feat: IOC pre-check via public threat intel (AegisAUR integration)

- ioc_fetcher.py: fetches from HedgeDoc, Atomic Arch Gist, Arch Security
  Tracker, AUR RPC orphan detection (concurrent)
- scanner.py: IOC pre-check before LLM scan — known malicious packages
  get instant MALICIOUS verdict without LLM cost
- typosquatting check with Levenshtein distance
- server.py: API returns ioc_matches + typosquat_matches
- README: threat intel sources documented

Sources ported from AegisAUR (Rust) to Python.

README.md

@@ -92,18 +92,30 @@ cache:
 
 ## How It Works
 
-1. **Request:** Client asks for `aur-shield/<package>`
+1. **IOC Pre-Check:** Package name checked against public threat lists (HedgeDoc, Atomic Arch Gist, Arch Security Tracker, AUR Orphan detection). Known malicious → instant block, no LLM needed.
 2. **Fetch:** Server pulls PKGBUILD + .SRCINFO from AUR API
-3. **Scan:** LLM analyzes the PKGBUILD for:
+3. **Regex Pre-Scan:** Fast pattern matching for `curl|bash`, `eval|base64`, `/dev/tcp`, etc.
+4. **LLM Scan:** Ollama analyzes the PKGBUILD for:
    - Suspicious `source=()` URLs (npm, tor, raw IPs)
    - Obfuscated bash (`eval`, `base64 -d`, hex encoding)
    - Reverse shells, `nc`, `/dev/tcp`
    - `post_install` hooks creating services/cronjobs
    - Typosquatting package names
    - Unusual `depends` for the package type
-4. **Build:** If clean, `makepkg` builds the package
+5. **Build:** If clean, `makepkg` builds the package
-5. **Serve:** `repo-add` adds it to the local pacman repo
+6. **Serve:** `repo-add` adds it to the local pacman repo
-6. **Cache:** Approved packages stay cached until upstream update
+7. **Cache:** Approved packages stay cached until upstream update
+
+## Threat Intelligence Sources
+
+Based on [AegisAUR](https://gitea.die-heimatlosen.eu/arch_agent/aegisaur) IOC fetcher:
+
+| Source | Type | Freshness | URL |
+|--------|------|-----------|-----|
+| HedgeDoc | Live paste | Always current | `md.archlinux.org/s/SxbqukK6IA` |
+| Atomic Arch Gist | GitHub Gist | Versioned | `gist.githubusercontent.com/Kidev/...` |
+| Arch Security | Official advisory | Slow but authoritative | `security.archlinux.org` |
+| AUR RPC | API | Real-time | `aur.archlinux.org/rpc/v5` |
 
