, reviewer, and implementation), supervisor behavior, handoff rules, and secure tool access. It also codifies file structure, data sources, and validation criteria so teams can reproduce, audit, and extend the campaign automation over time.

When to Use This AGENTS.md Template

• When orchestrating multi-channel marketing campaigns across email, paid media, social, and landing pages using AI agents.
• When you have a roster of specialized agents requiring clear handoffs, shared memory, and a single source of truth.
• When governance, risk, and security of tools and data are important to production campaigns.
• When you want a versioned operating manual that can be pasted into a repository as project context.

Copyable AGENTS.md Template

Paste this block into AGENTS.md to establish the project operating context for marketing campaign orchestration agents. The block below starts with the AGENTS.md header and includes all required sections.

# AGENTS.md

Project role
- Marketing Campaign Orchestrator: coordinates multi-agent activities across planning, content creation, data analysis, channel execution, and quality assurance to deliver aligned marketing campaigns with measurable outcomes.

Agent roster and responsibilities
- Planner Agent: defines campaign goals, target audience, channel mix, timing, budget envelope, and success criteria.
- ContentResearcher Agent: aggregates audience insights, brand guidelines, and content ideas; creates a creative brief for copy and visuals.
- Copywriter Agent: generates ad copy, email copy, landing page copy, and subject lines aligned to the brief.
- ChannelOps Agent: configures and schedules pushes to email, social, and paid channels; sets up tool integrations and automation rules.
- DataAnalyst Agent: collects metrics, monitors KPIs, and feeds dashboards and alerting rules.
- Implementation Agent: executes tasks through CI/CD-like pipelines, API calls, or platform interfaces.
- Reviewer Agent: validates outputs for brand safety, quality, and compliance; approves releases.

Supervisor or orchestrator behavior
- The Orchestrator maintains the campaign plan, tracks agent outputs, resolves conflicts, and ensures all artifacts are timestamped, sourced, and versioned.
- It enforces memory: outputs are stored in a shared memory store and referenced by evidence IDs.
- It enforces source-of-truth: data and decisions reference primary sources (e.g., analytics dashboards, approved briefs).

Handoff rules between agents
- Planning → ContentResearcher: share creative brief, audience segments, and success criteria.
- ContentResearcher → Copywriter: handoff briefs, tone, and approved assets.
- Copywriter → ChannelOps: handoff assets, scheduling, and deployment instructions.
- ChannelOps → DataAnalyst: handoff post-implementation data collection and monitoring criteria.
- DataAnalyst → Reviewer: handoff dashboards, anomalies, and recommended actions.

Context, memory, and source-of-truth rules
- Context is captured as memory entries with tags, IDs, and source URLs.
- All facts must cite primary sources; memory entries include a provenance field.
- Use a single source-of-truth for KPIs (analytics dashboards) and for briefs (brand guidelines repository).

Tool access and permission rules
- Agents may call permitted tools and APIs only through approved integrations listed in the project config.
- Secrets must be retrieved from a central vault and never logged.
- Tool usage must be auditable; every call produces a trace.

Architecture rules
- Maintain a modular pipeline: Planner, Research, Content, Channel, Analytics, and QA components.
- Use event-driven triggers and idempotent operations.
- All artifacts are versioned and stored in a central repository.

File structure rules
- /campaigns
  - /{campaign-id}
    - /agents
      - /planner
      - /content-researcher
      - /copywriter
      - /channel-ops
      - /data-analyst
      - /implementation
      - /reviewer
    - /data
    - /assets
    - /workflows
    - /docs

Data, API, or integration rules
- Use standard marketing APIs (email, social, ads) with scoped permissions.
- Validate API responses against schema and log anomalies.
- Respect rate limits and retries with backoff per provider.

Validation rules
- All outputs must pass content quality checks, brand alignment checks, and data validation against sources of truth.
- Confirm channel platform responses include success status and IDs.

Security rules
- Principle of least privilege for all agents and service accounts.
- Secrets stored in vault; no plaintext secrets in memory.
- Access to production channels requires explicit approval gates.

