# Third-Party Software Security Review — SCORPIOX CODE

| Field | Value |
|-------|-------|
| **Software** | SCORPIOX CODE |
| **Type** | AI-Powered Development Tool / CLI Platform |
| **Language** | Pure C (zero external dependencies) |
| **Audit Date** | 2026-04-30 |
| **Codebase Commit** | `fe4a27dca4c7a52f1735792badc534530cd24a84` |
| **Classification** | CONFIDENTIAL — For Corporate Review Only |

---

## 1. Executive Summary

This report consolidates findings from 14 specialized security audit agents examining the SCORPIOX CODE codebase at commit `fe4a27d`. The audit covers supply chain integrity, build provenance, network security, TLS configuration, credential handling, file I/O safety, buffer safety, memory safety, command injection, privilege management, Windows deployment scope, telemetry practices, install script security, and the WPF launcher application.

### Key Metrics

| Metric | Value |
|--------|-------|
| **Lines of Code** | 376,275 total (148,135 project / 228,140 vendor) |
| **Total Findings** | 155 |
| **Critical** | 0 |
| **High** | 0 |
| **Medium** | 14 |
| **Low** | 44 |
| **Informational** | 97 |
| **Overall Risk Rating** | **LOW** |
| **Windows-Filtered Findings** | 138 (0 Critical, 0 High, 10 Medium, 38 Low, 90 Info) |

### Assessment Summary

SCORPIOX CODE demonstrates **strong security engineering discipline** across a 148K-line pure-C codebase. The project achieves a safe-to-unsafe string function ratio of 3,464:1, enforces centralized TLS verification by default, uses `fork()+execvp()` over `system()` for process spawning, vendors only two well-maintained C libraries (mbedTLS 3.6.6, yyjson 0.12.0), and ships zero third-party runtime dependencies in its Windows distribution. Both telemetry systems are disabled by default with no mandatory data collection.

No critical or high-severity findings were identified. The 14 medium-severity findings are concentrated in three areas: (1) file I/O edge cases on multi-user Linux systems, (2) command injection in the Linux-only container init process (requires filesystem control), and (3) install script integrity gaps (same-origin checksums without independent code signing). Most medium findings do not apply to the primary Windows workstation deployment.

---

## 2. Deployment Scope — Windows Workstation vs Server-Side

### 2.1 Primary Deployment: Windows Workstation

The following binaries are compiled for and deployed on Windows:

| Binary | Purpose |
|--------|---------|
| `sx.exe` | Main TUI — AI coding assistant |
| `scorpiox.exe` | Symlink to sx.exe |
| `scorpiox-bash.exe` | Shell command execution (PowerShell/cmd.exe) |
| `scorpiox-busybox.exe` | Embedded Unix utilities |
| `scorpiox-screenshot.exe` | Screenshot capture |
| `scorpiox-vi.exe` | Minimal vi text editor |
| `scorpiox-config.exe` | Configuration manager |
| `scorpiox-websearch.exe` | Multi-engine web search |
| `scorpiox-fetch.exe` | URL content fetcher |
| `scorpiox-renderimage.exe` | Image rendering/display |
| `scorpiox-pwsh.exe` | PowerShell integration |
| `scorpiox-wsl.exe` | WSL integration |
| `scorpiox-tmux.exe` | Terminal session manager |
| `scorpiox-multiplexer.exe` | Terminal multiplexer |
| `scorpiox-voice.exe` | Voice recording + Whisper |
| `scorpiox-server.exe` | HTTP server for script execution |
| `scorpiox-otp.exe` | TOTP/HOTP generator |
| `scorpiox-transcript.exe` | Conversation viewer |
| `scorpiox-conv.exe` | Conversation manager |
| `scorpiox-rewind.exe` | Conversation checkpoint rewind |
| `scorpiox-debug.exe` | Diagnostics viewer |
| `scorpiox-logger.exe` | Log file viewer |
| `scorpiox-printlogs.exe` | Log export utility |
| `scorpiox-systemprompt.exe` | System prompt generator |
| `scorpiox-search.exe` | Code search (grep/rg/ag) |
| `scorpiox-agent.exe` | Sub-agent spawner |
| `scorpiox-tasks.exe` | Task manager |
| `scorpiox-skills.exe` | Skill manager |
| `scorpiox-planmode.exe` | Plan mode manager |
| `scorpiox-askuserquestion.exe` | Interactive question TUI |
| `scorpiox-gemini.exe` | Gemini provider |
| `scorpiox-mcp.exe` | Model Context Protocol |
| `scorpiox-beam.exe` | Beam sharing |
| `scorpiox-host.exe` | Host information |
| `scorpiox-openai.exe` | OpenAI provider |
| `scorpiox-email.exe` | Email client |
| `scorpiox-usage.exe` | Usage reporting (opt-in) |
| `scorpiox-emit-session.exe` | Session telemetry (opt-in) |
| `scorpiox-frp.exe` | Fast reverse proxy |
| `scorpiox-hook.exe` | Git hook manager |
| `scorpiox-vault-git.exe` | Git vault |
| `scorpiox-mirror-git.exe` | Git mirror |
| `scorpiox-dns.exe` | DNS server |
| `scorpiox-kql.exe` | KQL query engine |
| `scorpiox-sdk.exe` | SDK utilities |
| `scorpiox-server-email.exe` | Email server |
| `scorpiox-server-http2tcp.exe` | HTTP-to-TCP bridge |
| `scorpiox-claudecode-fetchtoken.exe` | Claude token fetcher |
| `scorpiox-claudecode-refreshtoken.exe` | Claude token refresh |
| `scorpiox-codex-refreshtoken.exe` | Codex token refresh |
| `scorpiox-claudecode-models.exe` | Claude model list |
| `scorpiox-codex-fetchtoken.exe` | Codex token fetcher |
| `scorpiox-gemini-fetchtoken.exe` | Gemini token fetcher |
| `scorpiox-server-fetchtoken.exe` | Server token fetcher |
| `libsx.dll` | Shared library for WPF P/Invoke |

