Valhuntir forensic investigation platform with varying levels of human-in-the-loop review and control based on your deployment needs. CLI and architecture reference.
Platform Documentation · CLI Reference
Public Beta — This project is undergoing active feature development. Backward compatibility with future releases is not guaranteed. Consider this a public beta for feature testing and evaluation rather than a production-ready tool for real case data.
In its simplest form, Valhuntir Lite provides Claude Code with forensic knowledge and instructions on how to enforce forensic rigor, present findings for human review, and audit actions taken. MCP servers enhance accuracy by providing authoritative information — a forensic knowledge RAG and a Windows triage database — plus optional OpenCTI threat intelligence and REMnux malware analysis.
Quick — Forensic discipline, MCP packages, and config. No databases (<70 MB):
git clone https://github.com/AppliedIR/sift-mcp.git
cd sift-mcp
./quickstart-lite.sh --quick
Recommended — Adds the RAG knowledge base (22,000+ records from 23 security sources) and Windows triage databases (2.6M baseline records). Requires ~14 GB disk space:
- ~7 GB — ML dependencies (PyTorch, CUDA) required by the RAG embedding model
- ~6 GB — Windows triage baseline databases (2.6M rows, decompressed)
- ~1 GB — RAG index, source code, and everything else
git clone https://github.com/AppliedIR/sift-mcp.git
cd sift-mcp
./quickstart-lite.sh
This one-time setup takes approximately 15-30 minutes depending on internet speed and CPU. Subsequent runs reuse existing databases and index.
claude
/welcome
To update an existing Valhuntir Lite installation, re-run the installer from an updated clone (first step changes into wherever you installed the sift-mcp repo on initial installation):
cd sift-mcp
git pull
./quickstart-lite.sh
The installer is idempotent — it reuses the existing venv, skips databases and RAG index if already present, and redeploys config files.
graph LR
subgraph analyst ["Analyst Machine"]
CC["Claude Code<br/>(human interface)"]
FR["forensic-rag-mcp<br/>Knowledge search"]
WTR["windows-triage-mcp<br/>Baseline validation"]
CC -->|stdio| FR
CC -->|stdio| WTR
end
graph LR
subgraph analyst ["Analyst Machine"]
CC["Claude Code<br/>(human interface)"]
FR["forensic-rag-mcp<br/>Knowledge search"]
WTR["windows-triage-mcp<br/>Baseline validation"]
OC["opencti-mcp<br/>Threat intelligence"]
CC -->|stdio| FR
CC -->|stdio| WTR
CC -->|stdio| OC
end
subgraph octi ["OpenCTI Instance"]
OCTI[OpenCTI]
end
subgraph remnux ["REMnux Workstation"]
RAPI["remnux-mcp API<br/>:3000"]
RMX["remnux-mcp<br/>Malware analysis"]
RAPI --> RMX
end
subgraph internet ["Internet"]
ML["MS Learn MCP<br/>(HTTPS)"]
ZE["Zeltser IR Writing MCP<br/>(HTTPS)"]
end
CC -->|"streamable-http"| RAPI
CC -->|"HTTPS"| ML
CC -->|"HTTPS"| ZE
OC -->|"HTTP(S)"| OCTI
| Connection | Protocol | Notes |
|---|---|---|
| Claude Code → forensic-rag-mcp | stdio | Local Python process, always present |
| Claude Code → windows-triage-mcp | stdio | Local Python process, always present |
| Claude Code → opencti-mcp | stdio | Local Python process, connects out to OpenCTI via HTTP(S) |
| opencti-mcp → OpenCTI Instance | HTTP(S) | opencti-mcp runs locally, calls out to the OpenCTI server |
| Claude Code → remnux-mcp | streamable-http | Remote, on its own REMnux workstation |
| Claude Code → MS Learn MCP | HTTPS | https://learn.microsoft.com/api/mcp — streamable-http type in .mcp.json |
| Claude Code → Zeltser IR Writing MCP | HTTPS | https://website-mcp.zeltser.com/mcp — streamable-http type in .mcp.json |
No gateway, no sandbox, no deny rules. Claude runs forensic tools directly via Bash. Forensic discipline is suggested and reinforced via prompt hooks and reference documents, but Claude Code can choose to ignore them. See the sift-mcp README for details and optional add-ons.
