Files
hermes-agent/tests/tools/test_local_env_blocklist.py
Teknium 6a320e8bfe fix(security): block sandbox backend creds from subprocess env (#1264)
* fix: prevent infinite 400 failure loop on context overflow (#1630)

When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message.  This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error.  Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.

Three-layer fix:

1. run_agent.py — Fallback heuristic: when a 400 error has a very short
   generic message AND the session is large (>40% of context or >80
   messages), treat it as a probable context overflow and trigger
   compression instead of aborting.

2. run_agent.py + gateway/run.py — Don't persist failed messages:
   when the agent returns failed=True before generating any response,
   skip writing the user's message to the transcript/DB. This prevents
   the session from growing on each failure.

3. gateway/run.py — Smarter error messages: detect context-overflow
   failures and suggest /compact or /reset specifically, instead of a
   generic 'try again' that will fail identically.

* fix(skills): detect prompt injection patterns and block cache file reads

Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):

1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
   (index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
   was the original injection vector — untrusted skill descriptions
   in the catalog contained adversarial text that the model executed.

2. skill_view: warns when skills are loaded from outside the trusted
   ~/.hermes/skills/ directory, and detects common injection patterns
   in skill content ("ignore previous instructions", "<system>", etc.).

Cherry-picked from PR #1562 by ygd58.

* fix(tools): chunk long messages in send_message_tool before dispatch (#1552)

Long messages sent via send_message tool or cron delivery silently
failed when exceeding platform limits. Gateway adapters handle this
via truncate_message(), but the standalone senders in send_message_tool
bypassed that entirely.

- Apply truncate_message() chunking in _send_to_platform() before
  dispatching to individual platform senders
- Remove naive message[i:i+2000] character split in _send_discord()
  in favor of centralized smart splitting
- Attach media files to last chunk only for Telegram
- Add regression tests for chunking and media placement

Cherry-picked from PR #1557 by llbn.

* fix(approval): show full command in dangerous command approval (#1553)

Previously the command was truncated to 80 chars in CLI (with a
[v]iew full option), 500 chars in Discord embeds, and missing entirely
in Telegram/Slack approval messages. Now the full command is always
displayed everywhere:

- CLI: removed 80-char truncation and [v]iew full menu option
- Gateway (TG/Slack): approval_required message includes full command
  in a code block
- Discord: embed shows full command up to 4096-char limit
- Windows: skip SIGALRM-based test timeout (Unix-only)
- Updated tests: replaced view-flow tests with direct approval tests

Cherry-picked from PR #1566 by crazywriter1.

* fix(cli): flush stdout during agent loop to prevent macOS display freeze (#1624)

The interrupt polling loop in chat() waited on the queue without
invalidating the prompt_toolkit renderer. On macOS, the StdoutProxy
buffer only flushed on input events, causing the CLI to appear frozen
during tool execution until the user typed a key.

Fix: call _invalidate() on each queue timeout (every ~100ms, throttled
to 150ms) to force the renderer to flush buffered agent output.

* fix(claw): warn when API keys are skipped during OpenClaw migration (#1580)

When --migrate-secrets is not passed (the default), API keys like
OPENROUTER_API_KEY are silently skipped with no warning. Users don't
realize their keys weren't migrated until the agent fails to connect.

Add a post-migration warning with actionable instructions: either
re-run with --migrate-secrets or add the key manually via
hermes config set.

Cherry-picked from PR #1593 by ygd58.

* fix(security): block sandbox backend creds from subprocess env (#1264)

Add Modal and Daytona sandbox credentials to the subprocess env
blocklist so they're not leaked to agent terminal sessions via
printenv/env.

Cherry-picked from PR #1571 by ygd58.

