agentleFS
mtarcuremtarcure/claude-vibe-squad/.agents/skills/error-based-ssti/SKILL.md

error-based-ssti

Skill220 starsChanged 17 days agoMITView on GitHub

Use during an operator-authorized web assessment when a suspected backend template or expression sink is blind or suppresses output: force reflected evaluation exceptions or build an HTTP-status oracle to demonstrate scoped code execution or data extraction with a benign control.

The file

Files
SKILL.md

Error-Based SSTI ("Successful Errors")

Detect and exploit server-side template injection (and code injection) in blind / output-blocked environments by deliberately forcing runtime errors that reflect the result of evaluated code in the exception message. When even error text is suppressed, use a boolean error-based oracle (evaluate-then-conditionally-throw, e.g. division-by-zero on a true bit) to exfiltrate bit-by-bit via HTTP 500-vs-200 status differentials — zero dependency on rendered output.

Source: corpus A §I.1 — Korchagin, "Successful Errors" (PortSwigger Top 10 Web Hacking Techniques of 2025, #1). Impact class: RCE / private-data exposure (intrinsic). Governing method: Phase-3a hypothesis lane of systematic-attacking; leads only into the verification spine. Only fired against operator-authorized in-scope targets, honoring rate limits.

Method

  1. Identify template/expression sinks (user input flowing into Jinja2/Mako, Twig/Smarty, Spring EL, Handlebars, etc.) via source-map recovery + semgrep taint rules; note which reflect output vs. are blind.
  2. For blind sinks, send error-forcing payloads per engine: Python AttributeError/mro probes (().__class__.__mro__[1].__subclasses__()), Twig include(...)/not-found errors, Spring EL type-conversion faults — and read the reflected exception, not the page body.
  3. For error-suppressed sinks, build a boolean oracle: an expression that evaluates a target bit and conditionally triggers a fault (e.g. 1/0 on true), then read the HTTP status. Iterate to exfiltrate a secret / prove code execution one bit at a time.
  4. Escalate carefully to a contained impact proof (read a scoped file, echo a unique token via an interactsh-client OOB callback) sufficient to demonstrate RCE — never destructive, never out-of-scope. Encode the winning payloads as custom nuclei templates for regression.
  5. Negative control: a benign input over the same path produces the baseline status/response, proving the differential is attacker-caused.

Acceptance

  • Blind vs. reflected template sinks are enumerated; the exploited sink names its engine.
  • Code execution or data exfiltration is proven via the error/boolean oracle with an OOB or scoped-read impact proof and a benign negative control.
  • Payloads are captured as reusable nuclei templates; finding deduped against known SSTI CVEs for the stack before submission.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.