For use cases where more definitive human-in-the-loop approval is desired, the full Valhuntir suite ensures accountability and enforces human review of findings through cryptographic signing, password-gated approvals, and multiple layered controls.
Full Valhuntir is LLM client agnostic — connect any MCP-compatible client through the gateway. Supported clients include Claude Code, Claude Desktop, LibreChat, Cherry Studio, and any MCP-only client that supports Streamable HTTP transport with Bearer token authentication. Forensic discipline is provided structurally at the gateway and MCP layer, not through client-specific prompt engineering, so the same rigor applies regardless of which AI model or client drives the investigation.
The examiner interacts with Valhuntir through three interfaces: the LLM client (AI-assisted investigation), the Examiner Portal (browser-based review and approval), and the vhir CLI (case management, evidence handling, and verification).
The typical deployment runs three VMs on a single host: SIFT (primary workstation), REMnux (malware analysis), and Windows (forensic tool execution). The examiner works on the SIFT VM — running the LLM client, the Examiner Portal in a browser, and the vhir CLI. REMnux and Windows are headless worker VMs. All three communicate over a VM-local network. Internet access is through NAT for external MCP services.
graph TB
subgraph host ["Host Machine"]
subgraph sift ["SIFT VM"]
CC["LLM Client<br/>(human interface)"]
BR["Browser<br/>(Examiner Portal)"]
CLI["vhir CLI"]
GW["sift-gateway :4508"]
CASE["Case Directory"]
CC -->|"streamable-http"| GW
BR -->|"HTTP"| GW
CLI --> CASE
end
subgraph remnux ["REMnux VM (optional)"]
RAPI["remnux-mcp :3000"]
end
subgraph winbox ["Windows VM (optional)"]
WAPI["wintools-mcp :4624"]
end
CC -->|"streamable-http"| RAPI
GW -->|"HTTPS"| WAPI
WAPI -->|"SMB"| CASE
end
subgraph internet ["Internet"]
ML["MS Learn MCP"]
ZE["Zeltser IR Writing MCP"]
OCTI["OpenCTI (if external)"]
end
CC -->|"HTTPS"| ML
CC -->|"HTTPS"| ZE
GW -.->|"HTTP(S)"| OCTI
REMnux and Windows VMs are optional. SIFT alone provides 79 MCP tools across 7 backends, the Examiner Portal, and full case management.
The sift-gateway aggregates 7 MCP backends as stdio subprocesses behind a single HTTP endpoint. Each backend is also available individually. The Examiner Portal is served by the gateway for browser-based review and approval.
graph LR
GW["sift-gateway :4508"]
FM["forensic-mcp<br/>26 tools · findings, timeline,<br/>evidence, discipline"]
CM["case-mcp<br/>15 tools · case management,<br/>audit queries, backup"]
RM["report-mcp<br/>6 tools · report generation,<br/>IOC aggregation"]
SM["sift-mcp<br/>6 tools · Linux forensic<br/>tool execution"]
RAG["forensic-rag<br/>3 tools · semantic search<br/>23K records"]
WT["windows-triage<br/>13 tools · offline baseline<br/>validation"]
OC["opencti<br/>10 tools · threat<br/>intelligence"]
CD["Examiner Portal<br/>browser review + commit"]
FK["forensic-knowledge<br/>shared YAML data"]
CASE["Case Directory"]
GW -->|stdio| FM
GW -->|stdio| CM
GW -->|stdio| RM
GW -->|stdio| SM
GW -->|stdio| RAG
GW -->|stdio| WT
GW -->|stdio| OC
GW --> CD
FM --> FK
SM --> FK
FM --> CASE
CM --> CASE
RM --> CASE
CD --> CASE
All findings and timeline events are staged as DRAFT by the AI. Only a human examiner can approve or reject them — via the Examiner Portal (browser) or the vhir CLI. Both paths produce identical HMAC-signed approval records. The AI cannot approve its own findings.
