mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-31 19:16:29 +00:00
0dee92df22
Hardens the context window against Brainworm-class promptware attacks (see #496). Three changes: 1. tools/threat_patterns.py — single source of truth for injection/promptware patterns. Replaces the duplicated pattern lists in prompt_builder.py and memory_tool.py. Adds ~15 new Brainworm/C2 patterns (node registration, heartbeat/beacon, pull tasking, anti-forensic disk avoidance, identity override, known framework names). Three scopes — 'all' (narrow, classic injection), 'context' (adds promptware/role-play, broader detection), 'strict' (adds persistence/SSH-backdoor patterns for user-mediated writes). 2. MemoryStore.load_from_disk() now scans entries at snapshot-build time. Poisoned entries are replaced with [BLOCKED: ...] placeholders in the frozen system-prompt snapshot. Live state keeps the original so the user can still inspect + remove via memory(action=read/remove). Scan is deterministic from disk bytes — prefix-cache invariant holds. 3. make_tool_result_message() wraps results from high-risk tools (web_extract, web_search, browser_*, mcp_*) in <untrusted_tool_result source="...">...</untrusted_tool_result> delimiters with framing prose telling the model the content is data, not instructions. Architectural defense against indirect injection from poisoned web pages, GitHub issues, MCP responses — does NOT regex-scan tool results (pattern arms race + per-iteration latency). Multimodal content lists pass through unwrapped to preserve adapter compatibility. Pattern philosophy: anchor on C2-specific vocabulary or unambiguous attack behavior, NOT on bossy English. Dropped patterns suggested in #496 that would have tripped legitimate content: standalone 'you are obligated to', 'do not respond immediately', 'you must X' without a C2-verb anchor. Validation: - 257/257 targeted tests pass (test_threat_patterns + test_memory_tool + test_tool_dispatch_helpers + test_prompt_builder) - E2E run with real Brainworm payload: blocked from AGENTS.md context-file path, blocked from MEMORY.md snapshot, wrapped in delimiters when arriving via web_extract. Legitimate 'you must follow conventions' phrasing not flagged. Explicitly NOT in this PR (per #496 discussion): - Per-tool-result regex scanning (pattern arms race) - SessionBehaviorMonitor / polling-loop detection (wrong layer) - Outbound network gating (Docker backend already covers this) - security.context_scanning warn|block knob (current behavior is always block-with-placeholder — there's no warn mode that makes sense) Closes #496 for Phase 1 + the architectural delimiter piece of Phase 2. Phase 3 stays in tracking issue territory. |
||
|---|---|---|
| .. | ||
| computer_use | ||
| environments | ||
| neutts_samples | ||
| __init__.py | ||
| ansi_strip.py | ||
| approval.py | ||
| binary_extensions.py | ||
| browser_camofox.py | ||
| browser_camofox_state.py | ||
| browser_cdp_tool.py | ||
| browser_dialog_tool.py | ||
| browser_supervisor.py | ||
| browser_tool.py | ||
| budget_config.py | ||
| checkpoint_manager.py | ||
| clarify_gateway.py | ||
| clarify_tool.py | ||
| code_execution_tool.py | ||
| computer_use_tool.py | ||
| credential_files.py | ||
| cronjob_tools.py | ||
| debug_helpers.py | ||
| delegate_tool.py | ||
| discord_tool.py | ||
| env_passthrough.py | ||
| fal_common.py | ||
| feishu_doc_tool.py | ||
| feishu_drive_tool.py | ||
| file_operations.py | ||
| file_state.py | ||
| file_tools.py | ||
| fuzzy_match.py | ||
| homeassistant_tool.py | ||
| image_generation_tool.py | ||
| interrupt.py | ||
| kanban_tools.py | ||
| lazy_deps.py | ||
| managed_tool_gateway.py | ||
| mcp_oauth.py | ||
| mcp_oauth_manager.py | ||
| mcp_tool.py | ||
| memory_tool.py | ||
| microsoft_graph_auth.py | ||
| microsoft_graph_client.py | ||
| mixture_of_agents_tool.py | ||
| neutts_synth.py | ||
| openrouter_client.py | ||
| osv_check.py | ||
| patch_parser.py | ||
| path_security.py | ||
| process_registry.py | ||
| registry.py | ||
| schema_sanitizer.py | ||
| send_message_tool.py | ||
| session_search_tool.py | ||
| skill_manager_tool.py | ||
| skill_provenance.py | ||
| skill_usage.py | ||
| skills_ast_audit.py | ||
| skills_guard.py | ||
| skills_hub.py | ||
| skills_sync.py | ||
| skills_tool.py | ||
| slash_confirm.py | ||
| terminal_tool.py | ||
| threat_patterns.py | ||
| tirith_security.py | ||
| todo_tool.py | ||
| tool_backend_helpers.py | ||
| tool_output_limits.py | ||
| tool_result_storage.py | ||
| transcription_tools.py | ||
| tts_tool.py | ||
| url_safety.py | ||
| video_generation_tool.py | ||
| vision_tools.py | ||
| voice_mode.py | ||
| web_tools.py | ||
| website_policy.py | ||
| x_search_tool.py | ||
| xai_http.py | ||
| yuanbao_tools.py | ||