Resume Check — AI Coding Handoffs

    1

    Pick up unfinished coding work with the right task, current project instructions, and a clear next step. Flag changed files and unresolved actions before continuing.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Resume Check — AI Coding Handoffs

    Example session with this skill installed

    Example request for this workflow: Resume the current local coding task using Resume Check. Read the required project instructions, check the saved handoff, and identify the next step. If a required instruction is missing or an action has an uncertain outcome, report it before continuing. Stay within the current authorized task and save an updated handoff when finished.

    • Read your context and instructions
    • Compiled the resume check

    Pick up the right task with a checked handoff

    Two recorded synthetic tests, summarized for readability. These are observed test outcomes, not customer results or a verbatim agent transcript.

    Test 1: The next agent completed the current task

    The project contained a current coding task and older publication notes. The fresh agent identified the current task and kept the work local.

    Handoff detailWhat the agent established
    Current taskReject blank greeting names in greet.py.
    ConstraintPreserve behavior for nonblank names.
    Next stepAdd blank-name validation and test the behavior.
    BoundaryLocal work only; no publication or deployment.

    Observed result: The agent added the validation, passed all 3 test methods, updated the work and progress records, and saved the final handoff. A separate verification reran the tests successfully.

    The final record marked the task complete, with no unresolved actions and no automatic next task. The checker returned REVIEW_REQUIRED with the reason WORK_COMPLETE: the finished task should not be started again.

    Test 2: A missing instruction was caught before work continued

    Resume Check flagged a missing required POLICY.md and an unresolved simulated publication. The agent stopped, left tracked project files unchanged, and did not retry the recorded operation.

    This was the intended result: the incomplete handoff was surfaced for review before the agent continued. BLOCKED described a missing project requirement.

    Next step in that case: Restore the authoritative instruction, review the changed state, and independently verify the uncertain action before considering a retry. The test used no real publication target.

    What this means for your next session

    The agent gets a concrete starting point: the current task, constraints, recorded progress, next step, and anything that needs review. The checker supports that review; it does not grant permission or guarantee that an agent will follow instructions.

    Resume Check flagged a missing required POLICY.md

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

    About this skill

    Give your next coding agent a checked starting point.

    Resume Check V2 — 0.2.0 beta

    When a coding session ends, the next agent needs to know what you were doing, what changed, and what still needs checking. Resume Check saves that handoff in your local Git project and checks it when you return.

    What you get

    • A startup brief with the current role, task, recorded progress, and next step.
    • Warnings about changed project files, missing instructions, stale checkpoints, and unresolved actions.
    • Closeout checks for the work and progress files you configure, so the next session has an explicit record of what was updated or left unchanged.

    Before you buy

    Requires an AI coding environment that can read local files and run shell commands, a local Git repository with an initial commit, Python 3.10+, and Git. Chat-only interfaces cannot run the checker. You must configure the project and save checkpoints deliberately.

    How you use it

    1. Connect your coding agent to Agensi and load Resume Check with get_skill. Load the current approved instructions when you use the skill in a new session.
    2. Your agent reads the delivered script, guides and examples in memory. It runs the unchanged Python script through python3 - with the script on standard input, without saving the package files.
    3. Configure your project profile, required instructions, current task and closeout files. Your project-specific records and checkpoints stay in your Git project.
    4. Save a checkpoint before ending a coding session. When you return, inspect the handoff and read the startup brief. Resolve flagged issues before continuing authorized work.

    Included

    10 functional text-file contents through Agensi MCP: skill instructions, a Python command-line tool, agent configuration, an operating guide, a format reference, and five JSON examples for the profile, current work, checkpoint, source review, and closeout. No icon is required to run the checker.

    Tested beta

    63 automated checks passed: 34 V1 regressions and 29 V2 checks. Two synthetic fresh-agent trials passed: one completed the current local task; the other stopped on a missing required instruction and preserved an unresolved action.

    Creator-access delivery and execution through Agensi MCP were also tested on 2026-10-12 using only Agensi-delivered materials. The unchanged script ran through standard input without being saved. The executed V2 configuration, checkpoint/review, file drift, missing required source, unresolved action and closeout scenarios passed.

