Files
hermes-agent/scripts/check_subprocess_stdin.py
Siddharth Balyan f9102e4ca4 Setup scan stays off backend globals; audit lint and RPC fixes (#129341)
* fix: userscan tmux and zellij socket probes pass the /tmp literal check

The four lines read where tmux and zellij put their sockets (and, on macOS,
measure the user's own temp dir). The POSIX fallback for those directories
is /tmp, so the literal is the fact being probed, not a scratch path. Mark
each line so the blocking lint step in lint.yml passes.

* fix: setup-profile and start_chat RPCs no longer block the RPC reader

onboarding.ensure_setup_profile / ensure_setup_session / reset_setup_profile
can run create_profile (a skill-tree copy), state.db writes and the first
import of the setup scanner. session.start_chat creates a session and runs
prompt.submit. Inline on the reader thread they left session.interrupt and
approval.respond unread in the pipe, the failure the _LONG_HANDLERS comment
names. profiles.create was already on the list for the same reason.

* fix: userscan children run with stdin closed and the served profile's env

The /initiate-setup builder runs the user scan on a backend thread, so
every child the scan spawned inherited the JSON-RPC stdin pipe (#14036)
and the launch profile's os.environ. HostAccess.run, copy_locked and the
PowerShell batch now pass stdin=subprocess.DEVNULL, and every spawn takes
its env from HostAccess.child_env(). host_facts passes
served_profile_child_env() into the run; the standalone run.py keeps the
process env.

check_subprocess_stdin.py now scans optional-skills/: the backend loads
initiate-setup's scripts in-process. The nested-dir skip matches paths
relative to the context dir, so a skill's own scripts/ folder is in
scope. Three model-run optional-skill scripts it flagged get stdin closed.

* refactor: split the userscan spec modules over 2,000 lines by topic

Six spec modules under initiate-setup's userscan passed the 2,000-line
gate. Each keeps its shared helpers and first families; the rest move
verbatim into <stem>_<topic>.py siblings that import the helpers they
use from the base module:

  apps_dev      -> apps_dev_agents, apps_dev_devenv
  browser_files -> browser_files_comms, browser_files_content
  usage_gaming  -> usage_gaming_games, usage_gaming_hardware
  linux_system  -> linux_system_security, linux_system_hardware
  linux_apps    -> linux_apps_dev, linux_apps_browser, linux_apps_files
  darwin_apps   -> darwin_apps_dev, darwin_apps_browser, darwin_apps_files

linux_system keeps health and usage: install_age reads usage's _wtmp and
usage reads health's _kdump_dirs. References through the ad./bf./ug.
aliases point at the module that now defines the name. specs/__init__
imports each sibling right after its base, so per-OS registration order
and the registry keys stay the same.

No behaviour change: `run.py --list` is byte-identical before and after,
and a temp-home scan under --os darwin/linux/windows gives the same fact
statuses, errors and insights.

* fix: the setup scan never patches process globals or sys.path in the backend

The /initiate-setup builder runs the user scan on a backend thread. It put
the scanner dir at sys.path[0] of the backend (so a top-level `run` or
`userscan` import anywhere resolved to the scanner), and it loaded
host.py's deny_reads_under (swaps builtins.open, io.open, sqlite3.connect
and subprocess.Popen) and override_env (rewrites os.environ HOME and
HERMES_HOME) into the backend's import graph.

- userscan/cli.py now holds override_env, the read guard, the aggregate
  HostAccess and scan_accounts. Only run.py imports it; runner.run() no
  longer retargets os.environ (overrides stay l0 data), and run.py applies
  override_env around run() for --home and friends.
- The package uses relative imports, and host_facts loads it once per
  process as initiate_setup_userscan, without touching sys.path.

Live: after an in-process scan sys.path, os.environ and the four patched
callables are unchanged and no top-level userscan module exists (before:
sys.path[0] was the scanner dir and userscan.host was importable).
`run.py --list` is unchanged; --all-users and --home give the same output.