sequenceDiagram
participant AI as LLM + MCP Tools
participant Case as Case Directory
participant Human as Examiner<br/>(Portal or CLI)
AI->>Case: record_finding() → DRAFT
AI->>Case: record_timeline_event() → DRAFT
Note over Case: Staged for review
Human->>Case: Review, edit, approve/reject
Human->>Case: Commit (password + HMAC signing)
Note over Case: Only APPROVED items<br/>appear in reports
Human->>Case: vhir report --full
The Examiner Portal is the primary review interface with 8 tabs: Overview (investigation progress and getting started guide), Findings (the core review workflow with provenance chain display), Timeline (chronological events with a color-coded ruler), Hosts (systems involved, aggregated from findings), Accounts (user/service accounts involved), Evidence (registered files with SHA-256 integrity verification), IOCs (indicators extracted from findings with category/status filters), and TODOs (outstanding tasks).
Examiners review findings and timeline events, edit fields (confidence, justification, observation, interpretation, MITRE IDs, IOCs, tags), approve or reject items, and commit decisions — all in the browser. Each finding displays its evidence artifacts with a provenance chain showing which registered evidence files were input, which tools processed them, and what output was extracted. Keyboard shortcuts (1-8 tabs, j/k navigate, a approve, r reject, e edit, Shift+C commit) enable fast review. The sidebar is resizable, and search matches across title, observation, host, and account fields. Light and dark themes are supported.
The Commit button (Shift+C) uses challenge-response authentication: the browser derives a PBKDF2 key from the examiner's password and proves knowledge via HMAC — the password never leaves the browser. Timeline events auto-created from findings follow the finding's approval status unless manually edited. IOCs auto-extracted from findings cascade when all source findings reach the same status. The CLI's vhir approve provides the same functionality from the terminal. Open the portal with vhir portal.
| Component | Runs on | Port | Purpose |
|---|---|---|---|
| sift-gateway | SIFT | 4508 | Aggregates SIFT-local MCPs behind one HTTP endpoint |
| forensic-mcp | SIFT | (via gateway) | Findings, timeline, evidence, TODOs, IOCs, discipline (26 tools) |
| case-mcp | SIFT | (via gateway) | Case management, audit queries, evidence registration, backup (15 tools) |
| report-mcp | SIFT | (via gateway) | Report generation with profiles, IOC aggregation, MITRE mapping (6 tools) |
| sift-mcp | SIFT | (via gateway) | Denylist-protected forensic tool execution on Linux/SIFT (6 tools) |
| forensic-rag-mcp | SIFT | (via gateway) | Semantic search across Sigma, MITRE ATT&CK, Atomic Red Team, and more |
| windows-triage-mcp | SIFT | (via gateway) | Offline Windows baseline validation |
| opencti-mcp | SIFT | (via gateway) | Threat intelligence from OpenCTI (10 tools) |
| Examiner Portal | SIFT | (via gateway) | 8-tab browser UI: overview, findings with provenance chains, timeline with ruler, hosts, accounts, evidence verification, IOCs, TODOs. Primary review UI. |
