Local AI for Cyber Security Finding phpIPAM LFI and myVesta ...

src/AI/AI-Assisted-Fuzzing-and-Vulnerability-Discovery.md

@@ -133,6 +133,37 @@ Implement a queue where confirmed PoV-validated patches and *speculative* patche
 
 ---
 
+## 6. Deterministic File-by-File AI Code Review
+
+A frequent failure mode in AI-assisted review is asking one agent to inspect a whole repository and hoping it chooses the right files and grep terms. A more reliable pattern is to **force repository coverage**:
+
+1. Enumerate source files.
+2. Send **one file at a time** plus minimal context (entrypoint, imports, nearby routes/callers).
+3. Require a **structured report** for each file: sources, sink, missing validation, exploit preconditions, and confidence.
+4. Deduplicate reports by **source→sink pattern** and manually validate the high-risk ones.
+
+This is token-heavy and noisy, but it is very effective at surfacing **simple high-impact source/sink bugs** that broad agentic reviews often miss.
+
+### Triage patterns worth prioritising
+
+- **Dynamic PHP include/dispatch**: attacker-controlled route/controller/module names flowing into `require_once()`, `include()`, or `include_once()` without strict allowlisting and path canonicalisation.
+- **Shell execution in admin/account workflows**: usernames, domains, FTP accounts, or other identifiers reaching `exec()`, `system()`, `shell_exec()`, `passthru()`, `proc_open()`, or `popen()`.
+- **Structured input fan-in**: the same parameter accepted from `$_GET`, `$_POST`, JSON, XML, or framework body parsers and later reused in filesystem or OS-command sinks.
+
+### Two practical bug classes this method finds well
+
+- **PHP controller-dispatch LFI/RCE**: if a request-controlled controller name is concatenated into `require_once()` with no allowlist/path normalisation, traversal sequences such as `../` can make PHP include an unintended local `.php` file. If the attacker can point the include at a planted or otherwise useful PHP file, the LFI becomes code execution. See [File Inclusion / Path Traversal](../pentesting-web/file-inclusion/README.md).
+- **Authenticated command injection in hosting/admin panels**: if account-management fields such as usernames are embedded into shell commands, a low-privileged authenticated user may turn a normal create/delete action into RCE. The impact increases when the panel executes the command as a more privileged service account. See [Command Injection](../pentesting-web/command-injection.md).
+
+### Practical review notes
+
+- Use the LLM as a **source/sink reviewer first**, not as an exploit generator.
+- Ask it to explicitly list the **attacker-controlled field**, the **normalization/validation gap**, and the **dangerous sink**.
+- For large repos, deterministic per-file review is often more reliable than a single autonomous agent run, even when the latter uses a stronger model.
+- Expect weaker results on **Broken Access Control** and other business-logic bugs where exploitability depends on cross-file assumptions, role semantics, or product-specific threat models.
+
+---
+
 ## Putting It All Together
 An end-to-end CRS (Cyber Reasoning System) may wire the components like this:
 
@@ -152,6 +183,9 @@ graph TD
 ---
 
 ## References
+* [Project Black - Local AI for Cyber Security: Finding phpIPAM LFI and myVesta Authenticated RCE](https://projectblack.io/blog/local-ai-for-cyber-security)
+* [Strix](https://github.com/usestrix/strix)
+* [GitHub Copilot community security-review skill](https://github.com/github/awesome-copilot/blob/main/skills/security-review/SKILL.md)
 * [Trail of Bits – AIxCC finals: Tale of the tape](https://blog.trailofbits.com/2025/08/07/aixcc-finals-tale-of-the-tape/)
 * [CTF Radiooo AIxCC finalist interviews](https://www.youtube.com/@ctfradiooo)
 {{#include ../banners/hacktricks-training.md}}