* refactor: userscan runner runs each layer as a named phase

runner._run was one closure-heavy function at cyclomatic complexity 66.
Its state moves onto a small _Pass object whose run_l1 / run_l2 (with
_run_python_l2) / run_l3 methods each hold one layer, moved verbatim;
_run composes them and builds the output. Highest CC in the file is now
21 (run_l3).

No behaviour change: temp-home scans under --os darwin/linux/windows give
the same fact ids, statuses, methods, tiers, coverage, insights and
output keys as before.

* chore: name the setup intro flag writer, document gui_tour presets, trim media.ts

- hermes_cli/setup_profile.py::mark_seen becomes mark_intro_seen. It
  writes the setup profile's intro state; agent/onboarding.py::mark_seen
  flips a config.yaml hint flag, and tui_gateway uses both. The RPC name
  onboarding.mark_seen is unchanged.
- tools-reference: gui_tour's `start` without `steps` runs the built-in
  tour (`preset` quick or full, default full).
- apps/desktop/src/lib/media.ts: drop the blank line at EOF (git diff --check).

* docs: describe the setup chat's automatic machine scan

The scan runs without a consent step when the desktop app creates the
setup profile (the owner kept it automatic). The desktop user guide now
says where it runs (the backend's machine), what it reads (T1: counts and
categories; finer values are dropped before storage), what reaches the
model (the derived fact block in the first setup message, sent to the
setup chat's inference provider) and where the cache lives
(insights/profile.json in the setup profile's home, 0600, 24 h).

The skill's own docs page is generated from SKILL.md, which is the model's
prompt, so the user-facing text lives in the desktop guide.

* fix: overlapping setup-profile RPCs no longer race on the RPC pool

ensure_setup_profile, ensure_setup_session and reset_setup_profile moved off the reader thread, so two kickoffs could overlap: both saw no setup profile, and the losers failed on the half-made profile (5 of 6 barrier-started calls returned ENOENT), or both saw an empty setup chat and seeded it twice. One module lock restores the one-at-a-time order; the prescan stays outside it.

* fix: the Linux locale probe reads date formats in a child, not via setlocale

locale.user ran setlocale(LC_TIME) on the scan thread inside the backend. setlocale is process-wide and not thread-safe, so strftime in other session threads briefly used the user's LC_TIME. The probe now runs 'locale d_fmt t_fmt' under LC_ALL=<locale> through HostAccess.run (new env overlay on child_env), after checking 'locale -a', because locale prints C formats for a locale that is not installed.

* docs: link the setup scan from the /initiate-setup and onboarding references

The desktop guide's scan section now says which search-engine names are kept and that the scan runs in the CLI and the gateway too. The /initiate-setup row and the profile_build bullet link to it.
2026-09-30 19:45:19 +05:30

