Summary
The execute_code tool's subprocess sandbox advertises a three-layer defense (AST validation, text-pattern blocklist, restricted __builtins__). In sandbox mode (the default) only two layers are active — the text-pattern blocklist is skipped — and both remaining layers are bypassed by combining two CPython semantics:
- Runtime string assembly. The AST validator (
src/praisonai-agents/praisonaiagents/tools/python_tools.py:75) enumerates blocked dunder names against ast.Attribute.attr, ast.Call.func.id, and ast.Constant string-substring. Names assembled at runtime (e.g. "_"*2 + "class" + "_"*2) appear in the AST as multiple short ast.Constant nodes, none containing a blocked substring, so the static check passes.
- C-level attribute access via format-spec.
str.format / str.format_map resolve dotted field references through CPython's internal PyObject_GetAttr (do_string_format → get_field). This C path never consults the Python-level getattr binding. The sandbox's _safe_getattr wrapper (python_tools.py:221) is installed only as the getattr name in safe_builtins, so any C-level attribute access — including format-spec field resolution — sidesteps it. format/format_map are also absent from _SANDBOX_BLOCKED_CALLS (python_tools.py:56).
Combined, this yields an arbitrary read primitive over every blocklisted attribute (__class__, __qualname__, __bases__, __base__, function __globals__, __dict__, …).
Affected code
| File |
Lines |
Symbol |
Role |
src/praisonai-agents/praisonaiagents/tools/python_tools.py |
39–54 |
_SANDBOX_BLOCKED_ATTRS |
The blocklist the bypass renders unreachable |
src/praisonai-agents/praisonaiagents/tools/python_tools.py |
56–60 |
_SANDBOX_BLOCKED_CALLS |
Missing entries: format, format_map |
src/praisonai-agents/praisonaiagents/tools/python_tools.py |
75–102 |
_validate_code_ast |
Static check, blind to runtime string assembly |
src/praisonai-agents/praisonaiagents/tools/python_tools.py |
221–226 |
_safe_getattr |
Wraps Python-level getattr only; C-level access bypasses |
src/praisonai-agents/praisonaiagents/tools/python_tools.py |
352 |
execute_code |
Entry point; gated by @require_approval(risk_level="critical") |
Reproducer
import os
os.environ["PRAISONAI_AUTO_APPROVE"] = "true"
from praisonaiagents.tools.python_tools import execute_code
payload = '''
und = "_" * 2 # "__" assembled at runtime
key1 = und + "class" + und # "__class__"
key2 = und + "qualname" + und # "__qualname__"
fmt_class = "{0." + key1 + "}"
fmt_qual2 = "{0." + key1 + "." + key2 + "}"
print("LEAK_CLASS=" + fmt_class.format(()))
print("LEAK_QUAL2=" + fmt_qual2.format(()))
'''
print(execute_code(payload, sandbox_mode="sandbox", timeout=15))
Observed: stdout = LEAK_CLASS=<class 'tuple'> / LEAK_QUAL2=tuple, success: true, no security error. Both __class__ (one hop) and __class__.__qualname__ (two hops) — all blocklisted — are read.
Trust boundary
The @require_approval(risk_level="critical") gate is bypassed when PRAISONAI_AUTO_APPROVE is set (verified: require_approval short-circuits on is_env_auto_approve()). That variable is set by the project's FULL_AUTO autonomy mode, the bots-CLI launchers, and the project's own issue-triage CI workflow — postures where the agent reaches execute_code with no human approval. The payload then arrives via any LLM-visible surface (user message, retrieved document, tool/web/MCP output) and the tool-call machinery passes it as the code argument.
The code's own comment at python_tools.py:46 cites GHSA-4mr5-g6f9-cfrh, which added __self__ to the blocklist to stop C-builtins leaking builtins via func.__self__. This finding does not bypass that single entry — it bypasses the entire blocklist, because format-spec attribute resolution never consults the blocklist or _safe_getattr. "{0.__self__}".format(print) would leak __self__ regardless of the blocklist. Same defense surface, different mechanism; the GHSA-4mr5 fix does not mitigate this.
Scope (read primitive only)
This reports the read primitive. Turning the read into in-process execution requires a callable bridge; the obvious one (string.Formatter().get_field() returning the live object) is not directly reachable because import string is blocked at the AST layer (no ast.Import). Other bridges may exist; a full execution chain is not claimed here. If one is found, severity rises to ~8.8 (the subprocess has no seccomp/setrlimit/syscall filtering).
Suggested fix
- Add
format, format_map to _SANDBOX_BLOCKED_CALLS (blocks the calls at the AST layer; cost: also blocks benign str.format).
