Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    MCP Migration Compatibility Checker (2025→2026)

    1

    Compare fixed MCP contract snapshots for revisions 2025-11-25 and 2026-07-28 and get local PASS, FAIL or NEED_USER evidence.

    Free

    2 usesSecurity scanned
    MCP Migration Compatibility Checker (2025→2026)

    MCP Migration Compatibility Checker (2025→2026)

    Example session with this skill installed

    Verify the migration compatibility between ./snapshots/old_contract.json and ./snapshots/new_contract.json. Output the results to the ./audit-logs directory.

    • Read your context and instructions
    • Compiled the mcp migration compatibility

    Audit complete. Status: FAIL. Found 2 BREAKING changes in tool schema required fields.

    Artifacts generated

    • compatibility.json (authoritative)
    • compatibility.md (summary)
    • evidence-receipt.json
    compatibility.json

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

    About this skill

    A contract migration can appear compatible until a renamed or unsupported field reaches release review. Compare one local 2025-11-25 baseline snapshot with one 2026-07-28 candidate snapshot and receive a deterministic compatibility decision plus five reviewable evidence files. This checker supports only that fixed revision boundary and does not connect to a live MCP server.

    The problem

    Manually verifying Model Context Protocol (MCP) contract compatibility during the 2025-to-2026 era migration is error-prone. Developers lack a deterministic way to audit breaking changes between local protocol snapshots without risking subjective LLM hallucinations.

    What it does

    • Performs deterministic comparison between 2025-11-25 and 2026-07-28 MCP snapshots.
    • Generates five auditable artifacts including compatibility reports and evidence receipts.
    • Enforces a fail-closed gate resulting in PASS, FAIL, or NEED_USER based on migration rules.
    • Validates JSON depth, value-node counts, and type-aware canonical semantics.
    • Detects breaking changes in tools, prompts, and schema annotations via local execution.

    Frameworks & tools

    Python, Model Context Protocol (MCP), JSON Schema, SHA-256 integrity hashing.

    Why this beats prompting it yourself

    Prompting an LLM to "check for compatibility" leads to missed edge cases in JSON schema evolution. This tool uses a rigid, script-based collector that prevents subjective upgrades of FAIL or NEED_USER statuses, ensuring 100% auditability.

    Use cases

    • Verifying tool and prompt compatibility before upgrading MCP server implementations.
    • Generating compliance evidence for internal architectural reviews.
    • Debugging malformed JSON schema transitions between protocol revisions.

    Known limitations

    Only supports the 2025-11-25 to 2026-07-28 revision boundary. Rejects network shares, remote volumes, and Windows UNC paths. Does not perform automatic fixes or tool execution.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.

    Supports only the 2025-11-25 to 2026-07-28 boundary. Requires local JSON snapshots. No remote MCP access, network paths, or automatic code fixes.

    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

    2 uses

    Times used to date

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 months ago

    • Free with an Agensi account

    Software
    Finding Id MCPTOOL 003 MCPSCHEMA 001 MCPPROMPT 003 MCPTOOL 006 MCPIDENTITY 001 MCPINPUT 001 MCP

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    File Scopes

    snapshots/*.json
    audit-logs/**

    Runs a bundled local Python CLI to compare MCP JSON snapshots from the local snapshots/ workspace and write bounded compatibility evidence to audit-logs/. Terminal permission is used only to launch the bundled CLI. No network, browser, environment-variable, credential, or remote MCP access.

    Listed2 months ago

    What's inside

    Frequently Asked Questions

    This skill provides a deterministic, script-based audit that compares specific protocol snapshots (2025-11-25 vs. 2026-07-28). Unlike LLMs, which may hallucinate or miss subtle JSON schema shifts, this tool uses type-aware canonical semantics to ensure 100% auditable migration proof.

    The skill generates five auditable artifacts, including detailed compatibility reports and SHA-256 integrity-hashed evidence receipts. These files serve as formal proof of compliance for internal architectural reviews and debugging.

    This tool is specifically scoped to the transition boundary between the 2025-11-25 and 2026-07-28 MCP revisions. It focuses on validating JSON depth, tool schemas, and prompt annotations rather than performing general code refactoring.

    The tool operates via a local execution environment to ensure data privacy and integrity. Please note that it does not support network shares, remote volumes, or Windows UNC paths; files must be stored on a local directory for the audit to run.

    No, this is a diagnostic and verification tool. It identifies breaking changes and enforces a fail-closed gate (PASS, FAIL, or NEED_USER), but it will not modify your source code or perform automatic fixes.

    New in LLM Ops

    Free