275 lines
10 KiB
Python

#!/usr/bin/env python3
"""Check that subprocess calls in TUI-context code specify stdin=.
When Hermes runs in TUI mode, the gateway child process communicates with
the Node.js parent over a JSON-RPC protocol on stdin. Subprocess calls that
inherit this fd can cause the gateway to exit with stdin EOF during tool
execution (issue #14036, PR #39257).
This script checks that all subprocess.run() and subprocess.Popen() calls
in TUI-context files (agent/, tools/, plugins/, tui_gateway/, optional-skills/) explicitly
set stdin= to prevent fd inheritance.
Exit codes:
0 — all calls are safe
1 — violations found
2 — script error
Usage:
python scripts/check_subprocess_stdin.py [--fix]
With --fix, prints the commands to add stdin=subprocess.DEVNULL to each
violation (does not modify files).
"""
from __future__ import annotations
import ast
import os
import re
import sys
from pathlib import Path
# Directories that run inside the TUI gateway child process.
TUI_CONTEXT_DIRS = [
"agent/",
"tools/",
"plugins/",
"tui_gateway/",
# The backend loads some optional-skill scripts in-process: the /initiate-setup
# builder runs initiate-setup's host_facts.py and its userscan package on a
# backend thread, so their children would inherit the JSON-RPC stdin too.
"optional-skills/",
]
# User plugin roots — scanned at runtime if they exist. Plugins load from
# ``get_hermes_home() / "plugins"`` (user) and ``./.hermes/plugins/`` (project,
# gated behind ``HERMES_ENABLE_PROJECT_PLUGINS``) — see
# ``hermes_cli/plugins.py:10-12``. The guard only checked the bundled
# ``plugins/`` dir, missing user-installed code that spawns subprocesses
# (gap reported in #67639).
#
# Import is deferred to ``main()`` (after ``os.chdir(repo_root)``) because
# this script runs as a standalone subprocess — ``hermes_constants`` isn't
# on ``sys.path`` until the repo root is added.
# subprocess and os APIs that inherit stdin by default when called without
# an explicit stdin= argument. The original regex only covered run/Popen
# (gap #1 in #67639); call, check_output, check_call, os.system, and
# asyncio.create_subprocess_* all inherit fd 0 equally.
_SUBPROCESS_PATTERNS = [
r"subprocess\.(run|Popen|call|check_output|check_call)\s*\([\"'a-zA-Z_\[\(]",
r"os\.system\s*\([\"'a-zA-Z_\[\(]",
r"asyncio\.create_subprocess_(exec|shell)\s*\([\"'a-zA-Z_\[\(]",
]
# Files with intentional stdin= override (e.g. input= creates a pipe).
# Format: "filepath:line" or just "filepath" to skip the whole file.
KNOWN_SAFE = {
"agent/shell_hooks.py", # uses input=stdin_json, creates a pipe
"plugins/security-guidance/patterns.py", # subprocess mentions are in reminder strings, not calls
}
# Inline marker that exempts a single subprocess call from this check.
# Put it in a comment on (or within) the call when the process MUST inherit
# stdin — e.g. an interactive login the user explicitly invokes. Travels with
# the line, so it survives edits that shift line numbers (unlike a pinned
# file:line entry).
EXEMPT_MARKER = "noqa: subprocess-stdin"
# Directory names skipped at any depth below a context dir. Matching against the
# path relative to that dir keeps a skill's own ``scripts/`` folder in scope.
SKIP_DIRS = {"tests"}
_SPLAT_RE = re.compile(r"\*\*\s*([A-Za-z_][A-Za-z0-9_]*)")
def _splat_carries_stdin(call_text: str, content: str) -> bool:
"""True when the call splats ``**name`` / ``**name(...)`` and ``name`` is defined in
the same file (assignment or ``def``) whose OWN expression/body sets ``stdin=``.
Shared kwargs helpers (``_RUN_KW = dict(..., stdin=DEVNULL)``, ``def _run_kwargs(): return
dict(..., stdin=DEVNULL)``) legitimately carry the guard; we only accept them when the
definition provably sets stdin= — never on the helper's name alone, and never because an
unrelated later call in the file happens to pass ``stdin=``.
"""
names = set(_SPLAT_RE.findall(call_text))
if not names:
return False
try:
tree = ast.parse(content)
except SyntaxError:
return False
for name in names:
node = None
for n in ast.walk(tree):
if isinstance(n, (ast.FunctionDef, ast.AsyncFunctionDef)) and n.name == name:
node = n
break
if isinstance(n, (ast.Assign, ast.AnnAssign)):
targets = n.targets if isinstance(n, ast.Assign) else [n.target]
if any(isinstance(t, ast.Name) and t.id == name for t in targets):
node = n.value if n.value is not None else n
break