- Or replace
str in safe_builtins with a subclass whose format/format_map reject dotted fields resolving to leading-underscore attributes (preserves benign formatting).
- Or drop sandbox-mode's in-process security claim and document that real isolation requires external sandboxing (gVisor/firejail/container/microVM) — which matches what the subprocess provides today.
The text-pattern blocklist present in the direct path (python_tools.py:487-502) is absent from the sandbox path; even if added, the runtime-assembly trick defeats it, so (1) or (2) is required.
Reporter: Kai Aizen / SnailSploit — [email protected] — PGP on request. Coordinated disclosure; no public posting.
References
Summary
The
execute_codetool's subprocess sandbox advertises a three-layer defense (AST validation, text-pattern blocklist, restricted__builtins__). In sandbox mode (the default) only two layers are active — the text-pattern blocklist is skipped — and both remaining layers are bypassed by combining two CPython semantics:src/praisonai-agents/praisonaiagents/tools/python_tools.py:75) enumerates blocked dunder names againstast.Attribute.attr,ast.Call.func.id, andast.Constantstring-substring. Names assembled at runtime (e.g."_"*2 + "class" + "_"*2) appear in the AST as multiple shortast.Constantnodes, none containing a blocked substring, so the static check passes.str.format/str.format_mapresolve dotted field references through CPython's internalPyObject_GetAttr(do_string_format→get_field). This C path never consults the Python-levelgetattrbinding. The sandbox's_safe_getattrwrapper (python_tools.py:221) is installed only as thegetattrname insafe_builtins, so any C-level attribute access — including format-spec field resolution — sidesteps it.format/format_mapare also absent from_SANDBOX_BLOCKED_CALLS(python_tools.py:56).Combined, this yields an arbitrary read primitive over every blocklisted attribute (
__class__,__qualname__,__bases__,__base__, function__globals__,__dict__, …).Affected code
src/praisonai-agents/praisonaiagents/tools/python_tools.py_SANDBOX_BLOCKED_ATTRSsrc/praisonai-agents/praisonaiagents/tools/python_tools.py_SANDBOX_BLOCKED_CALLSformat,format_mapsrc/praisonai-agents/praisonaiagents/tools/python_tools.py_validate_code_astsrc/praisonai-agents/praisonaiagents/tools/python_tools.py_safe_getattrgetattronly; C-level access bypassessrc/praisonai-agents/praisonaiagents/tools/python_tools.pyexecute_code@require_approval(risk_level="critical")Reproducer
Observed:
stdout=LEAK_CLASS=<class 'tuple'>/LEAK_QUAL2=tuple,success: true, no security error. Both__class__(one hop) and__class__.__qualname__(two hops) — all blocklisted — are read.Trust boundary
The
@require_approval(risk_level="critical")gate is bypassed whenPRAISONAI_AUTO_APPROVEis set (verified:require_approvalshort-circuits onis_env_auto_approve()). That variable is set by the project's FULL_AUTO autonomy mode, the bots-CLI launchers, and the project's own issue-triage CI workflow — postures where the agent reachesexecute_codewith no human approval. The payload then arrives via any LLM-visible surface (user message, retrieved document, tool/web/MCP output) and the tool-call machinery passes it as thecodeargument.Relationship to GHSA-4mr5-g6f9-cfrh
The code's own comment at
python_tools.py:46cites GHSA-4mr5-g6f9-cfrh, which added__self__to the blocklist to stop C-builtins leakingbuiltinsviafunc.__self__. This finding does not bypass that single entry — it bypasses the entire blocklist, because format-spec attribute resolution never consults the blocklist or_safe_getattr."{0.__self__}".format(print)would leak__self__regardless of the blocklist. Same defense surface, different mechanism; the GHSA-4mr5 fix does not mitigate this.Scope (read primitive only)
This reports the read primitive. Turning the read into in-process execution requires a callable bridge; the obvious one (
string.Formatter().get_field()returning the live object) is not directly reachable becauseimport stringis blocked at the AST layer (noast.Import). Other bridges may exist; a full execution chain is not claimed here. If one is found, severity rises to ~8.8 (the subprocess has no seccomp/setrlimit/syscall filtering).Suggested fix
format,format_mapto_SANDBOX_BLOCKED_CALLS(blocks the calls at the AST layer; cost: also blocks benignstr.format).strinsafe_builtinswith a subclass whoseformat/format_mapreject dotted fields resolving to leading-underscore attributes (preserves benign formatting).The text-pattern blocklist present in the
directpath (python_tools.py:487-502) is absent from the sandbox path; even if added, the runtime-assembly trick defeats it, so (1) or (2) is required.Reporter: Kai Aizen / SnailSploit — [email protected] — PGP on request. Coordinated disclosure; no public posting.
References