design-referral — Claude Code command: referral program from concept to launch
design-referral is a Claude Code command that designs referral programs with incentives, rules, and governance, producing a charter, stakeholder deck, and risk log.
Apache-2.0 ★ 450 updated 2026-04-03
.claude/commands/design-referral.mdWhat design-referral does
design-referral structures the build of a referral program from goals and audience through budget and payout mechanics. It pulls GTM targets, legal and finance constraints, picks a mechanic (one-sided or two-sided incentives, tiering, payout rhythm), and defines rules: who can refer, caps, fraud safeguards, compliance requirements.
You get a program charter (objective, mechanics, incentives, KPIs), a stakeholder alignment deck showing who owns what, a risk log flagging legal and compliance issues, and a JSON plan saved in .claude/plans for audit. The tool wires in agents—Referral Architect, Incentive Design, Fraud Detection—and escalates if legal blockers or budget overruns surface twice; remediation logs to the plan.
Runs on Claude Code. Also available for Codex, GitHub Copilot, and Gemini CLI via separate config paths.
What it helps with
- design a referral program from zero—objectives, mechanics, incentives
- align stakeholders and set clear ownership
- surface risks: legal, compliance, fraud
- build a launch plan with assets, messaging, training
- tie KPIs (CAC, conversion, fraud rates) to dashboards
Example prompts for the agent
- Design a referral program to acquire new customers from existing ones; budget is 25,000
- Build a partner referral mechanic focused on revenue expansion with geo limits in Europe
- Create an employee referral program for hiring with a quarterly time limit
How to install design-referral
Manually: put the files in place
.claude/commands/design-referral.mdCall it with /design-referral. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/design-referral.mdCall it with /prompts:design-referral. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/design-referral.prompt.mdManually: put the files in place
.gemini/commands/design-referral.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about design-referral
How do I install design-referral in Claude Code?
Save the command file to .claude/commands/design-referral.md. It will be available in Claude Code after reload.
What does design-referral output?
A program charter (objective, mechanics, incentives, rules, KPIs), a stakeholder alignment deck, a risk log, and a JSON plan stored in .claude/plans for audit and change tracking.
What inputs does the command need?
Required: goal (signups, pipeline, revenue, expansion), audience (customers, partners, employees, influencers), and budget. Optional: regions and launch timeline.
Source text
design-referral.md · Apache-2.0
--- name: design-referral description: Produces referral program blueprint with incentives, rules, and governance. usage: /referral-program-orchestration:design-referral --goal "pipeline" --audience "customers" --budget 25000 --- # Command: design-referral ## Inputs - **goal** – primary objective (signups, pipeline, revenue, expansion). - **audience** – target participants (customers, partners, employees, influencers). - **budget** – incentive budget cap or range. - **regions** – optional list of geo restrictions/compliance considerations. - **timeline** – planned launch window. ### GTM Agents Pattern & Plan Checklist > Mirrors GTM Agents orchestrator blueprint @puerto/plugins/orchestrator/README.md#112-325. - **Pattern selection**: Referral design typically uses a **diamond** (context ↔ mechanic modeling, converge into governance) or **pipeline**; document choice in the plan header. - **Plan schema**: Save `.claude/plans/plan-<timestamp>.json` capturing goals, incentives, dependency graph (legal, finance, ops, partner), error handling, and success criteria (pipeline, CAC, fraud rates). - **Tool hooks**: Reference `docs/gtm-essentials.md` stack—Serena for charter diffs, Context7 for compliance + incentive benchmarks, Sequential Thinking for trade-off analysis, Playwright for prototype referral journeys. - **Guardrails**: Default retry limit = 2 for modeling or policy failures; escalation ladder = Referral Architect → Legal/Finance → Exec sponsor. - **Review**: Run `docs/usage-guide.md#orchestration-best-practices-puerto-parity` before sign-off to confirm dependencies, approvals, deliverables. ## Workflow 1. **Context Gathering** – pull GTM targets, persona insights, legal/finance constraints. 2. **Mechanic Selection** – evaluate one-sided vs two-sided incentives, tiering, and cadence. 3. **Rules & Safeguards** – define eligibility, referral caps, fraud checks, and compliance requirements. 4. **Enablement Plan** – list assets, messaging, training, and tooling needed for launch. 5. **Measurement Framework** – connect KPIs to dashboards and define review cadence. ## Outputs - Referral program charter (objective, mechanics, incentives, rules, KPIs). - Stakeholder alignment deck outlining responsibilities and approvals. - Risk log highlighting compliance/legal considerations. - Plan JSON entry stored/updated in `.claude/plans` for audit trail. ## Agent/Skill Invocations - `referral-architect` – leads program design and governance. - `incentive-design` skill – structures payouts and thresholds. - `fraud-detection` skill – recommends safeguards. ## GTM Agents Safeguards - **Fallback agents**: document substitutes (e.g., Partner Ops covering Referral Architect) when decision-makers unavailable. - **Escalation triggers**: if compliance blockers or budget overruns appear twice, trigger GTM Agents rip-cord escalation and log remediation in plan JSON. - **Plan maintenance**: update plan JSON/change log whenever incentives, eligibility rules, or governance cadences change. ---
© gtmagents · Apache-2.0 · gtmagents/gtm-agents