HTTP Request Smuggling Practical Handbook
Paste the following prompt into your AI chat to install this skill:
Follow https://skillhub.cn/install/skillhub.md and install @user_3c6cb52e/request-smuggling-gpev.
About this skill
Problem being addressed
When a CDN, reverse proxy, and origin do not agree on where one HTTP request ends, a single logical request can be split into multiple requests or swallow bytes from a later request. This can produce abnormal request concatenation, cache-key misalignment, cross-user request interference, and hard-to-reproduce path or header anomalies during HTTP/2 to HTTP/1.1 downgrade. The issue is not plain header injection; it is a message boundary disagreement between hops over Content-Length, Transfer-Encoding, pseudo-headers, and body framing.
How the skill works
The skill treats request smuggling as a boundary-framing investigation rather than a generic injection workflow:
- CL.TE / TE.CL / TE.TE: compares how the front hop and back hop interpret length headers, chunked encoding, and
Transfer-Encodingobfuscation variants. - HTTP/2 downgrade: examines
:method,:path, duplicate headers, lowercase normalization, and handling ofConnection-related headers when H1 requests are reconstructed. - Client-side desync: evaluates browser
Fetch,no-cors, CORS, and connection reuse effects that may create request or response queue misalignment. - Tool-assisted verification: the material references tools such as
Burp Suite — HTTP Request Smuggler,defparam/smuggler, andsimple-http-smuggler-generatorfor generating probes, checking timing deltas, and assembling common variants. - Detection routing: uses a decision-tree style to confirm whether the behavior is an HTTP message-boundary issue before moving to adjacent areas such as CRLF injection, Web Cache Deception, SSTI, or SQLi.
Scope and cautions
This skill is useful for security engineers, SREs, and backend developers who suspect a front-end plus origin two-hop mismatch. It is best used for triage, reproduction, log comparison, and minimal probing, not for uncontrolled fuzzing in production. Validate the topology first, keep concurrency low, avoid affecting other users, and focus on parser disagreement, connection reuse, and cache-key or response-queue anomalies.
Use Cases
- Investigate request concatenation caused by proxy and origin disagreement over CL.TE or TE.CL boundaries.
- Audit HTTP/2 to HTTP/1.1 downgrade for pseudo-header mapping and duplicate-header handling that can split requests.
- Use minimal probes to verify whether browser Fetch, CORS, and connection reuse trigger client-side desync.
- After confirming a message-boundary issue, separate CRLF, cache deception, and injection-related routing.
Best For
- Backend engineers hardening gateway-to-origin security who need to locate request-boundary parsing differences.
- SREs investigating abnormal production request chains to determine whether proxy-origin desync caused them.
- Security engineers testing web apps who need to validate CL.TE, TE.CL, and HTTP/2 downgrade risks.
- Infrastructure engineers maintaining CDN or reverse-proxy rules who need to troubleshoot cache-key and response-queue misalignment.
Related Skills
Analyzes network captures from Wireshark, tcpdump, Fiddler, and Charles, then pinpoints TCP, HTTP, DNS, and TLS issues with filter expressions and remediation steps.
Lightweight Python Linux HIDS exposing SSH brute-force, web attack, and webshell alerts via MCP with ban controls.
Detects AI-generated images from Stable Diffusion, Midjourney, and GPT-4o via Tencent Cloud image moderation and returns risk scores.
Sets up a baseline Dynamo router, enables KV-aware routing when appropriate, and smoke tests the frontend endpoint locally or on Kubernetes.