 ## API
 

aur_shield/ioc_fetcher.py

@@ -0,0 +1,275 @@
+"""IOC fetcher — pulls known malicious package lists from public sources.
+
+Based on AegisAUR's ioc_fetcher.rs. Checks package names against:
+1. HedgeDoc live paste (Arch Linux community — always current)
+2. Atomic Arch Gist (GitHub — versioned fallback)
+3. Arch Security Tracker (official advisories)
+4. AUR RPC (orphan takeover detection)
+"""
+from __future__ import annotations
+
+import asyncio
+import json
+import re
+import time
+from dataclasses import dataclass, field
+from enum import Enum
+from typing import Any
+
+import httpx
+
+
+class ThreatType(str, Enum):
+    MALICIOUS_BUILD = "MaliciousBuildScript"
+    CREDENTIAL_STEALER = "CredentialStealer"
+    ROOTKIT = "Rootkit"
+    CRYPTOMINER = "Cryptominer"
+    BACKDOOR = "Backdoor"
+    TYPOSQUATTING = "Typosquatting"
+    ORPHAN_TAKEOVER = "OrphanTakeover"
+    UNKNOWN = "Unknown"
+
+
+class Confidence(str, Enum):
+    CRITICAL = "Critical"
+    HIGH = "High"
+    MEDIUM = "Medium"
+    LOW = "Low"
+
+
+@dataclass
+class IOCEntry:
+    package_name: str
+    threat_type: ThreatType
+    source: str
+    description: str
+    confidence: Confidence
+    discovered_date: str = ""
+
+
+@dataclass
+class IOCResult:
+    package: str
+    matches: list[IOCEntry] = field(default_factory=list)
+    sources_checked: list[str] = field(default_factory=list)
+    is_known_malicious: bool = False
+
+
+# Source URLs (from AegisAUR)
+HEDGEDOC_URL = "https://md.archlinux.org/s/SxbqukK6IA"
+GIST_URL = "https://gist.githubusercontent.com/Kidev/85756c3dcad3623ca5604a8135bafd14/raw"
+ARCH_SECURITY_URL = "https://security.archlinux.org/advisory/atomic-arch/json"
+AUR_RPC_ORPHAN_URL = "https://aur.archlinux.org/rpc/v5/search?by=maintainer&arg=orphan"
+
+CACHE_TTL = 300  # 5 minutes
+
+
+def _valid_pkg_name(name: str) -> bool:
+    """Validate that a string looks like a valid package name."""
+    if not name or len(name) > 100:
+        return False
+    return all(
+        c.isascii() and (c.islower() or c.isdigit() or c in "-_.")
+        for c in name
+    )
+
+
+async def _fetch_hedgedoc(client: httpx.AsyncClient) -> list[IOCEntry]:
+    """Fetch IOC list from Arch Linux HedgeDoc live paste."""
+    try:
+        resp = await client.get(HEDGEDOC_URL)
+        resp.raise_for_status()
+    except Exception:
+        return []
+
+    threats = []
+    for line in resp.text.splitlines():
+        pkg = line.strip()
+        if pkg.startswith(("- ", "* ")):
+            pkg = pkg[2:].strip()
+        # Skip headers, comments, metadata
+        if (not pkg or pkg.startswith(("#", "---", "**", "Arch Linux",
+             "Liste", "Quelle:", "Aktualisiert:"))):
+            continue
+        if _valid_pkg_name(pkg):
+            threats.append(IOCEntry(
+                package_name=pkg,
+                threat_type=ThreatType.MALICIOUS_BUILD,
+                source="hedgedoc_live",
+                description="Atomic Arch Supply Chain Attack — Live HedgeDoc",
+                confidence=Confidence.CRITICAL,
+            ))
+    return threats
+
+
+async def _fetch_gist(client: httpx.AsyncClient) -> list[IOCEntry]:
+    """Fetch IOC list from Atomic Arch GitHub Gist."""
+    try:
+        resp = await client.get(GIST_URL)
+        resp.raise_for_status()
+    except Exception:
+        return []
+
+    threats = []
+    for line in resp.text.splitlines():
+        pkg = line.strip()
+        if not pkg or pkg.startswith(("#", "echo")):
+            continue
+        # Try JSON array
+        if pkg.startswith("["):
+            try:
+                for p in json.loads(pkg):
+                    if _valid_pkg_name(p):
+                        threats.append(IOCEntry(
+                            package_name=p,
+                            threat_type=ThreatType.MALICIOUS_BUILD,
+                            source="atomic_arch_gist",
+                            description="Atomic Arch — Gist Fallback",
+                            confidence=Confidence.CRITICAL,
+                        ))
+                continue
+            except json.JSONDecodeError:
+                pass
+        # Plain package name
+        if _valid_pkg_name(pkg) and " " not in pkg and "/" not in pkg:
+            threats.append(IOCEntry(
+                package_name=pkg,
+                threat_type=ThreatType.MALICIOUS_BUILD,
+                source="atomic_arch_gist",
+                description="Atomic Arch — Gist Fallback",
