ProjectKeeper Context Handoff

    1

    Brief your coding agent with source-linked project context. Requires a separately installed local ProjectKeeper workbench and MCP connection.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+22 more

    ProjectKeeper Context Handoff

    Example session with this skill installed

    Use $projectkeeper-handoff to brief me on this project before continuing. Show the sources, freshness and next step; do not change files. Do not install software, change settings, or start an investigation. Use only this project and its connected ProjectKeeper server. Answer in English.

    • Read your context and instructions
    • Compiled the projectkeeper context handoff

    Recorded demo using a small controlled test project, not a customer project. Local filesystem link targets are omitted below; the full response text is retained.

    This project, Handoff verification, checks whether incoming agents can resume work using source-linked evidence.

    Goal and status: PLAN-1 calls for briefing the incoming agent with sources. Its sole work item,

    T-1 Brief incoming agent, remains recorded as

    In progress. Source: PLAN.md:5, ProjectKeeper src_04cbdd2aa02909f1.

    Recorded decision: Read context before changing files. Source: DECISIONS.md:3, src_e1f4e6e2a38abb42. ProjectKeeper does not trace these records to the owner’s words; owner acceptance is not established.

    Open issues: No breakpoints, send-backs, or flagged drift are recorded. One older indexed material has not been organized into fact records.

    Freshness is mixed. The pack says “as of October 13, 2026,” but its sources were read on

    October 19 at 08:50:25 UTC, at revision 3f01bd2ed7. It reports zero unprocessed changes while also describing the repository as “read but not organized yet.” Its saved “Keeper not connected” status is historical—the connected MCP server answered these reads. These inconsistencies limit confidence in the snapshot’s completeness.

    Next step: Review this briefing before selecting further work. The recorded plan specifies no implementation beyond the handoff; this briefing supplies that deliverable without changing its recorded status.

    Only ProjectKeeper’s read-only options, context, and source tools were used. No files, settings, or software were changed, and no investigation was started.

    PLAN.md

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

    About this skill

    Requires ProjectKeeper installed and running on your computer, an already organized project, and an agent with local stdio MCP support. This is a companion skill for that software; loading it from Agensi does not install the application.

    Pick up a long-running software project with its recorded direction in view. ProjectKeeper Context Handoff teaches your agent to read an existing ProjectKeeper project's context, resolve the relevant work and decisions, and check the original sources before continuing.

    Ask: “Use ProjectKeeper to brief me on this project before continuing. Show the sources, freshness and next step; do not change files.”

    The agent returns a concise briefing covering the current goal, relevant decisions, work status, open issues, the snapshot's freshness and a next step within your request. It distinguishes recorded plans and reports from verified behavior and owner acceptance.

    Requires ProjectKeeper 1.0.0 running on your computer, an already organized project, Node.js 24 or later, git 2.31 or later, and an agent that can use both skills and a local stdio MCP server. ProjectKeeper supports Windows and macOS; this skill's runtime checks and explicit skill invocation were verified on Windows with Node 24 and Codex CLI 0.162.0-alpha.17.2. macOS use and other hosts have not been tested for this skill. Loading the skill from Agensi does not install or start ProjectKeeper and does not give a cloud-only chat access to your computer.

    This skill reads existing project context through projectkeeper_options, projectkeeper_context and projectkeeper_get. It does not write your table, alter your project or launch a Keeper investigation. These three ProjectKeeper reading tools need no model call; your host agent's own model usage may still cost money. First-time project organization is separate, may use a provider key, can incur charges, and may send selected project/session material to the chosen model provider. Follow ProjectKeeper's setup instructions before first use.

    The ZIP contains a readable SKILL.md, a briefing example and MIT license, with no executable helpers, bundled server, keys or tracking. The skill does not direct project context to Agensi. Agensi's skill loading and usage records and your agent provider's data handling remain subject to their own policies.

    Source, installation and support: ProjectKeeper on GitHub and the 1.0.0 MCP guide. The companion skill source, example and MIT licence are publicly available from the author for local installation in other agents.

    How to use

    Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill passes our trust checks below.

    Security scanned

    Passed our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scanned

    Verified clean 1 day ago

    • Free to download with an account

    Requires

    ProjectKeeper 1.0.0 Local Workbench And Stdio MCP
    An already organized ProjectKeeper project
    Node.js 24 or later
    Git 2.31 or later
    Windows or macOS (this skill was tested on Windows)
    A Host Agent Supporting Skills And Local Stdio MCP
    Node.js
    macOS
    projectkeeper
    pk

    Listed1 day ago

    What's inside

    Frequently Asked Questions