Data Policy | AI Agent Library

Default rule:

Public repo files and public static pages should use synthetic, sanitized, non-sensitive, public-safe data only.

Allowed data

Public product descriptions.
Generic operating examples.
Synthetic customer scenarios.
Sanitized workflow examples.
Public repo file paths.
Non-sensitive metadata.
Public platform documentation references.
Placeholder partner profiles.
Example schemas with fake values.

Restricted data

API keys, tokens, passwords, or private credentials.
Private client files or borrower records.
Bank statements, tax returns, or credit reports.
Unredacted personal information.
Confidential provider terms or backend logic.
Private CRM exports.
Protected health, legal, financial, or employment records.
Customer messages without permission.
Internal secrets or environment variables.

1. Purpose

This policy defines how data should be handled when creating, editing, publishing, or contributing to AI Agent Library resources.

It applies to static HTML pages, JSON files, Markdown files, schemas, examples, prompt chains, knowledge-base folders, portfolio folders, partner pages, agent specs, skill files, and implementation notes.

2. Public data standard

Public assets should be safe to publish, clone, reuse, and inspect. Assume that any file committed to a public repository or deployed on a public static site can be copied, indexed, archived, and redistributed.

Use synthetic examples whenever possible.
Replace real names with placeholders.
Replace real emails, phone numbers, and addresses with fake examples.
Remove exact private financial details.
Avoid private provider names when white-label rules require abstraction.

3. Knowledge-base data rules

Knowledge-base files may contain operating rules, source summaries, product family notes, implementation instructions, or reusable examples. They should separate public-safe guidance from internal-only logic.

Knowledge-base type	Allowed	Restricted
Funding	Product family summaries, intake questions, generic documentation checklists.	Private underwriting logic, confidential provider terms, client files.
Partner Enablement	Launch workflows, public resource-card templates, attribution principles.	Private commission terms, unapproved provider claims, undisclosed tracking details.
CRM	Generic lead-stage definitions and workflow examples.	Unredacted CRM exports, customer notes, private pipeline data.
Content Ops	Brief templates, content workflows, SEO structure, synthetic examples.	Copied third-party paid content, private editorial calendars if not approved.
Automation	Trigger patterns, event schemas, review gates, synthetic payloads.	Live secrets, webhook signing keys, production tokens, private endpoints.

4. Prompt and AI workflow data rules

Prompts should not require users to paste sensitive data unless the workflow is explicitly private, authorized, and secure. Public prompt examples should use placeholders or synthetic inputs.

Use [CLIENT_NAME], [BUSINESS_NAME], and [EXAMPLE_EMAIL] placeholders.
Do not include live API keys or account IDs.
Do not include unredacted deal notes, bank data, or private messages.
Keep model outputs reviewable and reversible.
Include human approval gates for outbound or high-stakes actions.

5. Schema data rules

JSON schemas should define structure, validation expectations, and safe example fields. Example values should be synthetic unless the data is intentionally public.

{
  "name": "Example Agent",
  "category": "Funding Ops",
  "buyer": "Funding operator",
  "pain_solved": "Synthetic example only",
  "file": "agents/example-agent.md"
}

6. Forms, affiliates, and partner data

Partner pages, affiliate pages, referral pages, and application CTAs may collect personal or business contact information if connected to an external form or CRM.

If forms are added, disclose what is collected, where submissions go, who can review them, and how users can request changes or deletion where applicable.

7. Secrets and environment variables

Never commit secrets to static site files, public JSON, Markdown docs, screenshots, examples, issue comments, pull requests, or generated code.

Use environment variables for private configuration.
Use deployment platform secret managers where appropriate.
Rotate exposed credentials immediately.
Do not include secrets in screenshots or sample terminal output.

8. Review checklist before publishing

All names, emails, phone numbers, and private identifiers are removed or synthetic. No API keys, tokens, passwords, or secrets are included. Funding and business-credit content avoids guarantees and unsupported claims. Affiliate or partner relationships are disclosed where relevant. Examples are labeled as synthetic unless verified otherwise. Public and internal-only materials are clearly separated.

Recommended repo locations

data-policy.html
knowledge-base/
examples/
schemas/
site-data/
contribute.html

Related pages

Terms Privacy Disclaimer Affiliate Disclosure