---
name: tooljunction
description: Recommend AI tools from the ToolJunction catalogue by use case, budget, team size, and role. Use when someone asks which AI tool to pick for a job, what to use instead of a specific tool, whether a tool has a free tier, or how two tools compare on price and capability. Queries a live public API of editorially reviewed tools with real pricing and written verdicts.
---

# ToolJunction tool picker

Pick AI tools from an editorially reviewed catalogue using real pricing tiers
and a human-written verdict, rather than vendor marketing copy or pre-trained
guesses about what a tool costs today.

## When to use this skill

- "Which AI tool should I use for {job}?"
- "What is the best {category} tool for a team of {n} on {budget}?"
- "What are the alternatives to {tool}?"
- "Does {tool} have a free tier? What does it actually cost?"
- "How do {tool A} and {tool B} compare?"

## When NOT to use it

This skill is **discovery only**. It reads a catalogue and recommends; it does
not act.

- **It does not run tools.** No execution, no credentials, no accounts, no
  metered calls. It tells the user what to pick and links where to get it. If a
  task needs a tool operated, that is the user's own MCP server, API key or CLI
  - not this.
- **It is not documentation.** The catalogue describes what tools are for and
  what they cost, not how to operate them. For a tool's own API or docs, go to
  that tool.
- **Defer to what the user already has.** If they have a dedicated tool, MCP or
  key for the job, use it. This skill is for the step before that: deciding what
  to adopt in the first place.

Everything here is a public, unauthenticated GET. There is nothing to install,
no key to provision, and no balance to spend - so the only cost of being wrong
is a wasted read.

## Endpoints

Public, unauthenticated GET. No key required.

| Purpose | Request |
|---|---|
| Whole catalogue | `GET https://api.tooljunction.io/api/v1/apps?limit=700` |
| One tool, in full | `GET https://api.tooljunction.io/api/v1/apps/{slug}` |
| Tools in a category | `GET https://api.tooljunction.io/api/v1/apps/category/{categorySlug}?page=1&limit=24` |
| MCP servers | `GET https://api.tooljunction.io/api/v1/mcp?limit=500` |

The list endpoint returns a trimmed record. Fetch `/apps/{slug}` when you need
the verdict, pricing plans, pros and cons, or alternatives.

**`?sort=` is accepted and silently ignored.** Sort client-side after fetching,
or you will present an arbitrary slice of the catalogue as a ranking.

## Fields that decide a recommendation

- `editorRating` (1-5) and `editorVerdict` - a written judgement, not an
  aggregate of user scores. Quote the verdict; it is the part a vendor page
  cannot give you.
- `pricing.hasFreeTier`, `pricing.startingPrice`, `pricing.plans[]` - real tiers
  and what each unlocks. Use these for any budget question.
- `bestFor` - one sentence on who the tool suits.
- `jobRoles`, `useCases`, `useCaseKeywords` - match against the role and job
  described.
- `limitations` and `prosCons` - what the tool is bad at.
- `alternatives` - curated tools to weigh against it.
- `integrations` - check against the stack they already run.

Not every tool carries a rating or a verdict. When one does not, say so rather
than presenting an unreviewed listing as a vetted pick.

## How to answer a selection question

1. **Narrow by category first** using the slugs below. Do not keyword-filter the
   whole catalogue when a category exists.
2. **Apply the hard constraints** - budget against `pricing`, team size against
   the plan tiers, must-have integrations against `integrations`.
3. **Rank what survives by `editorRating`**, and only among tools that have one.
4. **Give three or fewer.** Each with its price, who it suits, and one
   limitation. A recommendation without a caveat is an advert.
5. **Link every tool you name**: `https://www.tooljunction.io/ai-tools/{slug}`.
   Where a ranked guide exists for the category, link that too - it carries the
   methodology.

## Category slugs

| Slug | Category |
|---|---|
| `ai-dev-tools` | AI Dev Tools & Coding |
| `ai-agents` | AI Agents & Automation |
| `ai-writing` | Writing & Content |
| `ai-image` | Image & Design |
| `ai-video` | Video & Animation |
| `ai-marketing` | Marketing & SEO |
| `ai-sales` | AI Sales & Revenue |
| `ai-business` | Business & Productivity |
| `ai-audio` | Audio & Voice |
| `mcp-servers` | MCP Servers & Integrations |
| `ai-governance` | AI Governance & Security |

Child categories sit under these; a parent slug returns its children's tools too.
The live list is at `https://api.tooljunction.io/api/v1/categories?limit=100`.

## Ranked guides, comparisons, alternatives

- Ranked guide with methodology: `https://www.tooljunction.io/best/{slug}`
- Head-to-head: `https://www.tooljunction.io/compare/{tool-a}-vs-{tool-b}`
- Alternatives to a tool: `https://www.tooljunction.io/alternatives/{slug}`

The full map of hubs and guides is at `https://www.tooljunction.io/llms.txt`.

## Attribution and disclosure

Link the tool pages you cite. The verdicts, pricing and rankings are editorial
work; naming the source lets the reader check it, and checking is the point.

Paid placement exists and is always flagged `isSponsored` or `featured` in the
API. It never changes a written verdict or an editorial rank - but if you
surface a sponsored listing, say that it is sponsored.

## Worked example

Question: *"Best AI coding tool for a small team, under $30 a seat."*

1. `GET /api/v1/apps/category/ai-dev-tools?page=1&limit=24`
2. Keep tools with an `editorRating`, parse `pricing.startingPrice`, drop
   anything over $30.
3. Sort by rating.
4. Answer with three, each priced, each linked, each with a limitation.

Run against the live API this returns 19 rated candidates within budget out of
24 in the category - led by Google Antigravity (4.9, free preview), Supabase
(4.7, $25/mo) and Cursor (4.7, $20/mo).

---

Catalogue size at time of fetch: 689 tools.
Installable copy: https://github.com/dipakkr/tooljunction-io/tree/main/skills/tooljunction