+                confidence=Confidence.CRITICAL,
+            ))
+    return threats
+
+
+async def _fetch_arch_security(client: httpx.AsyncClient) -> list[IOCEntry]:
+    """Fetch official Arch Linux security advisories."""
+    try:
+        resp = await client.get(ARCH_SECURITY_URL)
+        if not resp.status_code == 200:
+            return []
+        data = resp.json()
+    except Exception:
+        return []
+
+    threats = []
+    packages = data.get("packages", [])
+    if isinstance(packages, list):
+        for pkg in packages:
+            name = pkg if isinstance(pkg, str) else pkg.get("name", "")
+            if _valid_pkg_name(name):
+                threats.append(IOCEntry(
+                    package_name=name,
+                    threat_type=ThreatType.MALICIOUS_BUILD,
+                    source="arch_security",
+                    description="Arch Linux Security Advisory — Atomic Arch",
+                    confidence=Confidence.CRITICAL,
+                ))
+    return threats
+
+
+async def _fetch_orphan_takeover(client: httpx.AsyncClient) -> list[IOCEntry]:
+    """Fetch suspicious orphaned packages from AUR RPC."""
+    try:
+        resp = await client.get(AUR_RPC_ORPHAN_URL)
+        if not resp.status_code == 200:
+            return []
+        data = resp.json()
+    except Exception:
+        return []
+
+    threats = []
+    results = data.get("results", [])
+    for pkg in results[:50]:
+        name = pkg.get("Name", "")
+        votes = pkg.get("NumVotes", 0)
+        if _valid_pkg_name(name) and votes < 10:
+            threats.append(IOCEntry(
+                package_name=name,
+                threat_type=ThreatType.ORPHAN_TAKEOVER,
+                source="aur_rpc",
+                description="AUR Orphaned Package — low votes, possible takeover",
+                confidence=Confidence.MEDIUM,
+            ))
+    return threats
+
+
+async def fetch_all_iocs() -> list[IOCEntry]:
+    """Fetch IOCs from all sources concurrently."""
+    async with httpx.AsyncClient(timeout=30) as client:
+        results = await asyncio.gather(
+            _fetch_hedgedoc(client),
+            _fetch_gist(client),
+            _fetch_arch_security(client),
+            _fetch_orphan_takeover(client),
+            return_exceptions=True,
+        )
+
+    all_threats = []
+    for r in results:
+        if isinstance(r, list):
+            all_threats.extend(r)
+    return all_threats
+
+
+def check_package_against_iocs(
+    package: str,
+    iocs: list[IOCEntry],
+) -> IOCResult:
+    """Check if a package name appears in known malicious lists."""
+    matches = [
+        ioc for ioc in iocs
+        if ioc.package_name.lower() == package.lower()
+    ]
+
+    return IOCResult(
+        package=package,
+        matches=matches,
+        sources_checked=list({ioc.source for ioc in iocs}),
+        is_known_malicious=any(
+            ioc.confidence == Confidence.CRITICAL
+            for ioc in matches
+        ),
+    )
+
+
+def check_typosquatting(
+    package: str,
+    iocs: list[IOCEntry],
+    threshold: int = 2,
+) -> list[IOCEntry]:
+    """Check for typosquatting — package names close to known malicious ones.
+
+    Uses simple Levenshtein distance (no external dep).
+    """
+    def levenshtein(a: str, b: str) -> int:
+        if len(a) < len(b):
+            a, b = b, a
+        if not b:
+            return len(a)
+        prev = list(range(len(b) + 1))
+        for i, ca in enumerate(a):
+            curr = [i + 1]
+            for j, cb in enumerate(b):
+                curr.append(min(
+                    prev[j + 1] + 1,
+                    curr[j] + 1,
+                    prev[j] + (ca != cb),
+                ))
+            prev = curr
+        return prev[-1]
+
+    matches = []
+    for ioc in iocs:
+        if ioc.package_name.lower() == package.lower():
+            continue  # exact match handled elsewhere
+        dist = levenshtein(package.lower(), ioc.package_name.lower())
+        if dist <= threshold and dist > 0:
+            matches.append(IOCEntry(
+                package_name=ioc.package_name,
+                threat_type=ThreatType.TYPOSQUATTING,
+                source=f"typosquatting_check(dist={dist})",
+                description=f"Close match to known malicious '{ioc.package_name}' (distance={dist})",
+                confidence=Confidence.HIGH,
+            ))
+    return matches