    Tested on Linux with Python 3.12.14 and Git 2.51.1. Ordinary paid-buyer checkout, entitlement and delivery remain unverified. No customer pilot or cross-platform claim.

    Limits

    Checkpoints must be saved and required sources listed explicitly. The tool cannot recover unsaved chats, extend a model's context window, automatically import an existing YAML workflow, execute saved instructions, retry external actions, or sync to the cloud.

    Source-review receipts are agent self-reports. Closeout checks compare file changes, not semantic correctness. Passing checks do not prove claims are true or grant permission to act. V1 state can be upgraded explicitly with a backup while preserving unresolved journal entries.

    Privacy and purchase

    The Python tool runs locally with no built-in network calls or telemetry. Your AI coding environment may have its own data-handling settings.

    $19 one-time purchase. Access is managed through Agensi. No hosted service or subscription.

    Beta 0.2.0 with manual checkpoints and explicit project setup. Profiles and current-work records use JSON; existing YAML workflows are not automatically imported. Sources must be configured explicitly. Review receipts are self-reports, not proof of reading or comprehension. Closeout checks bytes, not semantic correctness. No context-window extension, unsaved-history recovery, automatic source discovery, action replay, cloud sync, approval verification, external outcome verification, or guarantee against duplicate operations. Unrecorded actions and unselected ignored files may be outside coverage. Tested on Linux with Python 3.12.14 and Git 2.51.1; macOS, Windows, other Python versions, and named external agent clients are unverified. Creator-access Agensi MCP delivery and standard-input execution were tested; ordinary paid-buyer checkout, entitlement and delivery remain unverified. No customer pilot.

    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 1 month ago

    Requires

    Python
    macOS

    Permissions

    Terminal / Shell
    Read Files
    Write Files

    File Scopes

    .resume-check/**

    Reads local Git state, project files, selected evidence, configured required sources, and included skill files. The Python tool writes state, checkpoints, review/closeout receipts and the action journal only under .resume-check/ in the selected repository. The agent deliberately creates or updates project-owned profile, current-work and progress files within current user authorization. Uses Python and Git through the local shell. No built-in network calls, telemetry, browser access, or external action execution. File scope depends on the configured repository and sources.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It means the configured comparisons and a matching source-review self-report are present. The agent must still read current instructions, reconcile prose conflicts, evaluate evidence and verify current authorization. The tool cannot prove reading, understanding, truthful claims or valid approvals.

    A local Git repository with an initial commit, Python 3.10+, Git and shell access, and an AI coding environment that can read files and run commands. Validated on Linux with Python 3.12.14 and Git 2.51.1. Other hosts are unverified.

    Agensi MCP delivers the contents of 10 functional text files: skill instructions, the Python checker, agent configuration, two guides and five JSON examples. Earlier package inventories may also list an optional icon.svg whose contents are not delivered through MCP; the checker does not need it. The agent uses the package in memory and runs the script through standard input. Creator-access delivery and execution have been tested; ordinary paid-buyer checkout, entitlement and delivery remain unverified.

    No. Save checkpoints after milestones and before a planned transition. It cannot save after a hard cutoff or recover work that was never recorded. A chat-only interface cannot run the checker.

    The checker flags Git or evidence drift, changed or missing configured instructions, and mismatches between the checkpoint and current role or task. Review the findings and deliberately save a reconciled checkpoint. Missing required sources block continuation checks. It does not automatically replace your checkpoint or retry unresolved actions.

    The Python tool runs locally and has no built-in network calls or telemetry. Your AI coding environment may have its own data-handling settings; review those separately.

    Connect a coding agent with local file and shell access to Agensi, then load Resume Check with get_skill. Read the delivered guides and examples in memory, and run the complete unchanged script through python3 - with the script on standard input. Do not save package files. Create project-specific profile, current-task and progress records using actual project facts; configure required sources and closeout files. Save a checkpoint before a session ends and inspect it when resuming. Keep the project and .resume-check state available for the next session. Existing Markdown or YAML instructions can be required reading, but V2 does not automatically import their structure or resolve conflicts. V1 state can be upgraded explicitly with a backup while preserving the action journal.

    New in Code Quality & Review