**Total: 55 Windows binaries** (53 executables + 1 DLL + 1 symlink)

### 2.2 Linux/Unix-Only Binaries (NOT deployed on Windows)

| Binary | Reason |
|--------|--------|
| `scorpiox-sshpass` | Unix PTY/signal handling |
| `scorpiox-traffic` | Unix raw sockets |
| `scorpiox-docs` | fork/pipe-based browser |
| `scorpiox-runtest` | fork()+execvp() test runner |
| `scorpiox-executecurl` | fork-based curl executor |
| `scorpiox-podman` | Podman container manager |
| `scorpiox-unshare` | Rootless container runtime (user namespaces) |
| `scorpiox-init` | Container init (PID 1) |
| `scorpiox-vm` | KVM virtual machine runner |
| `scorpiox-cron` | Crontab wrapper |
| `scorpiox-whatsapp` | WhatsApp CLI (fork/pipe/select) |

### 2.3 Windows-Filtered Findings Summary

Of the 155 total findings, **17 are Linux/Unix-only** and do not apply to Windows workstation deployments:

| Excluded Finding | Report | Severity | Reason |
|------------------|--------|----------|--------|
| NET-03: Container registry TLS verify | Network | LOW | Podman/container Linux-only |
| FIO-03: /tmp/sxmux predictable path | File I/O | MEDIUM | Unix socket path |
| MEM-02: scorpiox-unshare stack leak | Memory | MEDIUM | Linux-only binary |
| CJ-01: scorpiox-init descriptor injection | Command Inj. | MEDIUM | Linux-only binary |
| CJ-02: scorpiox-init INIT_TOOLS injection | Command Inj. | MEDIUM | Linux-only binary |
| PA-01: Missing CLONE_NEWIPC | Privilege | LOW | Linux namespaces |
| PA-02: Seccomp minimal blocklist | Privilege | LOW | Linux seccomp |
| PA-03: --privileged flag | Privilege | INFO | Linux container |
| PA-04: PR_SET_NO_NEW_PRIVS | Privilege | INFO | Linux-only |
| PA-05: Environment clearing in containers | Privilege | INFO | Linux-only |
| PA-07: Capability dropping | Privilege | INFO | Linux-only |
| PA-08: pivot_root | Privilege | INFO | Linux-only |
| PA-09: Overlayfs | Privilege | INFO | Linux-only |
| IS-04: --no-verify flag | Install | LOW | Linux/macOS script |
| IS-07: macOS ad-hoc codesign | Install | LOW | macOS only |
| IS-09: curl\|bash pattern | Install | LOW | Linux/macOS |
| Finding on bridge/ws2tcp.c traversal | File I/O | (partial) | Linux-only component |

