unsloth/studio/backend/core/inference
Daniel Han 2ae885ce72 studio: address sandbox hardening review findings
Round-2 fixes for ten issues surfaced by a 20-reviewer code review of
the initial hardening patches. Every change is detection-widening or a
false-positive narrowing; legitimate tool calls keep working.

Direct Python open now flows through _find_sensitive_paths so
open('/home/u/.aws/credentials').read() is gated the same as
os.system('cat ~/.aws/credentials'). The previous wiring covered only
the bash and shell-exec sides.

Both SignalEscapeVisitor and NetworkAndIoVisitor fail-closed once the
eval / exec literal recursion cap is reached. Wrapping a payload in
four or more nested literal exec layers no longer silently bypasses
inspection.

_find_sensitive_paths scans three projections of the command (raw,
backslash-normalised, shlex-dequoted) and recurses into nested
bash -c and cmd /c shells. Quote-spliced and Windows-backslash forms
of credential paths are all caught.

_HOME_PREFIX_RE adds Windows-style home prefixes (USERPROFILE,
HOMEDRIVE HOMEPATH, env:USERPROFILE, drive-letter Users) so cross-OS
hardening actually applies on Windows. Both sensitive-path regexes
now have a path-token start anchor so project-local lookalike paths
under workspace, fixtures, and tmp are not blocked.

Network host validation for sock.connect and the requests / urllib
FQ-prefix branch now use _extract_string_from_node instead of raw
ast.Constant checks, so concatenated and f-string literal hosts
resolve the same way the open gate already did.

pathlib.Path('/etc/shadow').open() is now inspected; the path is
extracted from the receiver constructor when node.args is empty.

The static-string resolver depth cap moves from 6 to 64, removing the
single-character literal-concat bypass while leaving the recursion
well inside CPython's default frame limit.

The SSH private-key regex gains a filename-end boundary so reading a
public ".pub" key stays allowed (legit developer action) while the
matching private key is still denied.

New regression tests cover one class per finding (TestFinding1 through
TestFinding10) plus updated nested-depth coverage; the previous
test_nested_depth_does_not_crash assertion was inverted by the
fail-closed change and has been replaced. Local sweep: 336 passed
(131 upstream + 205 hardening).
2026-05-19 11:07:16 +00:00
..
__init__.py Final cleanup 2026-03-12 18:28:04 +00:00
_html_to_md.py fix: studio web search SSL failures and empty page content (#4754) 2026-04-01 06:12:02 -07:00
anthropic_compat.py Studio: support images on /v1/messages (Anthropic-compat) (#5128) 2026-04-22 03:25:07 +04:00
audio_codecs.py Add native GGUF intake to Studio (#5246) 2026-05-04 11:46:18 +02:00
defaults.py studio: engage draft-mtp on vision MTP GGUFs (drop incorrect vision gate) (#5560) 2026-05-18 08:42:55 -07:00
external_provider.py fix(studio): handle expired OpenAI shell-tool containers without surfacing error in chat (#5547) 2026-05-18 05:47:57 -07:00
inference.py Pin bitsandbytes to continuous-release_main on ROCm (4-bit decode fix) (#4954) 2026-04-10 06:25:39 -07:00
key_exchange.py studio: API external provider support for chat (OpenAI, Mistral, Gemini, Cohere, Anthropic, OpenRouter, DeepSeek, custom providers) (#4706) 2026-05-14 16:13:59 +04:00
llama_cpp.py studio: emit one comma-chained --spec-type for CPU/Mac MTP path (#5575) 2026-05-19 03:16:05 -07:00
llama_server_args.py Studio: auto-enable MTP speculative decoding for MTP GGUFs (#5527) 2026-05-18 00:15:42 -07:00
mlx_inference.py MLX training support for Studio on Apple Silicon (#5340) 2026-05-14 05:24:20 -07:00
orchestrator.py Add native GGUF intake to Studio (#5246) 2026-05-04 11:46:18 +02:00
providers.py Polish/cloud to providers (#5450) 2026-05-15 19:29:21 +04:00
tools.py studio: address sandbox hardening review findings 2026-05-19 11:07:16 +00:00
worker.py studio: load cached GGUF models when fully offline (#5505) 2026-05-17 21:25:39 -07:00