| wintools-mcp | Windows | 4624 | Catalog-gated forensic tool execution on Windows (7 tools) |
| vhir CLI | SIFT | -- | Human-only: case init, evidence management, verification, exec. Approval also available via Examiner Portal. Remote examiners need SSH only for CLI-exclusive operations. |
| forensic-knowledge | anywhere | -- | Shared YAML data package (tools, artifacts, discipline) |
The gateway exposes each backend as a separate MCP endpoint. Clients can connect to the aggregate endpoint or to individual backends:
http://localhost:4508/mcp # Aggregate (all tools)
http://localhost:4508/mcp/forensic-mcp
http://localhost:4508/mcp/case-mcp
http://localhost:4508/mcp/report-mcp
http://localhost:4508/mcp/sift-mcp
http://localhost:4508/mcp/windows-triage-mcp
http://localhost:4508/mcp/forensic-rag-mcp
http://localhost:4508/mcp/opencti-mcp
graph LR
subgraph e1 ["Examiner 1 — SIFT Workstation"]
CC1["LLM Client<br/>(human interface)"]
BR1["Browser<br/>(human interface)"]
CLI1["vhir CLI"]
GW1["sift-gateway<br/>:4508"]
MCPs1["forensic-mcp · case-mcp · report-mcp<br/>sift-mcp · forensic-rag-mcp<br/>windows-triage-mcp · opencti-mcp"]
CASE1["Case Directory"]
CC1 -->|"streamable-http"| GW1
BR1 -->|"HTTPS"| GW1
GW1 -->|stdio| MCPs1
MCPs1 --> CASE1
CLI1 --> CASE1
end
subgraph e2 ["Examiner 2 — SIFT Workstation"]
CC2["LLM Client<br/>(human interface)"]
BR2["Browser<br/>(human interface)"]
CLI2["vhir CLI"]
GW2["sift-gateway<br/>:4508"]
MCPs2["forensic-mcp · case-mcp · report-mcp<br/>sift-mcp · forensic-rag-mcp<br/>windows-triage-mcp · opencti-mcp"]
CASE2["Case Directory"]
CC2 -->|"streamable-http"| GW2
BR2 -->|"HTTPS"| GW2
GW2 -->|stdio| MCPs2
MCPs2 --> CASE2
CLI2 --> CASE2
end
CASE1 <-->|"export / merge"| CASE2
cases/INC-2026-0219/
├── CASE.yaml # Case metadata (name, status, examiner)
├── evidence/ # Original evidence (read-only after registration)
├── extractions/ # Extracted artifacts
├── reports/ # Generated reports
├── findings.json # F-alice-001, F-alice-002, ...
├── timeline.json # T-alice-001, ...
├── todos.json # TODO-alice-001, ...
├── iocs.json # IOC-alice-001, ... (auto-extracted from findings)
├── evidence.json # Evidence registry
├── actions.jsonl # Investigative actions (append-only)
├── evidence_access.jsonl # Chain-of-custody log
├── approvals.jsonl # Approval audit trail
├── pending-reviews.json # Portal edits awaiting approval
└── audit/
├── forensic-mcp.jsonl
├── sift-mcp.jsonl
├── claude-code.jsonl # PostToolUse hook captures (Claude Code only)
└── ...
- Zeltser IR Writing MCP — Required for report generation. Configured automatically by
vhir setup client. URL: https://website-mcp.zeltser.com/mcp (HTTPS, no auth)
Requires Python 3.10+ and sudo access. The installer handles everything: MCP servers, gateway, vhir CLI, HMAC verification ledger, examiner identity, and LLM client configuration. When you select Claude Code, additional forensic controls are deployed (kernel-level sandbox, case data deny rules, PreToolUse guard hook, PostToolUse audit hook, provenance enforcement, password-gated human approval with HMAC signing). Other clients get MCP config only.
