More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Binary Diff: Cross-Version Symbol Migration

    1

    Carry your reverse-engineering results across a version bump: match named functions to their new addresses, verify each match, and ship a reviewable mapping file instead of re-reversing from scratch.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Binary Diff: Cross-Version Symbol Migration

    Binary Diff: Cross-Version Symbol Migration

    Example session with this skill installed

    Our service binary moved from v1.4.2 to v1.4.3, a patch release. I have 64 named functions from v1.4.2 and I need them carried across. How much migrates cleanly, and what needs real work?

    • Read your context and instructions
    • Compiled the binary diff

    MIGRATION TRIAGE — v1.4.2 → v1.4.3

    Triage score: 8/8 (full migration). Export overlap 92%, string table mostly shared, binary size up 3%, changelog a short patch list. Same architecture and toolchain family confirmed.

    Anchors (22 of 64): 14 exported functions resolved by name, 6 pinned by unique error strings, 2 callbacks pinned by registration call shape. Anchor coverage 34% — enough to proceed.

    Comparison pass (remaining 42): 31 candidates scored 8+ with a clear gap over the runner-up. 7 scored 5–7 or within 2 points of a runner-up — these go to the review queue. 4 scored below 5 and stay unmatched.

    Verification: string check passed for all 31 high scorers; the caller check confirmed 29, and 2 failed it (callers do not line up), so they drop to medium and join the review queue. Of the 7 ambiguous candidates, 4 passed the caller check and rise to high; 3 stay medium pending a spot-check.

    Result: 56 names migrated — 48 high confidence, 8 medium with review notes and addresses listed for spot-check. 8 unmatched, classified: 3 rewritten (a logging function whose internals changed), 2 split (an init function divided into setup and validation), 1 gone (a removed legacy path), 2 new (a telemetry sender and its helper). The mapping file records old and new addresses, confidence, and evidence per name, and the report asks a human to spot-check the 8 medium names before anyone builds on them. Human time: triage, review queue, three spot-checks — under two hours for 64 functions.

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

    About this skill

    You reversed version 1. Version 2 is out, and reversing it from scratch wastes the work you already did — most functions moved but did not change. This skill migrates your knowledge: a 30-minute triage scores whether migration is worth it at all, an anchor set of exported functions and unique constants gives you ground truth, and a multi-feature fingerprint (strings, constants, call shape, block structure) scores the top three candidates per function so ambiguity gets reviewed instead of guessed.

    Verification is a ladder, cheapest first: string check, caller check, side-by-side decompilation, runtime confirmation. Every name lands in a mapping file with its confidence and evidence, and the leftovers get classified — rewritten, split or merged, gone, or new — so nobody hunts for a function that no longer exists. A worked example migrates 64 named functions across a patch release in under two hours of human time, with 8 medium-confidence names flagged for spot-check. For authorized analysis of your own software, lab samples, or systems you own.

    • Large rewrites defeat matching — the triage will tell you to migrate anchors only, and the rest needs fresh analysis.
    • Cross-architecture or cross-compiler matching is out of scope; fingerprints do not transfer there.
    • Inlined functions have no address to match and are recorded as merged, not migrated.
    • A medium-confidence match is a flagged candidate, not a verified name — it needs a human spot-check.
    • Packed binaries must be unpacked first; fingerprints of a packer describe the packer.

    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 any payment.

    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 3 days ago

    Permissions

    No special permissions declared or detected

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    A byte diff tells you what changed; this migrates what you knew. It anchors on functions you can identify with certainty, scores candidates on several features at once, and verifies each name before applying it — then records the whole thing as a mapping file a reviewer can check.

    Run the triage. A major rewrite scores low and the skill routes you to anchors-only migration plus fresh analysis, rather than forcing matches that would mislabel functions.

    High confidence means two independent checks passed (string plus caller, at minimum). For the functions whose names will support conclusions someone acts on, the skill still asks for a side-by-side spot-check.

    New in Security