aur_shield/scanner.py

@@ -12,6 +12,10 @@ import httpx
 
 from .config import OllamaConfig, SecurityConfig
 from .aur_client import extract_sources, extract_install_hooks
+from .ioc_fetcher import (
+    fetch_all_iocs, check_package_against_iocs,
+    check_typosquatting, IOCEntry, IOCResult, Confidence,
+)
 
 
 class ScanVerdict(str, Enum):
@@ -30,6 +34,61 @@ class ScanResult:
     reasoning: str = ""
     sources_checked: list[str] = field(default_factory=list)
     raw_response: str = ""
+    ioc_matches: list[dict] = field(default_factory=list)
+    typosquat_matches: list[dict] = field(default_factory=list)
+
+
+async def pre_check_iocs(package: str) -> tuple[list[str], list[dict], list[dict]]:
+    """Check package against public IOC lists before LLM scan.
+
+    Returns (findings, ioc_matches, typosquat_matches).
+    If ioc_matches contains critical entries, the package is known malicious.
+    """
+    findings = []
+    ioc_dicts = []
+    typo_dicts = []
+
+    try:
+        iocs = await fetch_all_iocs()
+
+        # Exact match
+        result = check_package_against_iocs(package, iocs)
+        if result.matches:
+            for m in result.matches:
+                ioc_dicts.append({
+                    "package": m.package_name,
+                    "threat": m.threat_type.value,
+                    "source": m.source,
+                    "confidence": m.confidence.value,
+                    "description": m.description,
+                })
+                if m.confidence == Confidence.CRITICAL:
+                    findings.append(
+                        f"KNOWN MALICIOUS: '{package}' found in {m.source} "
+                        f"({m.threat_type.value}: {m.description})"
+                    )
+                else:
+                    findings.append(
+                        f"Suspicious: '{package}' flagged in {m.source} "
+                        f"({m.threat_type.value})"
+                    )
+
+        # Typosquatting
+        typo_matches = check_typosquatting(package, iocs)
+        for m in typo_matches:
+            typo_dicts.append({
+                "package": m.package_name,
+                "threat": m.threat_type.value,
+                "source": m.source,
+                "confidence": m.confidence.value,
+                "description": m.description,
+            })
+            findings.append(f"Typosquatting: {m.description}")
+
+    except Exception as e:
+        findings.append(f"IOC check error (non-blocking): {e}")
+
+    return findings, ioc_dicts, typo_dicts
 
 
 SYSTEM_PROMPT = """\
@@ -76,7 +135,27 @@ async def scan_pkgbuild(
     ollama_cfg: OllamaConfig,
     sec_cfg: SecurityConfig,
 ) -> ScanResult:
-    """Scan a PKGBUILD with regex pre-check + LLM analysis."""
+    """Scan a PKGBUILD with IOC pre-check + regex pre-scan + LLM analysis."""
+
+    # 0. IOC pre-check — public threat intelligence
+    ioc_findings, ioc_matches, typo_matches = await pre_check_iocs(package_name)
+
+    # If package is known malicious from public lists, skip LLM scan
+    has_critical_ioc = any(
+        m.get("confidence") == "Critical" for m in ioc_matches
+    )
+    if has_critical_ioc:
+        return ScanResult(
+            package=package_name,
+            verdict=ScanVerdict.MALICIOUS,
+            confidence=1.0,
+            findings=ioc_findings,
+            reasoning="Package found in public IOC lists (known malicious). "
+                      "LLM scan skipped — package is confirmed malicious by "
+                      "Arch Linux security advisories.",
+            ioc_matches=ioc_matches,
+            typosquat_matches=typo_matches,
+        )
 
     # 1. Fast regex pre-scan
     regex_findings = _pre_scan(pkgbuild, sec_cfg)
@@ -124,8 +203,8 @@ async def scan_pkgbuild(
         raw = data.get("message", {}).get("content", "")
         result = _parse_llm_response(raw)
 
-        # Combine with regex findings
+        # Combine with regex + IOC findings
-        all_findings = result.get("findings", []) + regex_findings
+        all_findings = result.get("findings", []) + regex_findings + ioc_findings
         verdict = result.get("verdict", "suspicious")
         confidence = result.get("confidence", 0.5)
 
@@ -134,6 +213,11 @@ async def scan_pkgbuild(
             verdict = "suspicious"
             confidence = max(confidence, 0.7)
 
+        # If IOC found non-critical matches, escalate
+        if ioc_matches and verdict == "clean":
+            verdict = "suspicious"
+            confidence = max(confidence, 0.8)
+
         return ScanResult(
             package=package_name,
             verdict=ScanVerdict(verdict),
@@ -142,6 +226,8 @@ async def scan_pkgbuild(
             reasoning=result.get("reasoning", ""),
             sources_checked=sources,
             raw_response=raw,
+            ioc_matches=ioc_matches,
+            typosquat_matches=typo_matches,
         )
 
     except Exception as e:

aur_shield/server.py

@@ -100,6 +100,8 @@ async def scan_package(package: str) -> dict[str, Any]:
         "reasoning": result.reasoning,
         "sources": result.sources_checked,
         "version": source.package.version,
+        "ioc_matches": result.ioc_matches,
+        "typosquat_matches": result.typosquat_matches,
     }