Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Bug Identification: Find the Bug Before You Fuzz

    1

    Static scanners hand you a thousand grep hits and a fuzzer burns CPU on code no attacker can reach.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Bug Identification: Find the Bug Before You Fuzz

    Bug Identification: Find the Bug Before You Fuzz

    Example session with this skill installed

    Audit NotesApp's C parser for memory bugs. I have the source. Where should I point a fuzzer first?

    • Read your context and instructions
    • Compiled the bug identification

    Bug identification pass: NotesApp parser

    Where to start (example run): One dangerous sink, reachable from untrusted file input, with a weak bounds check. Point the first fuzzing harness at parse_record().

    Attack surface

    ItemFinding
    Entry points3 file parsers take untrusted input
    Dangerous calls11 flagged (memcpy, strcpy, alloca, sprintf)
    Tainted to sink2 paths reach a sink unsanitized
    Highest prioritymemcpy with a length read from the file header

    The path that matters

    read_file() to parse_header() to record_len (untrusted) to parse_record() to memcpy(dst, src, record_len). There is no clamp on record_len between the header read and the copy.

    Next

    1. Build an AFL++ harness around parse_record() with a seed corpus of valid files.
    2. Add an ASAN build so overflows surface as clean reports.
    3. Patch-diff the last release: one bounds check was tightened, a likely N-day in older builds.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Static scanners hand you a thousand grep hits and a fuzzer burns CPU on code no attacker can reach. The hard part is deciding where the real bug is before you spend a week on it. This skill makes your agent map the surface, trace taint to a sink, and point the fuzzer at the paths that matter.

    What your agent does

    • Maps the attack surface: entry points, trust boundaries, privilege transitions and the parsers that take untrusted input.
    • Hunts dangerous functions: semantic search with CodeQL, Semgrep and Weggli, plus decompiler scripts like rhabdomancer and haruspex on binaries.
    • Traces taint to the sink: follows untrusted input through the code and reports the full data-flow path, not an isolated match.
    • Diffs patches for N-days: BinDiff, Ghidriff and Diaphora to isolate the modified bounds check in a recent patch.
    • Builds focused harnesses: AFL++, LibAFL or Syzkaller around the reachable sink, then triages the resulting crashes.

    Built-in guard rails

    • Plans, does not auto-run: it designs the audit and reads results you provide. You run the scanners and fuzzers.
    • No secrets needed: it works from source, binaries and scanner output.
    • Authorized targets only: it assumes you are permitted to test the code you point it at.

    What's inside

    One SKILL.md of about 7,300 words across 10 sections, covering source review, black-box testing, binary diffing and fuzzing setup, plus a LICENSE.txt.

    Works with Claude Code, Codex, Cursor, and any agent that loads SKILL.md.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.

    It guides the audit and triages findings; it does not run scanners or fuzzers for you. The tools it drives (CodeQL, AFL++, a decompiler) are yours to install. It assumes you are authorized to test the target.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 day ago

    Accounts
    Microsoft
    Software
    Claude Code
    orCodex
    orCursor
    Git
    Optional
    CodeQLSemgrepWeggli
    AFL++
    orLibAFL
    a decompiler for binary targets

    Permissions

    Terminal / Shell
    Network Access
    Write Files

    Allowed Hosts

    www.catalog.update.microsoft.com
    Listed1 day ago

    What's inside

    Frequently Asked Questions

    No. It plans the audit and reads results you provide. You run the scanners and fuzzers.

    No. It works from source, binaries and scanner output.

    An agent that loads SKILL.md. CodeQL, Semgrep, Weggli and a fuzzer if you want to run the pipeline it designs.

    No. It covers black-box and binary targets too: patch diffing, decompiler scripts and dynamic taint tracing.

    New in Code Quality & Review