Quick — Core platform only, no databases (~70 MB):
curl -fsSL https://raw.githubusercontent.com/AppliedIR/sift-mcp/main/quickstart.sh -o /tmp/vhir-quickstart.sh && bash /tmp/vhir-quickstart.sh
Recommended — Adds the RAG knowledge base (22,000+ records from 23 security sources) and Windows triage databases (2.6M baseline records), downloaded as pre-built snapshots. Requires ~14 GB disk space:
- ~7 GB — ML dependencies (PyTorch, CUDA) required by the RAG embedding model
- ~6 GB — Windows triage baseline databases (2.6M rows, decompressed)
- ~1 GB — RAG index, source code, and everything else
curl -fsSL https://raw.githubusercontent.com/AppliedIR/sift-mcp/main/quickstart.sh -o /tmp/vhir-quickstart.sh && bash /tmp/vhir-quickstart.sh --recommended
Custom — Individual package selection, OpenCTI integration, or remote access with TLS:
git clone https://github.com/AppliedIR/sift-mcp.git
cd sift-mcp
./setup-sift.sh
# Option 1: git clone
git clone https://github.com/AppliedIR/wintools-mcp.git; cd wintools-mcp
# Option 2: download ZIP (no git required)
Invoke-WebRequest https://github.com/AppliedIR/wintools-mcp/archive/refs/heads/main.zip -OutFile wintools.zip
Expand-Archive wintools.zip -DestinationPath .; cd wintools-mcp-main
Then run the installer:
.\scripts\setup-windows.ps1
All Valhuntir components are assumed to run on a private forensic network, protected by firewalls, and not exposed to incoming connections from the Internet or potentially hostile systems. The design assumes dedicated, isolated systems are used throughout.
Any data loaded into the system or its component VMs, computers, or instances runs the risk of being exposed to the underlying AI. Only place data on these systems that you are willing to send to your AI provider.
Outgoing Internet connections are required for report generation (Zeltser IR Writing MCP) and optionally used for threat intelligence (OpenCTI) and documentation (MS Learn MCP). No incoming connections from external systems should be allowed.
Valhuntir is designed so that AI interactions flow through MCP tools, enabling security controls and audit trails. Clients with direct shell access (like Claude Code) can also operate outside MCP, but vhir setup client deploys forensic controls for Claude Code: a kernel-level sandbox restricts Bash writes, deny rules block Edit/Write to case data files, a PreToolUse hook guards against Bash redirections to protected files, a PostToolUse hook captures every Bash command to the audit trail, provenance enforcement ensures findings are traceable to evidence, and an HMAC verification ledger provides cryptographic proof that approved findings haven't been tampered with. Valhuntir is not designed to defend against a malicious AI or to constrain the AI client that you deploy.
Most vhir CLI operations have MCP equivalents via case-mcp, forensic-mcp, and report-mcp. When working with an MCP-connected client, you can ask the AI to handle case management, evidence registration, report generation, and more — the AI operates through audited MCP tools rather than direct CLI invocation.
The commands below that require human interaction at the terminal (/dev/tty) cannot be delegated to the AI. These are intentional human-in-the-loop checkpoints — they use password entry, interactive review, or terminal confirmation to ensure the human examiner retains control over approval, rejection, and security-sensitive operations.
These commands read from /dev/tty directly and cannot be run by an AI client, even via Bash. This is by design — they are the human-in-the-loop controls that ensure the examiner reviews and approves all findings.
vhir approve # Interactive review of all DRAFT items
vhir approve F-alice-001 F-alice-002 T-alice-001 # Approve specific IDs
vhir approve F-alice-001 --edit # Edit in $EDITOR before approving
vhir approve F-alice-001 --note "Malware family unconfirmed" # Approve with examiner note
vhir approve --by jane # Filter to IDs with jane's examiner prefix
vhir approve --findings-only # Skip timeline events
vhir approve --timeline-only # Skip findings
vhir approve --review # Apply pending portal edits
Requires password entry via /dev/tty. Approved findings are HMAC-signed with a PBKDF2-derived key. The --review flag applies edits made in the Examiner Portal (stored in pending-reviews.json), recomputes content hashes and HMAC signatures, then removes the pending file. Alternatively, use the portal's Commit button (Shift+C) which performs the same operation via challenge-response authentication — the password never leaves the browser.
