Scoped agent builds with human approval gates
Productized implementation sprints, digital products and open reference systems. Scoped deliverables, inspectable code and partner rates dated to their source. No income promises.
Audit, architect, implement, verify. Then a person approves.
- 01 / 05
Audit
Read what exists before proposing a change.
scripts/audit-affiliate.mjs- run
- pnpm run audit:affiliate
- script
- node scripts/audit-affiliate.mjs
- 02 / 05
Architect
Pick one delivery route and write its boundary down.
data/income-models.json- routes
- services · products · systems
- updatedAt
- 2026-08-16
- 03 / 05
Implement
Ship deliverables as files with sizes anyone can check.
lib/pack-manifests.ts- file
- downloads/agentic-income-starter-kit.md
- size
- 2144 bytes · 91 lines
- 04 / 05
Verify
Run every check before anyone is asked to trust the result.
package.json- run
- pnpm run merge:gate
- script
- pnpm lint && pnpm test && pnpm type-check && pnpm links:check:static && pnpm verify:public && pnpm build
- 05 / 05
Human gate
Checkout stays closed until a person changes the catalog status.
products/catalog.json- catalog permits sale
- Agentic Income Blueprint
- catalog refuses checkout
- 5 of 6 paid products
- updatedAt
- 2026-11-03
One real run. Every boundary visible.
The open-source affiliate audit is a concrete reference workflow: known inputs, executable source, a reviewable output, and a human decision before anything becomes public.
node scripts/affiliate-audit.mjs --content=<site>/content --writeThe command reads files and writes an audit. It does not join a program, change a public page, send a message, or claim revenue.
Load request & evidence
Point the agent pipeline at customer repository content and program records. Accepts content paths, catalog data, and traffic context.
Output · Structured content paths + verified catalog recordsAnalyze & draft solution
The agent matches program aliases against site content, calculates gap coverage, and drafts recommendations without executing external calls.
Output · Prioritized gaps & draft recommendationsGenerate AUDIT.md artifact
The run outputs an inspectable AUDIT.md deliverable with affected paths, verification status, traffic context, and disclosure text.
Output · A reviewable Markdown deliverable artifactApprove before publishing
A human principal verifies terms, destination fit, and compliance before any public publishing or external partner commitment.
Output · Explicit approve or reject next action
Inspect the skill contract and executable script directly on GitHub. These links are the proof; the diagram is only the explanation.
Reference implementation only. It demonstrates the operating method; it is not customer evidence or a revenue result.
Three delivery routes for customer value.
Productized Implementation — Agent-as-a-Service
Customer job: Deploy custom AI agent workflows for your team. We scope, test, and deliver repo-committed code with human gates, while client teams maintain authority.
- Execution:
- delivery-bounded: scoped deliverables, concrete test suites, and explicit human approval gates before production execution
- Timeframe:
- scoped implementation sprint — fixed 2-day or custom timeframe based on acceptance checks
- Risk boundary:
- unbounded scope creep — clear boundaries, non-goals, and acceptance criteria prevent misaligned expectations
Digital Products — Self-Serve Blueprints & Kits
Customer job: Install production-proven agent architectures, prompt packs, and starter kits self-serve via merchant of record.
- Execution:
- self-directed implementation: executable assets, automated schemas, and step-by-step guides
- Timeframe:
- instant self-serve delivery — access blueprints and starter code immediately upon availability
- Risk boundary:
- lack of maintenance — products carry explicit versioning and maintenance promises to prevent staleness
Selective Affiliate & Open Core — Verified Proof Systems
Customer job: Inspect open-source reference implementations and query verified partner economics without hidden incentives.
- Execution:
- transparent research: disclosed partner recommendations and quarterly re-verified terms
- Timeframe:
- open access & verified partner tables — inspectable MIT code and dated commission terms
- Risk boundary:
- undisclosed incentives — every compensated destination is explicitly labeled with non-affiliate alternatives
We publish the rates and when we checked them.
| Program | Commission | Cookie window | Status | Verified |
|---|---|---|---|---|
| Higgsfield | up to 25% (rate rises with qualifying revenue) | not published | active | higgsfield.ai |
| Systeme.io | 60% lifetime recurring | not published | active | systeme.io |
| CapCut | up to 35% recurring (tiered by referral volume) | not published | active | capcut.com |
| ElevenLabs | 22% of first 12 months (11% on Business) | not published | active | elevenlabs.io |
| HeyGen | 35% for the first 3 months (Social Creator Program) | 30 days | verify | heygen.com |
| Writesonic | 20% of every payment for up to 12 months | 60 days | active | writesonic.com |
| Opus Clip | 25% for up to 12 months | not published | active | opus.pro |
| Synthesia | 25% of net payments for 12 months (Starter/Creator plans) | 60 days | active | synthesia.io |
| Topaz Labs | 20% (monthly plans: 20% of each payment for up to 12 months) | 30 days | active | topazlabs.com |
| Magnific | Rate not published officially | not published | verify | magnific.com |
| Descript | $25 flat per new subscription | 30 days | active | descript.com |
| Rytr | 30% for 12 months | not published | active | affiliates.rytr.me |
| Pollo AI | 30% on first purchases, renewals and upgrades | not published | active | pollo.ai |
| Whop | Seller-set per product (Whop's default global rate is 30%) | not published | verify | docs.whop.com |
Affiliate links disclosed. Rates verified between 16 Aug 2026 and 28 Sept 2026 against each program's own terms.
This site is an MCP server.
Reading about agentic income is research. Querying it is leverage.
- list_income_models
The three revenue delivery routes with customer jobs and deliverables.
- get_income_model
One route in full, including deliverables, scope, and next actions.
- search_partner_programs
The partner registry: rates, cookies, status, disclosed links.
- get_partner_program
Full record with commission note, source URL, verification date.
- list_products
The catalog with real availability — purchasable is false until checkout is live.
- fit_check
Deterministic recommendation: which route fits your hours, budget, audience, skills.
- get_playbook
The free starter kit and dated knowledge-track summaries.
Agents build. Humans hold the gate.
Partner rates dated and sourced.
We sell machinery and education — never outcomes.
The catalog renders from reality.
Availability follows the product catalog. Nothing says buy until checkout actually exists; today the free starter kit is the only live path. Paid prices and offer links appear only on products the catalog marks for sale.
Income Flywheel Starter Kit
free — delivered on-site, never sold through checkoutFreeAgentic Income Blueprint
for sale — private delivery via Polar$67Agent Team Pro
Waitlist · not yet on salePrivate Intel
Waitlist · not yet on saleClaims Guard Pack
Not yet on saleAgentic Business OS Pack Bundle
Not yet on saleVibe OS Pack
Not yet on sale
NO INCOME PROMISES — we sell machinery and education, never outcomes. Some research links are commission links, always disclosed inline.
Continue at agenticpassiveincome.com
That discipline lives at agenticpassiveincome.com — including its own MCP server your agent can query.