Add SSTI and Header Injection vulnerability skills

[MoDarK-MK]

Adds two new vulnerability skill files under strix/skills/vulnerabilities/:

• ssti.md — Server-side template injection across Jinja / Mako / Velocity / Freemarker / Thymeleaf (SpEL) / Twig / Smarty / Blade / ERB / Haml / Handlebars / Nunjucks / EJS. Engine fingerprinting, per-language gadget chains, sandbox escape patterns, RCE primitives, post-exploitation.
• header_injection.md — HTTP header injection covering CRLF / response splitting / request smuggling, cache poisoning, Host-header confusion, cookie tossing, X-Forwarded-* / X-Original-URL spoofing, Content-Type / encoding tricks, header-driven XSS and open redirects, HTTP/2 frame confusion.

Both files follow the established skill template (frontmatter + Attack Surface / High-Value Targets / Reconnaissance / Key Vulnerabilities / Bypass Techniques / Testing Methodology / Validation / False Positives / Impact / Pro Tips / Summary), matching the format used by sql_injection.md, csrf.md, ssrf.md, xss.md, and the rest of the existing vulnerability skills.

Test plan

• Skim both files end-to-end for prose feel vs peer skills
• Spot-check a couple of payloads (${T(java.lang.Runtime).getRuntime().exec('id')} for SpEL, {{cycler.__init__.__globals__.os.popen('id').read()}} for Jinja)
• Confirm strix/skills/__init__.py auto-discovers the new files (no registration needed for skills, per existing convention)

[MoDarK-MK]

and Header Injection vulnerability

[0xallam]

@greptile

[greptile-apps]

Greptile Summary

Adds two new vulnerability skill files (ssti.md and header_injection.md) following the established frontmatter + section template used by peer skills. Both files are broadly accurate and well-structured; earlier review rounds addressed the most significant payload errors (Velocity $Runtime, SpEL Process.toString, Jinja __import__ scope, and MRO index). Two minor issues remain: the standalone {{('__cl'+'ass__')}} snippet in ssti.md's filter-evasion section produces a string rather than an attribute bypass, and X-Accel-Redirect in header_injection.md's "Open Redirect" section is an Nginx internal file-serving directive (LFI impact) rather than an external redirect primitive.

Confidence Score: 5/5

Safe to merge — remaining findings are documentation-level P2s that do not affect runtime behaviour

All findings are P2; no P0/P1 issues remain after previous review rounds corrected the payload accuracy problems

Both files warrant a quick re-read of the two flagged snippets before merging, but neither requires blocking changes

Important Files Changed

Filename	Overview
strix/skills/vulnerabilities/ssti.md	Comprehensive SSTI skill covering engine fingerprinting, gadget chains, and sandbox escapes; previous review rounds addressed the most critical payload inaccuracies; one remaining minor issue with a misleading standalone string-concatenation bypass example in the filter-evasion section
strix/skills/vulnerabilities/header_injection.md	Well-structured header injection skill; one categorisation issue: X-Accel-Redirect is listed under Open Redirect via Headers but its actual impact is internal file read / LFI, not an external redirect

Prompt To Fix All With AI

Fix the following 2 code review issues. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 2
strix/skills/vulnerabilities/ssti.md:409-410
**`{{('__cl'+'ass__')}}` renders a string, not an attribute access**

The first example in this bullet concatenates two literals into the string `__class__` and renders it verbatim — it performs no attribute lookup and achieves no bypass on its own. A tester who copies it in isolation will get the output `__class__` rather than the class object, which could lead them to conclude string concatenation is an ineffective evasion. The only functional bypass here is the `|attr(...)` form shown immediately after; the bare concatenation snippet should be folded into that form or removed to avoid the confusion.

### Issue 2 of 2
strix/skills/vulnerabilities/header_injection.md:138
**`X-Accel-Redirect` is an LFI/file-read primitive, not an open-redirect**

`X-Accel-Redirect` is an Nginx internal-redirect header: when the upstream application sets it, Nginx serves the *local file path* specified in the value, bypassing normal request routing. Its impact is unauthorized file read / SSRF-to-internal-file, not an open redirect to an external URL. Listing it under "Open Redirect via Headers" misrepresents both the attack class and the impact, and could cause a tester to apply the wrong validation criteria (they'd look for an external redirect instead of an internal file being served). It fits better under a server-side file access or SSRF-adjacent section, or at minimum the parenthetical should note that the impact is internal file serving rather than redirect.

_{Reviews (5): Last reviewed commit: "add header-injection and ssti vulnerabil..." | Re-trigger Greptile}

[Copilot]

Pull request overview

This PR adds two new vulnerability skill documents under strix/skills/vulnerabilities, expanding the repository’s offensive-security guidance with playbooks for SSTI and HTTP header injection. It fits the existing skills library pattern by providing reconnaissance, validation, false-positive handling, and impact guidance for additional vulnerability classes.

Changes:

• Add a new SSTI skill covering engine fingerprinting, bypass techniques, and language-specific RCE primitives.
• Add a new HTTP header injection skill covering CRLF injection, cache poisoning, Host-header abuse, and proxy/header trust issues.
• Extend the vulnerability catalog with two new markdown-based reference guides that follow the existing skill-file structure.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 8 comments.

File	Description
strix/skills/vulnerabilities/ssti.md	Adds a new SSTI reference guide with engine-specific probes, exploitation paths, and validation steps.
strix/skills/vulnerabilities/header_injection.md	Adds a new header-injection reference guide covering response splitting, cache poisoning, forwarding-header abuse, and smuggling-related checks.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[0xallam]

@greptile

[0xallam]

@greptile

[0xallam]

@greptile

[0xallam]

@greptile