**Windows deployment finding count: 138** (0 Critical, 0 High, 10 Medium, 38 Low, 90 Info)

---

## 3. Changes Since Last Audit

**Previous Audit:** `output/2026-04-30_6851bc0` (commit `6851bc0`, 2026-04-30)

| Metric | Previous (`6851bc0`) | Current (`fe4a27d`) | Delta |
|--------|---------------------|---------------------|-------|
| **Total Findings** | 225 | 155 | **−70 (↓31%)** |
| **Critical** | 2 | 0 | **−2** |
| **High** | 6 | 0 | **−6** |
| **Medium** | 38 | 14 | **−24** |
| **Low** | 55 | 44 | **−11** |
| **Info** | 124 | 97 | **−27** |
| **Windows Findings** | 67 | 138* | +71 |
| **Overall Risk** | LOW-MEDIUM | LOW | **Improved** |

*\*Note: The Windows finding count increase reflects improved granularity in the Windows deployment analysis (14 agents vs 13 previously, with the addition of the WPF launcher audit contributing 22 findings). The actual security posture improved — all Critical and High findings were remediated.*

### Key Improvements Since Last Audit

1. **All 2 Critical findings eliminated** — zero critical findings remain
2. **All 6 High-severity findings eliminated** — zero high findings remain
3. **Medium findings reduced by 63%** (38 → 14)
4. **Commit `fe4a27d` specifically addresses** Windows command injection findings CJ-16, CJ-17, CJ-18 (per commit message: "security: fix 3 Windows command injection findings")
5. **Overall risk rating improved** from LOW-MEDIUM to LOW

---

## 4. Supply Chain & Dependencies

**Agent:** supply-chain | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 1 Low, 5 Info

### Key Findings

The project has an exceptionally minimal supply chain for a software platform of this scale:

- **Vendored libraries:** Only 2 — mbedTLS 3.6.6 (LTS, Apache-2.0/GPL-2.0+) and yyjson 0.12.0 (MIT)
- **No package manager dependencies** in the core C build (no npm, pip, cargo, go modules)
- **No pre-built binaries** in the source repository
- **No FetchContent, ExternalProject, or git submodules**
- **Windows release:** Exclusively native binaries (.exe, .dll) — no interpreted runtimes

An optional WhatsApp bridge component (`bridge/`) uses Bun/TypeScript with 2 npm dependencies, but this is **not** part of the core build system or Windows distribution.

| Finding | Severity | Description |
|---------|----------|-------------|
| SC-01 | LOW | Vendored library versions not tracked in a dedicated manifest file (versions are in source headers) |
| SC-02–SC-06 | INFO | Positive observations: current library versions, proper licensing, minimal config, no pre-built binaries, zero-dependency Windows distribution |

---

## 5. Build System & Code Provenance

**Agent:** build-provenance | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 2 Low, 5 Info

### Build Security Highlights

- **Compiler hardening:** `-Wall -Wextra -fstack-protector-strong -D_FORTIFY_SOURCE=2 -fcf-protection` (GCC)
- **Linker hardening:** Full RELRO (`-Wl,-z,relro,-z,now`), non-executable stack (`-Wl,-z,noexecstack`)
- **Static linking** by default (`SX_STATIC_LINK=ON`)
- **1,225 commits** with traceable history
- **Model header generation frozen** — Python code generator not run in CI, header checked into source

| Finding | Severity | Description |
|---------|----------|-------------|
| BP-01 | LOW | `bridge/Makefile` missing linker hardening flags (RELRO, noexecstack) |
| BP-02 | LOW | PIE not explicitly enabled for static builds (typical for static binaries) |
| BP-03–BP-07 | INFO | Positive: hardening flags present, deterministic build, frozen code generation, comprehensive CMake structure |

---

## 6. Network Endpoints

**Agent:** network-endpoints | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 3 Low, 7 Info

### Network Architecture

78 unique hardcoded endpoints across 43 source files. All external connections use HTTPS. Endpoints fall into three categories:

1. **First-party infrastructure** (`*.scorpiox.net`) — distribution, token services, API proxying
2. **Third-party LLM APIs** — Anthropic, OpenAI, Google AI, Z.AI (all HTTPS)
3. **Localhost services** — HTTP for local IPC only

| Finding | Severity | Description |
|---------|----------|-------------|
| NET-01 | LOW | Hardcoded public IP (`20.53.240.54`) for TCP token transport in WASM config; native config uses hostname |
| NET-02 | LOW | DNS server binds `0.0.0.0:53` by default (intentional; configurable via `DNS_LISTEN`) |
| NET-03 | LOW | Container registry TLS verification defaults to `false` (Linux-only, Podman context) |
| NET-04–NET-10 | INFO | Positive observations: all HTTPS, proper localhost binding, standard API endpoints |

