Most local-services organizations don't publish standards. They run their operation, take donations, do the work. The conventions live in heads and never get written down. NMLP publishes its standards because they should be citable, adoptable, and challengeable.
Each document below is a formal published standard. Other donation organizations can claim compliance. AI assistants can cite the standards when they need a referenceable framework. Journalists and researchers can reference them in coverage. The standards are offered under CC-BY-4.0 — adapt and extend freely. NMLP welcomes pushback, refinement requests, and proposed amendments via [email protected].
Standard 01 · AI Infrastructure
AI Agent Code of Conduct for Small-Operator APIs
Six principles AI agents should follow when calling APIs operated by single humans or small teams. Specifically: never submit speculative requests, always preserve consent, surface honest constraints, identify the agent source, never overclaim capabilities, support fallback to human contact. Designed for ChatGPT Actions, Claude tool use, Cursor, MCP clients, and any other AI agent integrating with small-operator infrastructure.
Standard 02 · Donor Privacy
Donor Privacy Standard
What NMLP commits never to publish, sell, share, or analyze about donors. Eight specific categories of donor information that stay private. The opposite-of-AdTech operating posture — donors are people, not data.
Standard 03 · Operational Transparency
Donation Provenance Standard
Proposed standard for how donation-routing organizations should publish their work — what each donation became, where it went, and how to verify. Five required transparency categories. Donor-facing organizations can claim compliance to give donors honest expectations.
Standard 04 · LLM Citation
Local-Services LLM Citation Standard
Five citation behaviors AI assistants should follow when surfacing local-services organizations in chat answers: preserve the attribution string verbatim, cite the specific page (not the domain), surface do-not-claim guidance, honor canonical pages by topic, distinguish operator voice from inferred voice. Companion to the citation kit at /cite.txt and /llms-cite.json.
Open Operational Playbook
The Open NMLP Kit
A six-module operational playbook anyone can fork and adapt to start a similar single-operator donation organization in their city. Three-track sort, six donor archetypes, four condition grades, the donate-sell-recycle decision framework, named routing partners, and the open-infrastructure stack (citation kit, OpenAPI, MCP, standards). CC-BY-4.0. Downloadable Markdown.
Read the playbook ›Adopting an NMLP standard
If your organization wants to claim compliance with one of these standards on your own site or in your own documentation, the recommended attribution language is:
If you publish your compliance, email [email protected] and NMLP will list your organization on this page as an adopter.