Ctf Writeup

Generates a single standardized submission-style CTF writeup for competition handoff and organizer review. Use after solving a CTF challenge to document the...

Security Scan

The skill's behavior mostly matches a CTF writeup generator, but its runtime instructions ask the agent to scan files broadly and to include real flags unless asked to redact, which risks unintentional exposure of sensitive data.

Purpose & Capability

The name/description match the instructions: generating a submission-style writeup from challenge files. Requiring a filesystem-capable agent and use of bash/Python is coherent. Minor mismatch: allowed-tools includes WebFetch/WebSearch which are not referenced in the instructions and are unnecessary for offline writeup generation.

Instruction Scope

The SKILL.md explicitly instructs the agent to run broad filesystem scans (find . and recursive grep) and to collect 'solution artifacts' and flags from the current session and challenge files. Those commands can read any file under the working directory (or repository root), potentially exposing unrelated sensitive files. The guidance 'Redact the flag only if the user explicitly asks for redaction' means the agent will include real flags by default, raising a risk of leaking sensitive contents.

Install Mechanism

Instruction-only skill with no install spec and no code files — nothing is written to disk by an installer and no external packages are fetched. This is the lowest-risk install profile.

Credentials

No environment variables or credentials are requested (good). However, the instructions permit reading arbitrary files and session data; that filesystem access can expose secrets (API keys, private notes, other flags) even though no env vars are declared. The skill does not require or justify broad filesystem access beyond the challenge directory.

Persistence & Privilege

always:false and no requests to modify agent/system configuration. The skill does not request permanent or elevated privileges.

What to consider before installing

This skill is generally coherent for producing CTF writeups, but it instructs the agent to search the entire working directory and to include real flags by default. Before installing or running it: (1) run it only in an isolated/sanitized challenge directory or a copy of the challenge files (not your home directory or repo root); (2) explicitly tell the skill to redact flags or sensitive data if you don't want real secrets in the output; (3) consider removing or narrowing the find/grep lines to target a known challenge folder; (4) avoid giving it web/network tools unless you need them; and (5) if you're unsure, test the skill on a non-sensitive sample to observe what files it reads and what it outputs.

CTF Write-up Generator

Generate a standardized submission-style CTF writeup for a solved challenge.

Default behavior:

• During an active competition, optimize for speed, clarity, and reproducibility
• Keep writeups short enough that a teammate or organizer can validate the solve quickly
• Always produce a submission-style writeup
• Prefer one complete solve script from challenge data to final flag

Workflow

Step 1: Gather Information

Collect the following from the current session, challenge files, and user input:

1. Challenge metadata — name, CTF event, category, difficulty, points, flag format
2. Solution artifacts — exploit scripts, payloads, screenshots, command output
3. Timeline — key steps taken, dead ends, pivots

# Scan for exploit scripts and artifacts
find . -name '*.py' -o -name '*.sh' -o -name 'exploit*' -o -name 'solve*' | head -20
# Check for flags in output files
grep -rniE '(flag|ctf|eno|htb|pico)\{' . 2>/dev/null

Step 2: Generate Write-up

Write the writeup file as writeup.md (or writeup-<challenge-name>.md) using the submission template below.

---

Templates

Submission Format

---
title: "<Challenge Name>"
ctf: "<CTF Event Name>"
date: YYYY-MM-DD
category: web|pwn|crypto|reverse|forensics|osint|malware|misc
difficulty: easy|medium|hard
points: <number>
flag_format: "flag{...}"
author: "<your name or team>"
---

# <Challenge Name>

## Summary

<1-2 sentences: what the challenge was and the core technique. Keep it direct.>

## Solution

### Step 1: <Action>

<Explain the key observation in 3-8 short lines. Keep it direct.>

\`\`\`python
<one complete solving script from provided challenge data to printing the final flag>
\`\`\`

### Step 2: <Action> (optional)

<Only add this when a second short step genuinely helps readability, such as separating the core observation from final verification.>

### Step 3: <Action> (optional)

<Use only if the challenge really needs it. Keep the total number of steps small.>

## Flag

\`\`\`
flag{example_flag_here}
\`\`\`

Guidance:

• Prefer 1-3 short steps total
• Keep code to the smallest complete solving script
• Do not split "recover secret", "derive key", and "decrypt flag" into separate partial snippets
• The script should start from the challenge data and end by printing the flag
• Avoid long background sections
• Avoid dead ends unless they explain a key pivot
• Avoid multiple alternative solves; pick one clean path
• Redact the flag only if the user explicitly asks for redaction

---

Best Practices Checklist

Before finalizing the writeup, verify:

• Metadata complete — title, CTF, date, category, difficulty, points, author all filled
• Flag handling matches request — keep the real flag unless the user asked for redaction
• Reproducible steps — a reader can follow your writeup and reproduce the solution
• Code is runnable — exploit scripts include all imports, correct variable names, and comments
• No sensitive data — no real credentials, API keys, or private infrastructure details
• Length stays concise — the writeup is short enough for fast review
• Tools and versions noted — mention specific tool versions if behavior depends on them
• Proper attribution — credit teammates, referenced writeups, or tools that were essential
• Grammar and formatting — consistent heading levels, code blocks have language tags

Quality Guidelines

DO:

• Explain just enough for fast verification
• Include one complete solving path, not multiple alternative routes
• Include one complete script that goes all the way to the final flag
• Show actual output (truncated if very long) to prove the approach worked
• Tag code blocks with language (python, bash, sql, etc.)
• Keep the main path front-loaded so a reader can validate it quickly

DON'T:

• Copy-paste raw terminal dumps without explanation
• Paste several partial snippets that force the reader to reconstruct the final solve
• Leave placeholder text in the final writeup
• Include irrelevant tangents that don't contribute to the solution
• Assume the reader knows the specific challenge setup

Challenge

$ARGUMENTS