Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Intent Replay

    1

    A project continuity skill that keeps intent, decisions, active tasks, verification state, and handoffs synchronized across different agents and LLMs so work can continue without losing important context.

    Free

    1 use5.0 (1 review)
    Intent Replay

    Intent Replay

    Example session with this skill installed

    Initialize Intent Replay in this repo and create a new task for adding Stripe integration.

    • Read your context and instructions
    • Compiled the intent replay

    I have bootstrapped the .project directory and established the tracking files. I also created a new active task for the Stripe integration.

    Project Context Initialized

    FileStatusDescription
    AGENTS.mdCreatedStanding orders for agent continuity.
    .project/INTENT.mdCreatedCore project goals and constraints.
    .project/NOW.mdCreatedCurrent snapshot and next steps.
    .project/DECISIONS.mdCreatedLog of architectural and logic choices.
    .project/tasks/active/001-stripe.mdCreatedTask tracking for Stripe integration.

    Next steps

    1. Define the specific Stripe products and checkout flow in .project/INTENT.md.
    2. Review the acceptance criteria in .project/tasks/active/001-stripe.md.
    3. Begin implementation of the Stripe webhook handler.

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

    About this skill

    A project continuity skill that preserves intent, decisions, active work, verification state, and handoffs across different agents and LLMs. It keeps important project context inside the repository so work can continue across tools and sessions without relying on a previous agent's conversation history.

    What It Does

    • Preserves project intent, scope, constraints, and desired outcomes.
    • Tracks the current objective, active tasks, blockers, and next action.
    • Records important decisions that future agents need to follow.
    • Separates observed, reported, planned, and unverified project state.
    • Maintains active and completed tasks with verification evidence.
    • Keeps project context synchronized across different agents and LLMs.
    • Provides a durable handoff structure without storing entire conversations.

    How It Works

    1. Checks for the Intent Replay structure before substantial project work begins.
    2. Bootstraps the required .project/ files and AGENTS.md integration when missing.
    3. Reads the current intent, handoff state, active task, and relevant decisions.
    4. Reconciles the recorded context with the actual repository and available Git state.
    5. Updates project context only when meaningful goals, decisions, tasks, or verification state change.
    6. Archives completed and verified tasks while preserving their history.
    7. Refreshes the current handoff so the next agent can safely continue the work.

    Package Contents

    The skill package contains the main SKILL.md file together with an assets/ directory containing the initial Intent Replay templates and a references/ directory containing the bootstrap rules, state contract, task lifecycle, continuity checks, and handoff guidance referenced by the skill. Keep the folder structure intact so references from SKILL.md continue to work correctly.

    Requirements

    • An AI agent or environment that can read and edit project files.
    • Access to the repository or project workspace being worked on.
    • Read and write access to AGENTS.md and the .project/ directory.
    • Access to relevant project files improves reconciliation and handoff accuracy.

    Frameworks & Tools

    Framework-agnostic and model-agnostic. Intent Replay works through Markdown files stored inside the project rather than depending on a particular LLM, coding agent, IDE, programming language, or application framework.

    It is especially useful for developers who move between multiple AI coding tools, agents, or sessions during the lifetime of a project.

    Why Use This Instead of Prompting It Yourself

    Normal AI development sessions keep important project context inside the current conversation. When you switch models, agents, or sessions, that context can disappear or become inconsistent. Intent Replay moves the important parts into the repository itself so the next agent can reconstruct the project's intent, decisions, active work, verification state, and next action before continuing.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Manual trigger required before commits. Does not automate Git actions. Requires filesystem write access. Limited to project-level context, not suited for global system configuration.

    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

    5.0
    1 review
    5
    1
    4
    0
    3
    0
    2
    0
    1
    0

    This skill has been used 1 time.

    Trust & safety

    Security scan

    Security scan passed 25 days ago

    • Free with an Agensi account

    Permissions

    Read Files
    Write Files
    Listed25 days ago

    What's inside

    Frequently Asked Questions

    Intent Replay eliminates 'context drift' by storing your project's goals, decisions, and task states in a standardized directory within your repo, ensuring no information is lost when you switch models or clear chat histories.

    This skill is framework-agnostic and works with any AI agent or IDE capable of file-system access (such as Claude Code, Cursor, or Windsurf), as it operates entirely through Markdown files in your repository.

    The skill functions via a `.project/` directory that stores your intent, task logs, and decision history; it initializes these files from templates and updates them dynamically as the agent works.

    You receive the core logic and standardized templates for the `.project/` directory structure, including the `INTENT.md`, `AGENTS.md`, and `NOW.md` handoff protocols.

    Intent Replay generates the necessary documentation and state files, but it does not automatically perform Git commands; you should trigger an update and include the generated files in your regular commit workflow.

    New in Agents & Orchestration

    Free