vhir reject F-alice-003 --reason "Insufficient evidence for attribution"
vhir reject F-alice-003 T-alice-002 --reason "Contradicted by memory analysis"
Requires password confirmation via /dev/tty.
vhir exec --purpose "Extract MFT from image" -- fls -r -m / image.E01
Requires /dev/tty confirmation. Logged to audit/cli-exec.jsonl. Use this for manual tool execution with audit trail when not operating through MCP.
vhir backup /path/to/destination # Back up case data (interactive)
vhir backup /path/to/destination --all # Include evidence + extractions
vhir backup /path/to/destination --include-evidence # Include evidence files
vhir backup /path/to/destination --include-extractions # Include extraction files
vhir backup --verify /path/to/backup/ # Verify backup integrity
Creates a timestamped backup with SHA-256 manifest. Verification checks every file hash against the manifest. The --all flag includes evidence and extraction files (which can be large). Without flags, interactive mode prompts per category with size estimates.
vhir unlock-evidence # Directory chmod 755, files remain 444
vhir evidence unlock
Requires /dev/tty confirmation. Unlocking evidence allows writes to the evidence directory.
vhir config --setup-password # Set approval password (PBKDF2-hashed, min 8 chars)
vhir config --reset-password # Reset password (requires current, re-signs ledger)
Password entry uses masked input via /dev/tty with termios. No echo, no stdin — the AI cannot read or supply the password.
vhir review --findings --verify # Cross-check content hashes + HMAC verification
vhir review --findings --verify --mine # HMAC verification for current examiner only
Verification requires the examiner's password to derive the HMAC key and confirm integrity.
The remaining commands can also be performed through MCP tools (case-mcp, forensic-mcp, report-mcp) when working with an MCP-connected client. The CLI equivalents are listed here for reference and for use outside MCP sessions.
vhir portal # Open the Examiner Portal in your browser
Opens the Examiner Portal for the active case. The portal is the primary review interface — examiners can review, edit, approve, reject, and commit findings entirely in the browser. Use the Commit button (Shift+C) to apply decisions with challenge-response authentication. Alternatively, vhir approve --review applies pending edits from the CLI.
vhir case init "Ransomware Investigation" # Create a new case
vhir case close INC-2026-02191200 # Close a case by ID
vhir case activate INC-2026-02191200 # Set active case
vhir case migrate # Migrate to flat layout (see below)
vhir review # Case summary (counts by status)
vhir review --findings # Findings table
vhir review --findings --detail # Full finding detail
vhir review --iocs # IOCs grouped by approval status
vhir review --timeline # Timeline events
vhir review --timeline --status APPROVED # Filter timeline by status
vhir review --timeline --start 2026-01-01 --end 2026-01-31 # Filter by date range
vhir review --timeline --type execution # Filter by event type
vhir review --evidence # Evidence registry and access log
vhir review --audit --limit 100 # Audit trail (last N entries)
vhir review --todos --open # Open TODOs
vhir todo # List open TODOs
vhir todo --all # Include completed
vhir todo add "Run volatility on server-04" --assignee jane --priority high --finding F-alice-003
vhir todo complete TODO-alice-001
vhir todo update TODO-alice-002 --note "Waiting on third party" --priority low
vhir register-evidence /path/to/image.E01 --description "Disk image from workstation"
vhir lock-evidence # All files chmod 444, directory chmod 555
vhir evidence register /path/to/image.E01 --description "Disk image"
vhir evidence list
vhir evidence verify
vhir evidence log [--path <filter>]
vhir evidence lock
vhir export --file steve-findings.json # Export findings for sharing
vhir merge --file jane-findings.json # Merge another examiner's findings
vhir report --full [--save <path>]
vhir report --executive-summary [--save <path>]