---

## 7. TLS/SSL Security

**Agent:** tls-security | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 3 Low, 9 Info

### TLS Architecture

All libcurl HTTPS connections route through a centralized `sx_curl_set_tls()` helper enforcing certificate verification by default. All mbedTLS contexts enforce TLS 1.2 minimum. The SMTP server correctly requires STARTTLS before AUTH on submission port 587.

| Finding | Severity | Description |
|---------|----------|-------------|
| TLS-01 | LOW | `SX_TLS_VERIFY` config allows disabling certificate verification (default: enabled, with warning) |
| TLS-02 | LOW | Opportunistic STARTTLS for MX delivery uses `VERIFY_OPTIONAL` (standard SMTP behavior per RFC 3207) |
| TLS-03 | LOW | TLS session tickets not explicitly disabled (default mbedTLS behavior, minimal risk) |
| TLS-04–TLS-12 | INFO | Positive: centralized TLS config, fail-closed on CA errors, STARTTLS enforcement on 587, consistent `sx_curl_set_tls()` usage across all callsites, plaintext HTTP only for localhost |

---

## 8. Hardcoded Credentials

**Agent:** credential-hardcode | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 2 Medium, 3 Low, 3 Info

### Assessment

**No actual secret values** (API keys, passwords, tokens) were found hardcoded in the source code. All sensitive configuration fields use empty-string defaults and are populated at runtime via environment variables or configuration file.

| Finding | Severity | Description |
|---------|----------|-------------|
| CRED-01 | MEDIUM | Hardcoded OAuth client IDs for OpenAI/Anthropic (public IDs, no client_secret; rotation requires rebuild) |
| CRED-02 | MEDIUM | Hardcoded infrastructure IP `20.53.240.54` in WASM config (infrastructure exposure, not a credential) |
| CRED-03 | LOW | Hardcoded internal service URLs in embedded config |
| CRED-04 | LOW | Embedded config key placeholders (empty-string defaults) |
| CRED-05 | LOW | Base64-encoded SMTP challenge strings (protocol constants, not secrets) |
| CRED-06–08 | INFO | Config architecture observations — centralized, runtime-populated |

---

## 9. File I/O & Data Handling

**Agent:** file-io | **Risk:** MEDIUM | **Findings:** 0 Critical, 0 High, 3 Medium, 5 Low, 5 Info

### File I/O Security Controls

- Centralized `sx_fopen()` sets `FD_CLOEXEC` on all file descriptors
- Consistent `mkstemp()`/`mkdtemp()` usage (31 call sites)
- `sx_config_is_sensitive()` redacts API keys/tokens in logs
- Config files written with `chmod(0600)`, directories with `mkdir(0700)`
- Traffic log headers redacted (x-api-key, Authorization)

| Finding | Severity | Description |
|---------|----------|-------------|
| FIO-01 | MEDIUM | API request/response bodies logged in plaintext to `api.log` (file is 0600; sensitive conversation content persists on disk) |
| FIO-02 | MEDIUM | Path traversal check uses `strstr(.., "..")` — double-encoding bypass possible in `ws2tcp.c` (Linux-only) and `scorpiox-server.c` |
| FIO-03 | MEDIUM | Predictable socket path `/tmp/sxmux` without UID qualification on fallback (**Linux-only**) |
| FIO-04–08 | LOW | Log rotation absent, no automatic log expiry, temp file cleanup on error paths |
| FIO-09–13 | INFO | Positive: FD_CLOEXEC, mkstemp, sensitive redaction, 0600 permissions, path traversal checks present |

---

## 10. Buffer Safety

**Agent:** buffer-safety | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 3 Low, 6 Info

### Buffer Safety Metrics

| Function | Count | Safety |
|----------|-------|--------|
| `snprintf` | 2,228 | ✅ Safe |
| `strncpy` | 601 | ✅ Safe |
| `memcpy` (bounded) | 514 | ✅ Safe |
| `fgets` | 107 | ✅ Safe |
| `sprintf` | 1 | ⚠️ Unsafe |
| `strcpy` / `strcat` / `gets` | 0 | ✅ Not used |

