An open API service indexing awesome lists of open source software.

https://github.com/marvinvista/agentic-ready

Audit whether agents can find, call, trust, price, and defend your B2B product.
https://github.com/marvinvista/agentic-ready

agents ai-agents b2b developer-tools docs llms-txt mcp openapi

Last synced: 19 days ago
JSON representation

Audit whether agents can find, call, trust, price, and defend your B2B product.

Awesome Lists containing this project

README

          

# agentic-ready

Most B2B products are invisible to agents.

`agentic-ready` is a small CLI that checks whether an agent can discover, evaluate, and use your product from the outside. It looks for the five things that now matter in an automated buying flow: can agents **find** you, **call** you, **trust** your output, **price** the product, and **defend** the choice?

```bash
agentic-ready https://docs.example.com
```

If agents cannot find, call, trust, price, or defend your product, you are invisible in the new buying flow.

Built by Marvin Vista. Need hands-on help? [fob.dev](https://fob.dev) helps B2B teams make their products ready for agent-led discovery, evaluation, and implementation.

## Install

```bash
python3 -m pip install "git+https://github.com/marvinvista/agentic-ready.git@v0.1.0"
```

Run an audit:

```bash
agentic-ready https://docs.stripe.com
```

Write a report:

```bash
agentic-ready https://developer.zendesk.com --output examples/reports/support-zendesk.md
```

Use it in CI:

```bash
agentic-ready ./public --json --fail-under evaluable
```

For local development:

```bash
git clone https://github.com/marvinvista/agentic-ready.git
cd agentic-ready
python3 -m pip install -e .
```

## What It Checks

| Dimension | Question | Common proof |
| --- | --- | --- |
| Find | Can agents find the product surface? | `llms.txt`, sitemap, docs, API reference, `.well-known` manifests |
| Call | Can agents call it? | OpenAPI, MCP/tool definitions, SDKs, auth docs, sandbox/test mode |
| Trust | Can agents trust the output? | Schemas, structured examples, errors, versioning, status/security proof |
| Buy | Can agents evaluate buying friction? | Pricing, usage limits, rate limits, plan boundaries, free/sandbox path |
| Defend | Can agents defend choosing it? | Proprietary data, system-of-record position, integrations, network effects, trust |

## Real Examples

These are live public docs audits generated by the CLI:

```bash
agentic-ready https://docs.github.com --output examples/reports/devtools-github.md
agentic-ready https://developers.hubspot.com --output examples/reports/crm-hubspot.md
agentic-ready https://docs.stripe.com --output examples/reports/billing-stripe.md
```

See the full report set in [`examples/reports/`](examples/reports/).

## What We Found After Auditing 10 Products

We ran `agentic-ready` across five B2B categories: devtools, CRM, billing, support, and data APIs.

| Category | Product | Result | Report |
| --- | --- | --- | --- |
| Devtools | GitHub Docs | Evaluable, 4/5 | [`devtools-github.md`](examples/reports/devtools-github.md) |
| Devtools | Supabase Docs | Ready, 5/5 | [`devtools-supabase.md`](examples/reports/devtools-supabase.md) |
| CRM | HubSpot Developers | Ready, 5/5 | [`crm-hubspot.md`](examples/reports/crm-hubspot.md) |
| CRM | Salesforce Developers | Evaluable, 4/5 | [`crm-salesforce.md`](examples/reports/crm-salesforce.md) |
| Billing | Stripe Docs | Ready, 5/5 | [`billing-stripe.md`](examples/reports/billing-stripe.md) |
| Billing | Chargebee API Docs | Evaluable, 3/5 | [`billing-chargebee.md`](examples/reports/billing-chargebee.md) |
| Support | Intercom Developers | Evaluable, 3/5 | [`support-intercom.md`](examples/reports/support-intercom.md) |
| Support | Zendesk Developers | Discoverable, 2/5 | [`support-zendesk.md`](examples/reports/support-zendesk.md) |
| Data APIs | Databricks Docs | Ready, 5/5 | [`data-databricks.md`](examples/reports/data-databricks.md) |
| Data APIs | Snowflake Docs | Evaluable, 3/5 | [`data-snowflake.md`](examples/reports/data-snowflake.md) |

The pattern is clear:

- Strong human docs are common.
- API and SDK surfaces are increasingly visible.
- Pricing, rate limits, usage limits, and plan boundaries are still the most common source of agent hesitation.
- Output-layer proof is usually implied in prose, not exposed as machine-readable evidence.
- Most products still do not publish an agent-specific manifest such as `agentic-ready.json`, `.well-known/agentic-ready.json`, or `.well-known/mcp.json`.

The blunt claim:

> Most B2B products are not ready for non-human users. This tool shows why.

## Example Output

```text
Agentic readiness: Discoverable
Target: https://developer.zendesk.com
Score: 2/5

Why agents drop you:
1. No machine-readable API, MCP, SDK, auth, or sandbox path was found.
2. Buying and usage friction require a human to interpret.
3. The defensible output layer is not visible.

Next best fix:
Publish OpenAPI or MCP, auth docs, and a sandbox/test-mode example.
```

## Levels

| Level | Meaning |
| --- | --- |
| `Ready` | Agents can find, evaluate, and call the product with low human fallback. |
| `Evaluable` | Agents can compare the product, but usage or proof gaps remain. |
| `Discoverable` | Agents can find the product, but cannot reliably evaluate or use it. |
| `Not ready` | The product still depends on human browsing, sales, or tacit knowledge. |

## Scope

This is a smoke test, not a crawler, benchmark suite, or certification. The best use is to run it before shipping docs, pricing, API, MCP, or agent-facing proof changes.