vhir report --timeline [--from <date> --to <date>] [--save <path>]
vhir report --ioc [--save <path>]
vhir report --findings F-alice-001,F-alice-002 [--save <path>]
vhir report --status-brief [--save <path>]
vhir audit log [--limit 100] [--mcp sift-mcp] [--tool run_command]
vhir audit summary
vhir service status # Show running backends + health
vhir service start forensic-rag # Start a backend
vhir service stop windows-triage # Stop a backend
vhir service restart sift-mcp # Restart a backend
vhir case migrate # Migrate primary examiner data to flat layout
vhir case migrate --examiner alice # Specify examiner
vhir case migrate --import-all # Merge all examiners' data
vhir config --examiner "jane-doe" # Set examiner identity
vhir config --show # Show current configuration
vhir update # Pull latest, reinstall, redeploy, restart
vhir update --check # Check for updates without applying
vhir update --no-restart # Skip gateway restart after update
vhir join --sift SIFT_URL --code CODE # Join from remote machine using join code
vhir join --sift SIFT_URL --code CODE --wintools # Join as wintools machine (registers backend)
Exchange a one-time join code for gateway credentials. Run on the remote machine (analyst laptop or Windows forensic workstation). The join code is generated on SIFT via vhir setup join-code. Credentials are saved to ~/.vhir/config.yaml with restricted permissions (0600).
vhir setup client # Interactive client configuration (recommended)
vhir setup test # Test MCP server connectivity
Generate Streamable HTTP config for your LLM client:
vhir setup client # Interactive wizard
vhir setup client --client=claude-code --sift=http://127.0.0.1:4508 -y # Local solo
vhir setup client --sift=SIFT_IP:4508 --windows=WIN_IP:4624 # SIFT + Windows
For remote orchestrator setups (Path 2), remote examiners run a platform-specific setup script that creates a ~/vhir/ workspace with MCP config, forensic controls, and discipline docs:
# Linux
curl -sSL https://raw.githubusercontent.com/AppliedIR/valhuntir/main/setup-client-linux.sh \
| bash -s -- --sift=https://SIFT_IP:4508 --code=XXXX-XXXX
# macOS
curl -sSL https://raw.githubusercontent.com/AppliedIR/valhuntir/main/setup-client-macos.sh \
| bash -s -- --sift=https://SIFT_IP:4508 --code=XXXX-XXXX
# Windows
Invoke-WebRequest -Uri https://raw.githubusercontent.com/AppliedIR/valhuntir/main/setup-client-windows.ps1 -OutFile setup-client-windows.ps1
.\setup-client-windows.ps1 -Sift https://SIFT_IP:4508 -Code XXXX-XXXX
Always launch your LLM client from ~/vhir/ or a subdirectory. Forensic controls only apply when started from within the workspace. To uninstall, re-run the setup script with --uninstall (Linux/macOS) or -Uninstall (Windows).
Claude Desktop's config file supports stdio transport only. The mcp-remote bridge is used to connect to the gateway. vhir setup client --client=claude-desktop generates the correct mcp-remote config automatically.
| Client | Platforms | Config file | Extras |
|---|---|---|---|
| Claude Code | Linux, macOS, Windows | ~/vhir/.mcp.json or ~/.claude.json (SIFT) |
CLAUDE.md, settings.json, sandbox, audit hooks |
| Claude Desktop | macOS, Windows | claude_desktop_config.json (see note) |
Requires mcp-remote bridge. Project instructions from AGENTS.md |
| Cherry Studio | Linux, macOS, Windows | JSON import (manual) | baseUrl field, streamableHttp type (camelCase) |
| LibreChat | Any (browser) | librechat.yaml (mcpServers section) |
Valhuntir generates librechat_mcp.yaml reference to merge |
| Other | Any | vhir-mcp-config.json |
Manual integration |
Claude Code on Windows requires Git for Windows (provides Git Bash) or WSL. Claude Desktop is not available on Linux. Claude Desktop's config file supports stdio transport only — the vhir setup client wizard generates mcp-remote bridge configs automatically. Config path is platform-specific: ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows. Any MCP client that supports Streamable HTTP transport with Bearer token authentication headers will work — the gateway is not client-specific.