**Safe:Unsafe ratio = 3,464:1**

| Finding | Severity | Description |
|---------|----------|-------------|
| BUF-01 | LOW | Single `sprintf` in SHA256 hex encoding (bounded by construction — 32 iterations × 2 bytes into 65-byte buffer) |
| BUF-02 | LOW | Yamux stream buffer `uint32_t` doubling without overflow guard (requires authenticated FRP tunnel; window limits make impractical) |
| BUF-03 | LOW | IMAP FETCH response stack buffer truncation on large envelopes (8KB buffer; silently truncates, no overflow) |
| BUF-04–09 | INFO | Positive: zero strcpy/strcat/gets, consistent snprintf, bounded memcpy, proper fgets usage |

---

## 11. Memory Safety

**Agent:** memory-safety | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 2 Medium, 2 Low, 4 Info

### Memory Management Practices

The vast majority of `malloc`/`calloc`/`realloc` calls have proper NULL checks. `realloc` calls use a temporary variable pattern to avoid pointer loss.

| Finding | Severity | Description |
|---------|----------|-------------|
| MEM-01 | MEDIUM | `sx_api_init()` leaks `h->api` allocation if `h->display` calloc fails (redundant early return; one-time init path) |
| MEM-02 | MEDIUM | `scorpiox-unshare.c` leaks `stack` (8MB) on pipe failure (**Linux-only**; process exits shortly after) |
| MEM-03 | LOW | Terminal resize unchecked `calloc` return (graceful fallback — old buffers retained) |
| MEM-04 | LOW | HTTP header realloc pattern correct but missing explicit overflow check on count multiplication |
| MEM-05–08 | INFO | Positive: consistent NULL checks, proper realloc pattern, no use-after-free in critical paths, proper free-before-reassign |

---

## 12. Command Injection

**Agent:** command-injection | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 2 Medium, 3 Low, 5 Info

### Remediation Progress

The commit under review (`fe4a27d`) specifically fixes 3 Windows command injection findings (CJ-16, CJ-17, CJ-18). Most `system()` calls have been replaced with `fork()+execvp()` or `CreateProcess` on Windows.

| Finding | Severity | Description |
|---------|----------|-------------|
| CJ-01 | MEDIUM | `scorpiox-init` tool descriptor fields injected into shell commands (**Linux-only**; requires filesystem control of mounted volumes) |
| CJ-02 | MEDIUM | `scorpiox-init` `INIT_TOOLS` env var passed to `sx_system_safe()` unsanitized (**Linux-only**; requires env control) |
| CJ-03 | LOW | Windows: `SCORPIOX_INDEX_URL` config interpolated into `sx_popen_no_window()` (requires admin-level config modification) |
| CJ-04 | LOW | `scorpiox-server.c` Python path interpolation in command string (server-side, configurable path) |
| CJ-05 | LOW | Residual `sx_system_safe()` calls in non-critical paths (3 remaining call sites) |
| CJ-06–10 | INFO | Positive: CreateProcess on Windows, fork+execvp on Linux, metacharacter validation, quoted arguments |

---

## 13. Privilege & Access Control

**Agent:** privilege-access | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 3 Low, 12 Info

### Security Architecture

The container runtime (`scorpiox-unshare`) uses Linux user namespaces, pivot_root, overlayfs, and seccomp BPF. The runtime avoids setuid, uses `PR_SET_NO_NEW_PRIVS`, drops capabilities, clears environment before exec, and uses `fork+execvp` instead of `system()`.

**Important context:** The container runtime runs **trusted first-party agents on a private network** — not untrusted workloads. Findings are rated against this threat model, not against generic container security checklists.

| Finding | Severity | Description |
|---------|----------|-------------|
| PA-01 | LOW | Missing `CLONE_NEWIPC` in container namespace flags (Linux-only; defense-in-depth) |
| PA-02 | LOW | Seccomp filter blocks only 2 syscalls — default-allow policy (appropriate for trusted-agent model) |
| PA-03 | LOW | No cgroup resource limits on containers (trusted workloads; DoS from own agents is self-inflicted) |
| PA-04–15 | INFO | Positive: no setuid, PR_SET_NO_NEW_PRIVS enforced, capabilities dropped, environment cleared, fork+execvp, pivot_root with overlayfs, --privileged properly guarded |

