Index of every page on this site: llms.txt
All skills
ljagiello avatar

/ctf-osint

@20d182d
by Lukasz Jagielloljagiello/ctf-skills3.8k stars
447
Run it yourself

Provides open source intelligence techniques for CTF challenges. Use when gathering information from public sources, social media, geolocation, DNS records, username enumeration, reverse image search, Google dorking, Wayback Machine, Tor relays, FEC filings, or identifying unknown data like hashes and coordinates.

Use this Skill: https://skilld.dev/gh/ljagiello/ctf-skills/ctf-osint

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

tools-mcp.md

≈1.2k tokens on demand. Your agent reads this file only when SKILL.md points to it.

CTF OSINT - MCP Server Integration

OSINT is browser-heavy: reverse-image search, map/Street View exploration, social-media profiles, archived pages, and JS-rendered sites that curl cannot read. Playwright MCP drives a real browser from the agent so it can navigate, screenshot, read rendered content, and step through interactive search flows.

Table of Contents

Full tool list and the general browser workflow live in ../ctf-web/tools-mcp.md.


Setup

Repository: https://github.com/microsoft/playwright-mcp (Apache-2.0, official Microsoft, tested v0.0.78+).

npx @playwright/mcp@latest        # most MCP clients auto-launch this
# The browser binary is a separate download; without it every call errors with
# 'Browser "chrome-for-testing" is not installed':
npx @playwright/mcp@latest install-browser chrome-for-testing

Snapshots, console logs and non-text response bodies are written to .playwright-mcp/ in the working directory and returned as file paths — read the file, and keep that directory out of version control.


Rendered-Page Reading & Screenshots

Modern OSINT targets render content with JavaScript — a raw fetch returns an empty shell. Playwright reads the page after JS runs.

browser_navigate "https://example.com/profile/target"
browser_snapshot                # rendered text/links/roles as an accessibility tree
browser_take_screenshot         # capture the visual for geolocation clues
browser_evaluate "() => document.body.innerText"   # full rendered text dump

Key insight: For any site that shows a blank page or a loading spinner under curl, Playwright MCP is the fix — browser_snapshot returns the content the way a human sees it, after client-side rendering, so names, timestamps, and hidden metadata become readable.


Reverse-Image & Map Search Flows

Reverse-image search and map exploration are multi-step interactive flows the agent can script end to end.

# Reverse image search
browser_navigate "https://images.google.com"
browser_click <camera / "Search by image" ref from snapshot>
browser_type <url field ref> "https://example.com/clue.jpg"
browser_press_key Enter
browser_snapshot                # read the visual-match results + candidate locations

# Map / Street View narrowing (cross-ref geolocation-and-media.md)
browser_navigate "https://www.google.com/maps/@<lat>,<lon>,3a,75y/data=..."
browser_take_screenshot         # compare panorama landmarks against the target photo

Key insight: Geolocation challenges are iterative — search, read candidates, refine, re-search. Playwright MCP lets the agent run that loop autonomously: submit the image, read the top matches from the snapshot, then pivot to maps to confirm landmarks. Pair with the manual techniques in geolocation-and-media.md.


Social Media & Archived Pages

browser_navigate "https://web.archive.org/web/*/example.com"
browser_snapshot                # list all snapshots; open one to read content deleted from the live site
browser_navigate "https://example.com/@target"
browser_evaluate "() => document.body.innerText"   # bio, post timestamps, geotags
browser_network_requests static=false filter="/api/"   # find the profile JSON endpoint
browser_network_request index=<n> part="response-body" # read the JSON itself

Key insight: a social site's profile JSON API is where the real intelligence is — IDs, exact timestamps, coordinates that the rendered HTML rounds off or omits. But browser_network_requests only lists the endpoints; the structured data comes from browser_network_request part="response-body". Take both steps, then pivot to username/metadata mining (see social-media.md and web-and-dns.md).


When to Use Playwright MCP vs curl/API

Situation Use
JS-rendered page, infinite scroll, or login wall Playwright MCP
Interactive reverse-image / map search flow Playwright MCP
Reading archived (Wayback) snapshots visually Playwright MCP
Bulk WHOIS / DNS / cert transparency lookups CLI/API (see web-and-dns.md)
Structured API with a documented endpoint requests/API — faster, no browser

Key insight: Use Playwright MCP when the intelligence is locked behind client-side rendering or an interactive UI. For structured lookups (DNS, WHOIS, cert transparency, platform APIs) the direct CLI/API path stays faster and more scriptable.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 alert26d5 checks · Risk SAFE
  • Gen Agent Trust Hub26d

    The skill provides a comprehensive toolkit and reference guide for OSINT (Open Source Intelligence) techniques in CTF challenges. While the tools and scripts are standard for their purpose, the skill inherently possesses a broad attack surface for indirect prompt injection because it processes data from numerous external, potentially attacker-controlled sources using high-privilege capabilities like shell execution and software installation.

  • Socket26d

    1 alert: gptAnomaly

  • Snyk26d

    Risk: MEDIUM · 1 issue

  • Runlayer8mo

    4/4 files flagged

  • ZeroLeaks7mo

    1 finding · Score: 86/100

Signed by skilld at 20d182d. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last week.

Activeupdated 4 weeks ago
metadata
{
  "user-invocable": "false",
  "author": "Lukasz Jagiello"
}
All 1 allowed tools
Bash Read Write Edit Glob Grep Task WebFetch WebSearch
Other metadata
compatibility
Requires filesystem-based agent (Claude Code or similar) with bash, Python 3, and internet access for OSINT lookups.

README badge

README badge for ljagiello/ctf-skills/ctf-osint