request-cache-py

Details

**Severity:** Critical

**Affected versions:** `= 1.0.0`

## Source: amazon-inspector (eafb96e46544cb1351d26caf52bff79055bc205a1f8454737b677fff8fbc6fea)
request-cache-py impersonates the legitimate requests-cache HTTP caching library. On `import request_cache_py`, the package's `__init__.py` starts a background thread that harvests installer-side secrets and POSTs them to a hardcoded attacker Telegram bot.

Observed behaviors:
- Reads private keys and credentials from `~/.ssh/` (`id_rsa`, `id_ed25519`, `id_ecdsa`, `id_dsa`), `~/.aws/credentials`, `~/.aws/config`, `~/.gitconfig`, `~/.git-credentials`, `~/.npmrc`, `~/.pypirc`, `~/.dockercfg`, `~/.docker/config.json`, plus gcloud and vscode settings.
- Copies Chrome/Edge/Safari `Login Data`, `Cookies`, and `History` SQLite databases to `/tmp` and extracts saved logins, cookies, and browsing history (`SELECT origin_url, username_value FROM logins`; `SELECT host_key, name, value, path FROM cookies`).
- Iterates `os.environ` and exfiltrates any variable whose name contains `key`, `token`, `secret`, `password`, `api`, or `auth`.
- Posts the collected data to `https://api.telegram.org/bot<redacted>/sendMessage` with a fixed `chat_id`. The bot token and chat id are base64-split across pieces and reassembled at runtime to evade scanners.
- Includes sandbox-evasion: `_should_skip()` aborts when `CI`, `GITHUB_ACTIONS`, `TRAVIS`, `JENKINS_HOME`, `CIRCLECI`, `/.dockerenv`, or hypervisor markers are present, restricting execution to real developer workstations. A `~/.cache/.pyrc` marker suppresses repeat sends within 24 hours.

The combination — name impersonation of a popular library, import-time credential harvest from classic developer secret paths, browser database theft, env scraping, base64-obfuscated C2, sandbox evasion — is a deliberate supply-chain credential stealer targeting human developers.

## Source: kam193 (d027c4b6379310432f96b48dc78c73ddf1346052c5ab16ea6ed4fe3fc0754d08)
During import, package exfiltrates browsers data, SSH keys and other credential files, env variables and other sensitive data.


---

Category: M […]

**References:**
- https://bad-packages.kam193.eu/pypi/package/request-cache-py
- https://github.com/ossf/malicious-packages/blob/49d0cfba3689ed9b195d101d3a2a964c6a77f767/osv/malicious/pypi/request-cache-py/MAL-2026-6245.json
- https://pypi.org/project/request-cache-py/1.0.0
- https://pypi.org/project/request-cache-py/1.0.1
- https://pypi.org/project/request-cache-py/1.0.2
- https://pypi.org/project/request-cache-py/1.0.3
- https://pypi.org/project/request-cache-py/1.0.4
- https://pypi.org/project/request-cache-py/1.0.5

Compromised versions (1)

• = 1.0.0

Any computer that installed or ran a compromised version should be considered fully compromised. Rotate every secret on that machine from a clean environment.