---

## 14. Windows Deployment Analysis

**Agent:** windows-deployment | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 1 Medium, 3 Low, 9 Info

### Windows Security Highlights

- `CreateProcess`-based execution (no `cmd.exe /c` with user input)
- Metacharacter validation on shell arguments
- SHA256-verified self-updates
- User-scoped installation (no admin/UAC required)
- Proper argument quoting in Windows command construction

| Finding | Severity | Description |
|---------|----------|-------------|
| WIN-01 | MEDIUM | `scorpiox-bash.c` writes temp `.bat` files in `%TEMP%` for cmd.exe execution (TOCTOU risk in shared temp; mitigated by user-scoped %TEMP%) |
| WIN-02 | LOW | `scorpiox-wsl.exe` SHA256 verification bypass possible on parse error (dead code path — `check_for_update()` never called) |
| WIN-03 | LOW | DLL search order — `SetDllDirectory` used but `libsx.dll` loaded from `%LOCALAPPDATA%\scorpiox` (trusted path) |
| WIN-04 | LOW | `scorpiox-multiplexer` uses named pipes without explicit ACL (defaults to creator-only access on Windows) |
| WIN-05–13 | INFO | Positive: CreateProcess usage, SHA256 update verification, user-scoped PATH, no admin required, proper quoting |

---

## 15. Telemetry and Tracking

**Agent:** telemetry-tracking | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 0 Medium, 1 Low, 4 Info

### Verdict: No Tracking by Default

Both telemetry systems (`USAGE_TRACKING` and `EMIT_SESSION_TRACKING`) are **disabled** (`0`) in all configuration tiers: compiled defaults, embedded config, and shipped config file. No network telemetry occurs out of the box.

**Mandatory Data Collection: NONE** — All telemetry is fully opt-in.

| System | Default | Data Collected (when enabled) |
|--------|---------|-------------------------------|
| `USAGE_TRACKING` | OFF | Token counts, model name, session UUID, hostname, OS (no conversation content) |
| `EMIT_SESSION_TRACKING` | OFF | Full conversation messages, tool calls/results (privacy-sensitive; requires explicit opt-in) |

| Finding | Severity | Description |
|---------|----------|-------------|
| TEL-01 | LOW | `USAGE_TRACKING` guard logic uses negative check — if config value is NULL/missing, tracking would proceed (defaults to `"0"` so safe in practice) |
| TEL-02–05 | INFO | Positive: both systems disabled by default, gating at both caller and binary level, no mandatory collection, opt-in architecture |

---

## 16. Install Script & Distribution Security

**Agent:** install-script | **Risk:** MEDIUM | **Findings:** 0 Critical, 0 High, 3 Medium, 7 Low, 7 Info

### Distribution Security Architecture

- All downloads use HTTPS with HSTS on `dist.scorpiox.net`
- SHA256 checksum verification enforced by default on all platforms
- Per-commit build provenance via `build-info.json`
- Windows installer uses user-scoped PATH (no admin required)

| Finding | Severity | Description |
|---------|----------|-------------|
| IS-01 | MEDIUM | Same-origin checksums — binary and SHA256 hash served from same server; no independent code signing (GPG, Sigstore) |
| IS-02 | MEDIUM | `index.txt` branch control file has no integrity protection (server compromise could redirect to malicious branch) |
| IS-03 | MEDIUM | `verify_sha256_update()` parse-error bypass in `scorpiox-wsl.c` (dead code path — function never called) |
| IS-04 | LOW | `--no-verify` flag allows checksum bypass (Linux/macOS only) |
| IS-05 | LOW | Non-atomic installation — partial state on interruption |
| IS-06 | LOW | No HSTS on `get.scorpiox.net` installer endpoint |
| IS-07 | LOW | macOS binary uses ad-hoc codesign (no Developer ID) |
| IS-08 | LOW | No rollback mechanism on installation failure |
| IS-09 | LOW | `curl\|bash` pattern susceptible to partial download execution (Linux/macOS) |
| IS-10 | LOW | No reproducible build process |
| IS-11–17 | INFO | Positive: SHA256 enforced by default, HTTPS with HSTS on dist server, per-commit provenance, user-scoped Windows install |