Every approval, rejection, and command execution is logged with examiner identity. Resolution order:
| Priority | Source | Example |
|---|---|---|
| 1 | --examiner flag |
vhir approve --examiner jane-doe F-jane-001 |
| 2 | VHIR_EXAMINER env var |
export VHIR_EXAMINER=jane-doe |
| 3 | ~/.vhir/config.yaml |
examiner: jane-doe |
| 4 | VHIR_ANALYST env var |
Deprecated fallback |
| 5 | OS username | Warns if unconfigured |
| Repo | Purpose |
|---|---|
| sift-mcp | Monorepo: 11 SIFT packages (forensic-mcp, case-mcp, report-mcp, sift-mcp, sift-gateway, case-dashboard, forensic-knowledge, forensic-rag, windows-triage, opencti, sift-common) |
| wintools-mcp | Windows forensic tool execution (7 tools, 31 catalog entries) |
| Valhuntir | CLI, architecture reference |
vhir update # Pull latest code, reinstall packages, redeploy controls, restart gateway
vhir update --check # Check for updates without applying
vhir update --no-restart # Update without restarting the gateway
The update command pulls the latest code from both repos (sift-mcp and vhir), reinstalls all packages, redeploys forensic controls, restarts the gateway, and runs a connectivity smoke test.
Both modes share the same Python venv, triage databases, and RAG index. Full Valhuntir adds the gateway (7 MCP backends behind one HTTP endpoint), 4 additional MCP servers (forensic-mcp, case-mcp, report-mcp, sift-mcp), a web-based review portal (Examiner Portal), structured case management, sandbox enforcement, and HMAC-signed approvals.
To upgrade, run setup-sift.sh from your existing sift-mcp clone. The
installer reuses the existing venv and databases. Lite case data (markdown
files) does not auto-migrate to full case data (structured JSON) — start
fresh or transfer findings manually.
Never place original evidence on any Valhuntir system. Only use working copies for which verified originals or backups exist. Valhuntir workstations process evidence through AI-connected tools, and any data loaded into these systems may be transmitted to the configured AI provider. Treat all Valhuntir systems as analysis environments, not evidence storage.
Evidence integrity is verified by SHA-256 hashes recorded at registration. Examiners can optionally lock evidence to read-only via vhir evidence lock. Proper evidence integrity depends on verified hashes, write blockers, and chain-of-custody procedures that exist outside this platform.
Case directories can reside on external or removable media. ext4 is preferred for full permission support. NTFS and exFAT are acceptable but file permission controls (read-only protection) will be silently ineffective. FAT32 is discouraged due to the 4 GB file size limit.
While steps have been taken to enforce human-in-the-loop controls, it is ultimately the responsibility of each examiner to ensure that their findings are accurate and complete. The AI, like a hex editor, is a tool to be used by properly trained incident response professionals. Users are responsible for ensuring their use complies with applicable laws, regulations, and organizational policies. Use only on systems and data you are authorized to analyze.
This software is provided "as is" without warranty of any kind. See LICENSE for full terms.
MITRE ATT&CK is a registered trademark of The MITRE Corporation. SIFT Workstation is a product of the SANS Institute.
Architecture and direction by Steve Anson. Implementation by Claude Code (Anthropic).
I do DFIR. I am not a developer. This project would not exist without Claude Code handling the implementation. While an immense amount of effort has gone into design, testing, and review, I fully acknowledge that I may have been working hard and not smart in places. My intent is to jumpstart discussion around ways this technology can be leveraged for efficiency in incident response while ensuring that the ultimate responsibility for accuracy remains with the human examiner.
MIT License - see LICENSE