if node is None:
return False
# stdin appears as a keyword (dict(stdin=...)) or as a dict-literal key ({"stdin": ...})
# somewhere INSIDE this definition — not merely nearby in the file.
has = any(
(isinstance(sub, ast.keyword) and sub.arg == "stdin")
or (isinstance(sub, ast.Constant) and sub.value == "stdin")
for sub in ast.walk(node)
)
if not has:
return False
return True
def find_subprocess_calls(content: str, filepath: str) -> list[dict]:
"""Find all subprocess/os/asyncio calls missing stdin= in content."""
violations = []
lines = content.split("\n")
# Match only actual function calls — not comments, docstrings, or prose.
# Multiple patterns cover subprocess.run/Popen/call/check_output/check_call,
# os.system, and asyncio.create_subprocess_exec/shell.
patterns = [re.compile(p) for p in _SUBPROCESS_PATTERNS]
for i, line in enumerate(lines):
# Skip comments.
stripped = line.lstrip()
if stripped.startswith("#"):
continue
# Skip lines where the match is inside backticks (docstring references).
if "``subprocess" in line:
continue
if not any(p.search(line) for p in patterns):
continue
# Collect the full call (may span multiple lines).
call_start = i
paren_depth = 0
found_open = False
call_lines = []
for j in range(i, min(i + 30, len(lines))):
call_lines.append(lines[j])
for ch in lines[j]:
if ch == "(":
paren_depth += 1
found_open = True
elif ch == ")":
paren_depth -= 1
if found_open and paren_depth == 0:
call_text = "\n".join(call_lines)
# Already has stdin= → safe.
if "stdin=" in call_text:
break
# Has input= → creates a pipe, safe.
if "input=" in call_text:
break
# Splats a same-file kwargs helper whose definition
# sets stdin= → the guard travels with the helper.
if _splat_carries_stdin(call_text, content):
break
# Inline exemption marker on the call itself or within
# the few comment lines immediately above it → the call
# intentionally inherits stdin.
window_start = max(0, i - 4)
preceding = "\n".join(lines[window_start:i])
if EXEMPT_MARKER in call_text or EXEMPT_MARKER in preceding:
break
violations.append({
"file": filepath,
"line": i + 1,
"snippet": line.strip()[:120],
})
break
else:
continue
break
return violations
def main() -> int:
fix_mode = "--fix" in sys.argv
repo_root = Path(__file__).resolve().parent.parent
os.chdir(repo_root)
# Add repo root to sys.path so we can import hermes_constants (this script
# runs as a standalone subprocess, not as a module).
sys.path.insert(0, str(repo_root))
from hermes_constants import get_hermes_home
all_violations = []
for tui_dir in TUI_CONTEXT_DIRS:
dirpath = repo_root / tui_dir
if not dirpath.exists():
continue
for py_file in dirpath.rglob("*.py"):
rel = str(py_file.relative_to(repo_root))
# Skip known-safe files. ``relative_to`` returns a host-separated
# path (backslashes on Windows) while KNOWN_SAFE is forward-slash.
if py_file.relative_to(repo_root).as_posix() in KNOWN_SAFE:
continue
# Skip test files inside tools/ etc.
if SKIP_DIRS & set(py_file.relative_to(dirpath).parts):
continue
content = py_file.read_text(encoding="utf-8-sig")
violations = find_subprocess_calls(content, rel)
all_violations.extend(violations)
# Scan user plugin directories (Gap 1: guard missed user-installed
# plugins in get_hermes_home()/plugins/ and project plugins in
# ./.hermes/plugins/, where code like ori/hooks.py can spawn
# subprocesses with inherited stdin — #67639).
plugin_roots: list[Path] = [get_hermes_home() / "plugins"]
if os.environ.get("HERMES_ENABLE_PROJECT_PLUGINS"):
plugin_roots.append(Path.cwd() / ".hermes" / "plugins")
seen_roots: set[Path] = set()
for plugin_root in plugin_roots:
resolved = plugin_root.resolve()
if resolved in seen_roots or not resolved.is_dir():
continue
seen_roots.add(resolved)
for py_file in resolved.rglob("*.py"):
rel = str(py_file)
if py_file.name in ("conftest.py",) or "/tests/" in rel:
continue
try:
content = py_file.read_text(encoding="utf-8-sig")
except Exception:
continue
violations = find_subprocess_calls(content, rel)
all_violations.extend(violations)
if all_violations:
print(f"❌ {len(all_violations)} subprocess calls missing stdin=:")
for v in all_violations:
print(f" {v['file']}:{v['line']}: {v['snippet']}")
if fix_mode:
print("\nAdd stdin=subprocess.DEVNULL to each call above.")
return 1
else:
print("✅ All TUI-context subprocess calls have explicit stdin=")
return 0
if __name__ == "__main__":
sys.exit(main())