---

## 17. WPF Launcher (scorpioxcode.exe)

**Agent:** wpf-launcher | **Risk:** LOW | **Findings:** 0 Critical, 0 High, 1 Medium, 5 Low, 16 Info

### WPF Security Highlights

- Zero NuGet package dependencies (all .NET Framework BCL)
- P/Invoke bindings with proper free-after-copy pattern
- SHA256-verified dependency downloads
- Static callback delegate pinning (prevents GC collection)
- No embedded browser (eliminates XSS-class risks)
- Deterministic build enabled

| Finding | Severity | Description |
|---------|----------|-------------|
| WPF-01 | MEDIUM | `CharSet.Ansi` marshalling corrupts non-ASCII input to UTF-8 native API (functional bug with minor security implications) |
| WPF-02 | LOW | Unbounded null-terminator scan in `PtrToStringUtf8` (trusted first-party library contract) |
| WPF-03 | LOW | Install script URL lacks explicit `https://` scheme |
| WPF-04 | LOW | Binary and checksum from same origin (no code signing) |
| WPF-05 | LOW | Remote telemetry sends machine name and file paths (minor PII) |
| WPF-06 | LOW | `AutoUpdater` SHA256 verification — same-origin checksum pattern |
| WPF-07–22 | INFO | Positive: zero NuGet deps, proper P/Invoke safety, SHA256 downloads, no unsafe code, deterministic builds |

---

## 18. Consolidated Risk Matrix

### Medium-Severity Findings

| # | Area | Finding | Severity | Windows | Recommendation |
|---|------|---------|----------|---------|----------------|
| 1 | Credentials | Hardcoded OAuth client IDs (public, no secret) | MEDIUM | ✅ | Make configurable for rotation |
| 2 | Credentials | Hardcoded infrastructure IP in WASM config | MEDIUM | ✅ | Use DNS hostname consistently |
| 3 | File I/O | API bodies logged in plaintext (0600 perms) | MEDIUM | ✅ | Add config to disable body logging; log rotation |
| 4 | File I/O | Path traversal double-encoding bypass | MEDIUM | ✅ | Use `realpath()` validation consistently |
| 5 | File I/O | Predictable `/tmp/sxmux` socket path | MEDIUM | ❌ | Always append UID to path |
| 6 | Memory | `sx_api_init()` allocation leak on error | MEDIUM | ✅ | Remove redundant early return (line 108) |
| 7 | Memory | `scorpiox-unshare` stack leak on pipe fail | MEDIUM | ❌ | Add `free(stack)` before error return |
| 8 | Command Inj. | `scorpiox-init` descriptor field injection | MEDIUM | ❌ | Use fork+execvp or validate with allowlist |
| 9 | Command Inj. | `scorpiox-init` INIT_TOOLS env injection | MEDIUM | ❌ | Add `is_safe_shell_arg()` validation |
| 10 | Install | Same-origin checksums (no code signing) | MEDIUM | ✅ | Implement GPG or Sigstore signing |
| 11 | Install | `index.txt` branch control unprotected | MEDIUM | ✅ | Add integrity verification for index file |
| 12 | Install | SHA256 parse-error bypass (dead code) | MEDIUM | ✅ | Fix parse-error return handling or remove dead code |
| 13 | Windows | Temp `.bat` file TOCTOU in scorpiox-bash | MEDIUM | ✅ | Use unique temp filenames with mkstemp pattern |
| 14 | WPF | CharSet.Ansi corrupts non-ASCII to UTF-8 API | MEDIUM | ✅ | Use manual UTF-8 marshalling for string parameters |

### Low-Severity Findings (Summary by Area)

| Area | Count | Key Themes |
|------|-------|------------|
| Supply Chain | 1 | Version tracking manifest |
| Build Provenance | 2 | Bridge Makefile hardening, PIE |
| Network Endpoints | 3 | WASM hardcoded IP, DNS bind, registry TLS |
| TLS Security | 3 | SX_TLS_VERIFY toggle, opportunistic STARTTLS, session tickets |
| Credentials | 3 | Internal service URLs, config placeholders, SMTP constants |
| File I/O | 5 | Log rotation, temp cleanup, error-path handling |
| Buffer Safety | 3 | Single sprintf, yamux overflow guard, IMAP truncation |
| Memory Safety | 2 | Terminal resize, header realloc |
| Command Injection | 3 | Config URL interpolation, Python path, residual sx_system_safe |
| Privilege & Access | 3 | CLONE_NEWIPC, seccomp depth, cgroup limits |
| Windows Deployment | 3 | Dead-code SHA256 bypass, DLL search, named pipe ACL |
| Telemetry | 1 | Usage tracking guard logic |
| Install Script | 7 | --no-verify, non-atomic install, HSTS, macOS codesign, curl\|bash, rollback, reproducibility |
| WPF Launcher | 5 | PtrToStringUtf8 bounds, install URL scheme, same-origin checksums, telemetry PII, auto-updater |

