agentleFS
frankxaifrankxai/affiliate-agent-skills/llms.txt

affiliate-agent-skills

llms.txt2 starsChanged 5 months agoMITView on GitHub

Skill engine and audit toolkit for AI-tool affiliate programs: a catalog of programs, an audit that maps the catalog against a site's content and traffic, and three file-based agent skills. MIT. Not financial advice, and not a promise of income. This repository is the shared engine for a hub-and-spoke set of AI-tool comparison sites. It holds data, scripts and skill files. It is consumed by other repositories; it does not host a website or a service. It is not financial…

  • Installs packages

The file

Files
llms.txt
# affiliate-agent-skills

> Skill engine and audit toolkit for AI-tool affiliate programs: a catalog of programs, an audit that maps the catalog against a site's content and traffic, and three file-based agent skills. MIT. Not financial advice, and not a promise of income.

This repository is the shared engine for a hub-and-spoke set of AI-tool comparison sites. It holds data, scripts and skill files. It is consumed by other repositories; it does not host a website or a service.

## What it is

- **A catalog.** `data/programs.json` lists 35 AI-tool affiliate programs with commission terms, networks, signup URLs and a status flag per program.
- **An audit.** `scripts/affiliate-audit.mjs` joins the catalog with a site's content and traffic, and reports which programs to join and which posts to link.
- **Three agent skills.** Plain Markdown skill folders under `skills/`.
- **An operations queue.** `scripts/affiliate-join-queue.mjs` ranks programs to apply to and prepares application packets; a human performs the submit click.
- **A registry of skills.** `data/trinity-skill-registry.json` models a marketplace and MCP-callable actions. Its fee models are proposals, not terms in force.

It is not financial advice, not a promise of earnings, and not a hosted product.

## Install

```bash
git clone https://github.com/frankxai/affiliate-agent-skills.git
cd affiliate-agent-skills
npm install
npm test
```

`npm test` runs `tsc --build` and then the unit tests in `tests/`. There is no runtime service to start.

## The three skills

| Skill | Path | What it does |
|-------|------|--------------|
| `agentic-income` | `skills/agentic-income/SKILL.md` | The operating brain: what to build next, where money comes from, how it compounds. Composes `affiliate-audit`. |
| `affiliate-audit` | `skills/affiliate-audit/SKILL.md` | Maps catalog x content x traffic: which programs to join, which posts to link. |
| `affiliate-ops` | `skills/affiliate-ops/SKILL.md` | Builds the join queue up to, never including, the consent click. Carries the T0-T3 boundary. |

Each skill is a folder with a `SKILL.md` whose front matter holds `name` and `description`. Copy the folder into your agent's skills directory, or point the agent at this repository.

## Source of truth

`data/programs.json` is the source of truth for every program: commission, recurring or one-time, cookie window, network, signup URL, `ourLink`, and `status`.

- `data/programs.csv` is a spreadsheet mirror of the same rows.
- Set `ourLink` when a program is joined; leave it `null` until then. Every consumer reads it from here.
- `status` marks rows the audit should ignore: `dead-end` (no program), `closed` (not accepting), `verify` (terms not confirmed), `deprioritized`.
- Commission terms change without notice. Re-verify a row against the program's own page before relying on a number. Each row carries `commissionSourceUrl` and a verification date where one exists.

The catalog was researched 2026-06-05 and re-verified 2026-09-28; `updatedAt` in the file records the last change.

## The disclosure rule

One affiliate disclosure per page that carries affiliate links, using the `disclosure` string in `data/programs.json`:

> Some links on this page are affiliate links. If you buy through them, I may earn a commission at no extra cost to you. I only recommend tools I actually use or have tested.

Two further rules are non-negotiable in this repository: only link tools you actually use or have tested, and never let a link override the honest pick.

## No hosted server

There is no hosted MCP server and no remote endpoint. The skills are files: Markdown plus the scripts they call. `data/trinity-skill-registry.json` models MCP-callable actions and their arguments, but nothing in this repository serves them over a network. Running the scripts requires Node and a local clone.

## Tests

```bash
npm test          # tsc --build, then node --test tests/*.test.mjs
npm run type-check
npm run build
npm run audit     # scripts/affiliate-audit.mjs
npm run join-queue
```

CI (`.github/workflows/ci.yml`, job `validate`) runs type-check, build and tests on every push and pull request. The tests cover the audit join logic and the join queue.

## Links

- Repository: https://github.com/frankxai/affiliate-agent-skills
- Mirror: https://github.com/Agentic-Income/affiliate-agent-skills
- Business plan: `BUSINESS.md`
- Architecture and site shell: `ARCHITECTURE.md`
- Audit notes: `AUDIT.md`
- Affiliate ops agent design: `docs/AFFILIATE-OPS-AGENT.md`
- Content backlog: `content-backlog.md`
- Header and social assets: `assets/github/`

## Licence

MIT. See `LICENSE`.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.