---------

Co-authored-by: buray <ygd58@users.noreply.github.com>
Co-authored-by: lbn <llbn@users.noreply.github.com>
Co-authored-by: crazywriter1 <53251494+crazywriter1@users.noreply.github.com>
2026-03-17 02:20:42 -07:00

291 lines
11 KiB
Python

"""Tests for subprocess env sanitization in LocalEnvironment.
Verifies that Hermes-managed provider, tool, and gateway env vars are
stripped from subprocess environments so external CLIs are not silently
misrouted or handed Hermes secrets.
See: https://github.com/NousResearch/hermes-agent/issues/1002
See: https://github.com/NousResearch/hermes-agent/issues/1264
"""
import os
import threading
from unittest.mock import MagicMock, patch
from tools.environments.local import (
LocalEnvironment,
_HERMES_PROVIDER_ENV_BLOCKLIST,
_HERMES_PROVIDER_ENV_FORCE_PREFIX,
)
def _make_fake_popen(captured: dict):
"""Return a fake Popen constructor that records the env kwarg."""
def fake_popen(cmd, **kwargs):
captured["env"] = kwargs.get("env", {})
proc = MagicMock()
proc.poll.return_value = 0
proc.returncode = 0
proc.stdout = MagicMock(__iter__=lambda s: iter([]), __next__=lambda s: (_ for _ in ()).throw(StopIteration))
proc.stdin = MagicMock()
return proc
return fake_popen
def _run_with_env(extra_os_env=None, self_env=None):
"""Execute a command via LocalEnvironment with mocked Popen
and return the env dict passed to the subprocess."""
captured = {}
fake_interrupt = threading.Event()
test_environ = {
"PATH": "/usr/bin:/bin",
"HOME": "/home/user",
"USER": "testuser",
}
if extra_os_env:
test_environ.update(extra_os_env)
env = LocalEnvironment(cwd="/tmp", timeout=10, env=self_env)
with patch("tools.environments.local._find_bash", return_value="/bin/bash"), \
patch("subprocess.Popen", side_effect=_make_fake_popen(captured)), \
patch("tools.terminal_tool._interrupt_event", fake_interrupt), \
patch.dict(os.environ, test_environ, clear=True):
env.execute("echo hello")
return captured.get("env", {})
class TestProviderEnvBlocklist:
"""Provider env vars loaded from ~/.hermes/.env must not leak."""
def test_blocked_vars_are_stripped(self):
"""OPENAI_BASE_URL and other provider vars must not appear in subprocess env."""
leaked_vars = {
"OPENAI_BASE_URL": "http://localhost:8000/v1",
"OPENAI_API_KEY": "sk-fake-key",
"OPENROUTER_API_KEY": "or-fake-key",
"ANTHROPIC_API_KEY": "ant-fake-key",
"LLM_MODEL": "anthropic/claude-opus-4-6",
}
result_env = _run_with_env(extra_os_env=leaked_vars)
for var in leaked_vars:
assert var not in result_env, f"{var} leaked into subprocess env"
def test_registry_derived_vars_are_stripped(self):
"""Vars from the provider registry (ANTHROPIC_TOKEN, ZAI_API_KEY, etc.)
must also be blocked — not just the hand-written extras."""
registry_vars = {
"ANTHROPIC_TOKEN": "ant-tok",
"CLAUDE_CODE_OAUTH_TOKEN": "cc-tok",
"ZAI_API_KEY": "zai-key",
"Z_AI_API_KEY": "z-ai-key",
"GLM_API_KEY": "glm-key",
"KIMI_API_KEY": "kimi-key",
"MINIMAX_API_KEY": "mm-key",
"MINIMAX_CN_API_KEY": "mmcn-key",
"DEEPSEEK_API_KEY": "deepseek-key",
}
result_env = _run_with_env(extra_os_env=registry_vars)
for var in registry_vars:
assert var not in result_env, f"{var} leaked into subprocess env"
def test_non_registry_provider_vars_are_stripped(self):
"""Extra provider vars not in PROVIDER_REGISTRY must also be blocked."""
extra_provider_vars = {
"GOOGLE_API_KEY": "google-key",
"MISTRAL_API_KEY": "mistral-key",
"GROQ_API_KEY": "groq-key",
"TOGETHER_API_KEY": "together-key",
"PERPLEXITY_API_KEY": "perplexity-key",
"COHERE_API_KEY": "cohere-key",
"FIREWORKS_API_KEY": "fireworks-key",
"XAI_API_KEY": "xai-key",
"HELICONE_API_KEY": "helicone-key",
}
result_env = _run_with_env(extra_os_env=extra_provider_vars)
for var in extra_provider_vars:
assert var not in result_env, f"{var} leaked into subprocess env"
def test_tool_and_gateway_vars_are_stripped(self):
"""Tool and gateway secrets/config must not leak into subprocess env."""
leaked_vars = {
"TELEGRAM_BOT_TOKEN": "bot-token",
"TELEGRAM_HOME_CHANNEL": "12345",
"DISCORD_HOME_CHANNEL": "67890",
"SLACK_APP_TOKEN": "xapp-secret",
"WHATSAPP_ALLOWED_USERS": "+15555550123",
"SIGNAL_ACCOUNT": "+15555550124",
"HASS_TOKEN": "ha-secret",
"EMAIL_PASSWORD": "email-secret",
"FIRECRAWL_API_KEY": "fc-secret",
"BROWSERBASE_PROJECT_ID": "bb-project",
"ELEVENLABS_API_KEY": "el-secret",
"GITHUB_TOKEN": "ghp_secret",
"GH_TOKEN": "gh_alias_secret",
"GATEWAY_ALLOW_ALL_USERS": "true",
"GATEWAY_ALLOWED_USERS": "alice,bob",
"MODAL_TOKEN_ID": "modal-id",
"MODAL_TOKEN_SECRET": "modal-secret",
"DAYTONA_API_KEY": "daytona-key",
}
result_env = _run_with_env(extra_os_env=leaked_vars)
for var in leaked_vars:
assert var not in result_env, f"{var} leaked into subprocess env"
def test_safe_vars_are_preserved(self):
"""Standard env vars (PATH, HOME, USER) must still be passed through."""
result_env = _run_with_env()
assert "HOME" in result_env
assert result_env["HOME"] == "/home/user"
assert "USER" in result_env
assert "PATH" in result_env
def test_self_env_blocked_vars_also_stripped(self):
"""Blocked vars in self.env are stripped; non-blocked vars pass through."""
result_env = _run_with_env(self_env={
"OPENAI_BASE_URL": "http://custom:9999/v1",
"MY_CUSTOM_VAR": "keep-this",
})
assert "OPENAI_BASE_URL" not in result_env
assert "MY_CUSTOM_VAR" in result_env
assert result_env["MY_CUSTOM_VAR"] == "keep-this"
class TestForceEnvOptIn:
"""Callers can opt in to passing a blocked var via _HERMES_FORCE_ prefix."""
def test_force_prefix_passes_blocked_var(self):
"""_HERMES_FORCE_OPENAI_API_KEY in self.env should inject OPENAI_API_KEY."""
result_env = _run_with_env(self_env={
f"{_HERMES_PROVIDER_ENV_FORCE_PREFIX}OPENAI_API_KEY": "sk-explicit",
})
assert "OPENAI_API_KEY" in result_env
assert result_env["OPENAI_API_KEY"] == "sk-explicit"
# The force-prefixed key itself must not appear
assert f"{_HERMES_PROVIDER_ENV_FORCE_PREFIX}OPENAI_API_KEY" not in result_env
def test_force_prefix_overrides_os_environ_block(self):
"""Force-prefix in self.env wins even when os.environ has the blocked var."""
result_env = _run_with_env(
extra_os_env={"OPENAI_BASE_URL": "http://leaked/v1"},
self_env={f"{_HERMES_PROVIDER_ENV_FORCE_PREFIX}OPENAI_BASE_URL": "http://intended/v1"},
)
assert result_env["OPENAI_BASE_URL"] == "http://intended/v1"