---

## 19. Conclusion & Recommendations

### Overall Assessment

SCORPIOX CODE achieves a **LOW** overall risk rating with **zero Critical and zero High findings** across 155 total observations from 14 audit agents. This represents a **significant improvement** from the previous audit (225 findings including 2 Critical and 6 High). The commit under review specifically addresses Windows command injection vulnerabilities, demonstrating active security remediation.

### Top Recommendations (Priority Order)

1. **Implement independent code signing** (GPG or Sigstore) for release artifacts to establish a trust anchor independent of the distribution server (IS-01)
2. **Fix `CharSet.Ansi` marshalling** in the WPF launcher to prevent non-ASCII input corruption (WPF-02)
3. **Add log rotation and configurable body logging** to prevent indefinite sensitive conversation persistence (FIO-01)
4. **Use `realpath()` validation** consistently for all path traversal checks (FIO-02)
5. **Remove dead code** in `scorpiox-wsl.c` `check_for_update()` or fix the SHA256 parse-error handling (IS-03)
6. **Add input validation** for `scorpiox-init` descriptor fields and `INIT_TOOLS` environment variable (CJ-01, CJ-02, Linux-only)
7. **Fix the `sx_api_init()` memory leak** by removing the redundant early return at line 108 (MEM-01)
8. **Expand seccomp blocklist** with additional commonly-blocked syscalls for defense-in-depth (PA-02, Linux-only)

### Strengths

- Exceptional buffer safety discipline (3,464:1 safe-to-unsafe ratio)
- Centralized TLS configuration with fail-closed defaults
- Zero third-party runtime dependencies in Windows distribution
- Both telemetry systems disabled by default with no mandatory collection
- Active security remediation visible in commit history
- Clean supply chain — only 2 vendored libraries, both current LTS/stable releases
- Comprehensive security hardening flags in build system

---

## Appendix: Audit Methodology

### Agents Deployed

| # | Agent | Scope |
|---|-------|-------|
| 1 | supply-chain | Package managers, vendored libraries, dependency analysis |
| 2 | build-provenance | Build system, compiler/linker flags, code generation |
| 3 | network-endpoints | Hardcoded URLs, IPs, domains, ports, binding addresses |
| 4 | tls-security | Certificate verification, protocol versions, cipher suites |
| 5 | credential-hardcode | API keys, passwords, tokens, secrets in source |
| 6 | file-io | File operations, temp files, permissions, data at rest |
| 7 | buffer-safety | sprintf/strcpy/strcat usage, bounds checking, format strings |
| 8 | memory-safety | malloc/realloc NULL checks, use-after-free, double-free, leaks |
| 9 | command-injection | system(), popen(), exec*() with untrusted input |
| 10 | privilege-access | setuid, root requirements, sandboxing, namespaces |
| 11 | windows-deployment | Windows binary enumeration, platform-specific risks |
| 12 | telemetry-tracking | Phone-home behavior, analytics, data collection defaults |
| 13 | install-script | Installation scripts, update mechanism, distribution integrity |
| 14 | wpf-launcher | WPF desktop application security (P/Invoke, downloads, telemetry) |

### Analysis Approach

- **Static analysis** of 224 non-vendor C/H source files (148,135 lines)
- **Configuration review** of embedded defaults, shipped config files, and build system
- **Threat model-aware rating**: Findings rated against the actual deployment context (trusted first-party agents on private network), not against generic container/server security checklists
- **Platform-scoped findings**: Each finding tagged with platform applicability (Windows/Linux/macOS/All)
- **Delta analysis**: Comparison with previous audit at commit `6851bc0` to track remediation progress

### Severity Definitions

| Severity | Definition |
|----------|------------|
| **Critical** | Remotely exploitable without authentication; immediate data breach risk |
| **High** | Exploitable with minimal prerequisites; significant security impact |
| **Medium** | Exploitable with specific prerequisites (local access, config modification, filesystem control); moderate impact |
| **Low** | Defense-in-depth hardening; theoretical risk with minimal practical exploitability |
| **Info** | Positive security observation or architectural note |