token-config-checker
# token-config-checker
[简体中文](#简体中文) | [English](#english)
---
## 简体中文
用于检测包含 `access_token` / `refresh_token` / `id_token` 的 JSON 配置文件是否还有效。
### 提供内容
- `scripts/check_token_configs.py`:基础版,支持离线检测和可选在线探测
- `scripts/check_token_configs_plus.py`:增强版,支持脱敏输出、失败原因分类、三分类保存、写入报告文件
### 快速用法
#### 检测单个文件
```bash
python3 scripts/check_token_configs.py /path/to/file.json
```
#### 批量检测目录
```bash
python3 scripts/check_token_configs.py /path/to/dir
```
#### 在线探测 access_token
```bash
python3 scripts/check_token_configs.py /path/to/file.json --probe
```
#### 增强版:按 valid / no_quota / invalid 三分类保存
```bash
python3 scripts/check_token_configs_plus.py ./tokens \
--probe \
--probe-url https://api.openai.com/v1/models \
--save-valid-dir ./valid_configs \
--save-no-quota-dir ./no_quota_configs \
--save-invalid-dir ./invalid_configs \
--group-by-type \
--out report.txt
```
### 输出重点
- `valid_offline`:离线结构和过期时间是否正常
- `online_valid`:在线探测是否通过
- `failure_class`:失败分类,例如 `token_expired` / `online_invalid_or_revoked`
- `bucket`:最终分类(`valid` / `no_quota` / `invalid`)
### 注意事项
- 这类配置文件非常敏感,不要分享原始文件。
- 对 Codex 配置,在线结果通常比离线判断更重要。
- 默认探测地址使用 OpenAI 官方接口;如果你使用私有中转或自定义面板,请手动传入 `--probe-url`。
- 建议开启 `--out` 输出报告,便于复查。
---
## English
Use this skill to validate JSON config files containing `access_token` / `refresh_token` / `id_token`.
### Included scripts
- `scripts/check_token_configs.py`: basic version with offline checks and optional online probing
- `scripts/check_token_configs_plus.py`: enhanced version with redacted output, failure classification, three-bucket saving, and report generation
### Quick usage
#### Check one file
```bash
python3 scripts/check_token_configs.py /path/to/file.json
```
#### Batch scan a directory
```bash
python3 scripts/check_token_configs.py /path/to/dir
```
#### Online probe access_token
```bash
python3 scripts/check_token_configs.py /path/to/file.json --probe
```
#### Enhanced: save configs into valid / no_quota / invalid buckets
```bash
python3 scripts/check_token_configs_plus.py ./tokens \
--probe \
--probe-url https://api.openai.com/v1/models \
--save-valid-dir ./valid_configs \
--save-no-quota-dir ./no_quota_configs \
--save-invalid-dir ./invalid_configs \
--group-by-type \
--out report.txt
```
### Key outputs
- `valid_offline`: whether the file looks structurally valid offline
- `online_valid`: whether online probing succeeded
- `failure_class`: failure reason classification
- `bucket`: final bucket (`valid` / `no_quota` / `invalid`)
### Notes
- These files are highly sensitive. Do not share raw configs.
- For Codex-style configs, online probe results are usually more important than offline checks.
- Default probe target uses the OpenAI official endpoint; if you use a private relay or custom panel, pass your own `--probe-url` explicitly.
- Prefer `--out` to generate a reusable report.
标签
skill
ai