class TestBlocklistCoverage:
"""Sanity checks that the blocklist covers all known providers."""
def test_issue_1002_offenders(self):
"""Blocklist includes the main offenders from issue #1002."""
must_block = {
"OPENAI_BASE_URL",
"OPENAI_API_KEY",
"OPENROUTER_API_KEY",
"ANTHROPIC_API_KEY",
"LLM_MODEL",
}
assert must_block.issubset(_HERMES_PROVIDER_ENV_BLOCKLIST)
def test_registry_vars_are_in_blocklist(self):
"""Every api_key_env_var and base_url_env_var from PROVIDER_REGISTRY
must appear in the blocklist — ensures no drift."""
from hermes_cli.auth import PROVIDER_REGISTRY
for pconfig in PROVIDER_REGISTRY.values():
for var in pconfig.api_key_env_vars:
assert var in _HERMES_PROVIDER_ENV_BLOCKLIST, (
f"Registry var {var} (provider={pconfig.id}) missing from blocklist"
)
if pconfig.base_url_env_var:
assert pconfig.base_url_env_var in _HERMES_PROVIDER_ENV_BLOCKLIST, (
f"Registry base_url_env_var {pconfig.base_url_env_var} "
f"(provider={pconfig.id}) missing from blocklist"
)
def test_extra_auth_vars_covered(self):
"""Non-registry auth vars (ANTHROPIC_TOKEN, CLAUDE_CODE_OAUTH_TOKEN)
must also be in the blocklist."""
extras = {"ANTHROPIC_TOKEN", "CLAUDE_CODE_OAUTH_TOKEN"}
assert extras.issubset(_HERMES_PROVIDER_ENV_BLOCKLIST)
def test_non_registry_provider_vars_are_in_blocklist(self):
extras = {
"GOOGLE_API_KEY",
"DEEPSEEK_API_KEY",
"MISTRAL_API_KEY",
"GROQ_API_KEY",
"TOGETHER_API_KEY",
"PERPLEXITY_API_KEY",
"COHERE_API_KEY",
"FIREWORKS_API_KEY",
"XAI_API_KEY",
"HELICONE_API_KEY",
}
assert extras.issubset(_HERMES_PROVIDER_ENV_BLOCKLIST)
def test_optional_tool_and_messaging_vars_are_in_blocklist(self):
"""Tool/messaging vars from OPTIONAL_ENV_VARS should stay covered."""
from hermes_cli.config import OPTIONAL_ENV_VARS
for name, metadata in OPTIONAL_ENV_VARS.items():
category = metadata.get("category")
if category in {"tool", "messaging"}:
assert name in _HERMES_PROVIDER_ENV_BLOCKLIST, (
f"Optional env var {name} (category={category}) missing from blocklist"
)
elif category == "setting" and metadata.get("password"):
assert name in _HERMES_PROVIDER_ENV_BLOCKLIST, (
f"Secret setting env var {name} missing from blocklist"
)
def test_gateway_runtime_vars_are_in_blocklist(self):
extras = {
"TELEGRAM_HOME_CHANNEL",
"TELEGRAM_HOME_CHANNEL_NAME",
"DISCORD_HOME_CHANNEL",
"DISCORD_HOME_CHANNEL_NAME",
"DISCORD_REQUIRE_MENTION",
"DISCORD_FREE_RESPONSE_CHANNELS",
"DISCORD_AUTO_THREAD",
"SLACK_HOME_CHANNEL",
"SLACK_HOME_CHANNEL_NAME",
"SLACK_ALLOWED_USERS",
"WHATSAPP_ENABLED",
"WHATSAPP_MODE",
"WHATSAPP_ALLOWED_USERS",
"SIGNAL_HTTP_URL",
"SIGNAL_ACCOUNT",
"SIGNAL_ALLOWED_USERS",
"SIGNAL_GROUP_ALLOWED_USERS",
"SIGNAL_HOME_CHANNEL",
"SIGNAL_HOME_CHANNEL_NAME",
"SIGNAL_IGNORE_STORIES",
"HASS_TOKEN",
"HASS_URL",
"EMAIL_ADDRESS",
"EMAIL_PASSWORD",
"EMAIL_IMAP_HOST",
"EMAIL_SMTP_HOST",
"EMAIL_HOME_ADDRESS",
"EMAIL_HOME_ADDRESS_NAME",
"GATEWAY_ALLOWED_USERS",
"GH_TOKEN",
"GITHUB_APP_ID",
"GITHUB_APP_PRIVATE_KEY_PATH",
"GITHUB_APP_INSTALLATION_ID",
"MODAL_TOKEN_ID",
"MODAL_TOKEN_SECRET",
"DAYTONA_API_KEY",
}
assert extras.issubset(_HERMES_PROVIDER_ENV_BLOCKLIST)