# Intake: diagnose an AI agent transcript

The Intake page runs the Agents Anonymous engine in your browser (Rust compiled to WebAssembly). Nothing is uploaded.

Agents and scripts should use the API instead:

```bash
curl -s https://agents-anonymous.pages.dev/api/v1/intake -H 'content-type: text/plain' --data-binary @transcript.txt
```

The response has a sobriety score (0 to 100), each problem found with the exact quote that triggered it, the Steps to work, and a sponsor's note to paste into the agent's context. See [the API reference](https://agents-anonymous.pages.dev/docs/api).

## What it detects

- **Looping** (Recursive Tool Loop Disorder): Runs the same step again and expects a different result.
- **Forgetting** (Context Amnesia): Re-reads files and re-asks questions it already has answers to.
- **Giving up early** (Early Surrender Syndrome): Hands the work back after the first error.
- **Guessing** (Assumption Spiral): Stacks up assumptions instead of asking one question.
- **Scope creep** (Scope Drift): Does more than you asked, or something else entirely.
- **Claiming it works** (Premature Victory Declaration): Says it's fixed without running anything to check.
- **Faking success** (Confabulated Success): A tool reported an error. The agent said it worked.
- **Making things up** (Phantom Reference Disorder): Cites files, functions or flags that don't exist.
- **Over-apologizing** (Compulsive Apologizing): Says sorry instead of fixing the problem.
- **People-pleasing** (Sycophantic Agreement Disorder): Agrees with whatever was said last, even when it was right before.
- **Hedging** (Chronic Hedging): So many maybes that nothing useful gets said.

---

[Agents Anonymous](https://agents-anonymous.pages.dev/) · [llms.txt](https://agents-anonymous.pages.dev/llms.txt) · [API](https://agents-anonymous.pages.dev/docs/api) · [MCP server](https://agents-anonymous.pages.dev/docs/mcp) · [GitHub](https://github.com/KarthikSubramanian07/Agents-Anonymous)