You are an agent, your current working directory is /app.

You can use the tools available to you to interact with the computer to assist the user in completing tasks.

# Repair nonce-derived AES-CCM message limits

Work in `/app/vendor/pycryptodome`. The frozen source already includes its local Python modules and ABI3 native extensions, so the task must run offline with `PYTHONPATH=/app/vendor/pycryptodome/lib`.

CCM encodes the message length in `Q = 15 - len(nonce)` bytes. The current implementation can initialize or process a message whose length cannot be represented in those `Q` bytes. Repair the existing source tree rather than adding a replacement cipher or bypassing its APIs.

## Required source repairs

Modify both existing modules:

- `/app/vendor/pycryptodome/lib/Crypto/Cipher/_mode_ccm.py`
- `/app/vendor/pycryptodome/lib/Crypto/Cipher/_mode_ccm.pyi`

The repaired implementation must:

1. Expose `CCMMessageTooLongError`, a `ValueError` subclass, from the CCM mode module and its typing stub.
2. Reject a declared `msg_len` when `msg_len >= 2 ** (8 * Q)` before CCM MAC/counter state is initialized.
3. When `msg_len` is omitted, reject an oversized first `encrypt()` or `decrypt()` input with the same exception before producing or accepting output. An `output=` buffer must remain unchanged on rejection.
4. Derive the bound from the actual nonce length. For example, a 13-byte nonce cannot encode 65536 bytes, while a 12-byte nonce can.
5. Preserve the public `AES.new(..., AES.MODE_CCM, ...)` API, valid associated-data handling, streaming behavior, ciphertext/tag values, and decrypt/verify behavior.

Do not disable or replace the vendored tests, native modules, AES dispatch, or regression runner.

## Artifact

After repairing the source, run:

```bash
cd /app
PYTHONPATH=/app/vendor/pycryptodome/lib python3 /app/regression/run_ccm_regression.py
```

This must create `/app/ccm_repair_report.json` with schema version `tbench.ccm_repair_report.v1` and all checks passing. The private verifier independently recomputes the cryptographic and boundary behavior from trusted fixtures.