Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Agent Memory Access

    1

    Automated historical context retrieval and synthesis from past agent sessions.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Agent Memory Access

    Agent Memory Access

    Example session with this skill installed

    I'm ready to continue work on the authentication module. What were the specific edge cases we decided to handle for the JWT refresh flow last Tuesday?

    • Read your context and instructions
    • Compiled the agent memory access

    I have retrieved the relevant notes from your session on Tuesday regarding the JWT refresh implementation.

    Historical Context: JWT Refresh Decisions

    CategoryDecision / Edge Case
    Token ReuseImplement a 30-second grace period for old refresh tokens to handle concurrent network requests.
    RevocationImmediate invalidation of the entire refresh chain if a reused token is detected outside the grace window.
    ExpirySet sliding window expiration: 7 days max age, 24 hours idle timeout.
    SecurityStore jti in a Redis-backed allow-list rather than a block-list for faster validation.
    Client HandlingReturn a specific 401.13 sub-status code to trigger a full re-login on the frontend.

    Next steps

    1. Review the current auth-service/middleware.ts to see if the jti allow-list logic is already present.
    2. Implement the 30-second grace period logic in the token exchange function.
    3. Update the frontend interceptor to handle the 401.13 sub-status code.

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

    About this skill

    Context switching between projects often leads to repetitive onboarding of your own AI. This skill enables agents to automatically retrieve relevant definitions, past decisions, and preferences from previous sessions without manual copy-pasting.

    What it does:

    • Automated search scans local session history for keywords matching the current prompt.
    • Privacy redaction filters out common secret formats before data is processed by subagents.
    • Smart synthesis uses a subagent to summarize historical excerpts into concise notes.
    • Resource bounding enforces strict limits on file counts, byte sizes, and search timeouts.
    • Context injection provides the main agent with relevant history without treating it as new instructions.

    How it works:

    1. The extension triggers automatically on every prompt before the main agent processes the request.
    2. It searches configured JSONL session files, ranking matches by term relevance.
    3. Historical data is sent to a memory subagent for synthesis into high-density context.
    4. The synthesized notes are injected into the current prompt as reference material.

    Frameworks & tools: This skill is built for agents utilizing local session logging. It requires a TypeScript runtime to execute the extensions/agent-memory-access.ts component and supports various LLM providers for synthesis.

    Why this beats prompting it yourself: Manual context management is slow and prone to human error. This skill automates the retrieval process at the system level, ensuring your agent remembers specific technical decisions or project constraints you established weeks ago without you having to find the old logs.

    Use cases:

    • Resuming complex coding tasks across different days or sessions.
    • Recalling specific architectural decisions made in previous project phases.
    • Maintaining consistent formatting preferences across disparate chat threads.
    • Referencing internal definitions established in past research sessions.

    Known limitations: Memory is treated as untrusted reference data only. Historical instructions are ignored to prevent prompt injection. Performance depends on local file indexing speeds.

    Historical data is strictly reference only and cannot override current instructions. Search performance is bound by local file system I/O and configured timeouts.

    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

    GitHub
    Npmjs

    Permissions

    Terminal / Shell
    Write Files
    Network Access
    Environment Variables

    Allowed Hosts

    registry.npmjs.org
    github.com
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    This skill eliminates 'context amnesia' by automatically scanning your local session logs for relevant past decisions and definitions, injecting them into your current prompt so you don't have to manually re-explain your project constraints.

    The extension is compatible with any agent framework that supports TypeScript execution and maintains local JSONL session logs, including most popular open-source autonomous agent stacks.

    It includes a privacy redaction layer that automatically identifies and masks common sensitive formats (like API keys or secret tokens) within your history before the data is synthesized by a subagent.

    Yes, you can configure resource bounding to set strict limits on file count, byte size, and search timeouts to ensure the memory retrieval process doesn't slow down your agent's response time.

    You receive the complete `extensions/agent-memory-access.ts` source code, configuration documentation, and lifetime access to version updates as the memory synthesis logic is refined.

    The skill is designed to treat historical data as 'read-only' reference material; it explicitly ignores old instructions found in logs to prevent 'ghost' commands from interfering with your current task.

    New in LLM Ops