Testing rules
- Unit tests for decision logic; integration tests for API calls; end-to-end tests for a full campaign run.
- Mock data should reflect realistic audience segments and performance metrics.

Deployment rules
- Deploy agent components in small, traceable steps.
- Feature flags govern publishing to production channels.
- Rollback plan includes reverting deployed configurations and clearing queued tasks.

Human review and escalation rules
- Human review is required for high-risk actions (new audience segments, budget increases beyond threshold, or creative assets failing QA).
- Escalate to campaign owner if metrics deviate beyond acceptable variance.

Failure handling and rollback rules
- If a step fails, roll back to the last successful state and notify the Orchestrator.
- Maintain partial state for post-mortem analysis; do not silently drop steps.

Things Agents must not do
- Do not bypass approvals or shortcuts.
- Do not modify the campaign brief after it has been signed off without re-approval.
- Do not access production data outside approved channels.

Recommended Agent Operating Model

The agent roster defines clear roles and decision boundaries. The Planner coordinates strategy; the ContentResearcher and Copywriter generate and validate creative; ChannelOps executes delivery; DataAnalyst monitors outcomes; Reviewer ensures quality. Handoffs are explicit; escalation paths are defined to the Orchestrator. This model supports multi-agent orchestration with predictable outputs and auditable traces.

Recommended Project Structure

/campaigns
  /<campaign-id>
    /agents
      /planner
      /content-researcher
      /copywriter
      /channel-ops
      /data-analyst
      /implementation
      /reviewer
    /data
    /assets
    /workflows
    /docs
    /tools

Core Operating Principles

• Explicit roles and clear handoffs between agents to prevent drift.
• Single source of truth for KPIs, briefs, and assets.
• Idempotent actions and traceable outputs for auditability.
• Least privilege and secret management for tool access.
• Fail-fast with safe rollbacks and human escalation when needed.

Agent Handoff and Collaboration Rules

The Planner, ContentResearcher, Copywriter, ChannelOps, DataAnalyst, Reviewer, and Implementation agents follow explicit handoff and collaboration rules to ensure consistent state and auditable decisions.

Tool Governance and Permission Rules

Tool access is governed by a central policy. All API calls and edits are logged and auditable. Secrets are kept in a vault. Approvals gates are required for production deployments.

Code Construction Rules

Implementation follows modular, testable components with clear interfaces. Use idempotent operations and avoid side effects without explicit consent.

Security and Production Rules

Security policies emphasize least privilege, secret management, and approved production channels. Production deployments require review and approval gates, with rollback plans.

Testing Checklist

• Unit tests for decision logic and function outputs.
• Integration tests for API calls with mock responses.
• End-to-end tests simulating a campaign run with expected KPIs.
• Manual exploratory testing for edge cases in creative assets and audience targeting.

Common Mistakes to Avoid

• Over-automation without human review for high-risk actions.
• Unclear handoffs leading to duplicated work or drift.
• Neglecting source-of-truth and provenance in memory entries.
• Bypassing approvals or using production channels without gating.

FAQ

What is the purpose of this AGENTS.md Template for marketing campaigns?

This AGENTS.md Template defines roles, workflows, and governance for marketing campaign orchestration with AI agents, enabling multi-agent coordination and handoffs.

How many agents are typically involved in the roster?

The roster includes Planner, ContentResearcher, Copywriter, ChannelOps, DataAnalyst, Implementation, and Reviewer, with room for domain-specific specialists.

How are handoffs between planner, implementer, and reviewer managed?

Handoffs follow a stated sequence with evidence IDs, briefs, assets, and validation results recorded in memory and tied to primary sources.

What are the security and permission guidelines for tool access?

Access is restricted to approved integrations, secrets are vaulted, and all actions are auditable with traceable calls and role-based permissions.

How is success validated and outputs measured?

Success is measured against campaign KPIs sourced from analytics dashboards, with validation checks on asset quality, compliance, and channel deliverability.