IDE Memory MCP
Persistent, cross-IDE project memory for AI agents via MCP.
Free
Works with the AI Agents you already use
IDE Memory MCP
Example session with this skill installed
Initialize memory for this project and tell me what we were working on last.
- Read your context and instructions
- Compiled the ide memory mcp
I have initialized the project memory and retrieved the active context. It looks like you were in the middle of refactoring the authentication middleware.
Project Memory Summary
| Section | Size | Last Updated | Status |
|---|---|---|---|
| overview | 1.2kb | 2 days ago | Current |
| decisions | 0.8kb | 5 days ago | Current |
| active_context | 2.4kb | 1 hour ago | Active |
| progress | 1.5kb | 1 hour ago | Active |
Current Work State (active_context)
The primary focus is replacing the legacy JWT verification in /src/middleware/auth.ts with the new @auth/core implementation.
- Completed: Installed dependencies, drafted new schema.
- In Progress: Updating
validateSessionto handle the new token format. - Blockers: Waiting for confirmation on the redirect URL for the OAuth callback.
Recent Decisions
- 2023-10-24: Switched to
@auth/coreto simplify multi-provider support. - 2023-10-20: Chose Redis for session caching to handle horizontal scaling.
Next steps
- Finish the
validateSessionlogic inauth.ts. - Run the existing test suite to check for regressions in the login flow.
- Update the
progresssection once the middleware tests pass.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop losing context every time you start a new chat or switch IDEs. This skill interfaces with the ide-memory MCP server to store architectural decisions, project progress, and active work states in local Markdown files, ensuring your agent remembers the project across Claude Code, Cursor, VS Code, and Windsurf.
What it does
- Persistent Context maintains a project-specific memory that survives chat resets and IDE restarts.
- Cross-IDE Sync shares the same memory files across any MCP-compatible environment including Cursor and Claude Desktop.
- Automated Bootstrapping scans READMEs and config files to initialize memory for new projects.
- Selective Retrieval loads only relevant sections like
decisionsoractive_contextto save tokens. - Historical Snapshots recovers previous versions of memory sections using built-in versioning.
How it works
- Initialize the project by calling
init_projectat the start of a session to detect the root path and git remote. - Review the summary table to identify stale sections or high-level project status without loading full content.
- Read specific sections or query the memory for keywords to pull relevant context into the current chat.
- Update memory after significant changes or decisions to ensure the next session starts with an accurate state.
Frameworks & tools
Requires the ide-memory MCP server installed locally. Works with any stack including Node.js, Python, Rust, and Go by parsing standard project manifests.
Why this beats prompting it yourself
Manual context setting is repetitive and error-prone. This skill automates the retrieval and storage of technical debt, rationale, and progress, preventing "hallucinated" project states that often occur in long-running development threads.
Use cases
- Resuming Work after a break by reloading the exact
active_contextand TODO list from the previous day. - Onboarding Agents to a new codebase by having the tool scan and summarize the architecture into memory.
- Tracking Decisions to ensure the agent remembers why a specific library or pattern was chosen months ago.
- Cross-Tool Handoff moving from a terminal-based agent to a GUI IDE while keeping the current task state intact.
Known limitations
Requires local installation of the ide-memory MCP server. Does not store secrets or credentials. Performance may degrade if individual memory sections exceed 10k characters without pruning.
Requires ide-memory MCP server. No support for storing secrets or .env values. Manual pruning required for sections exceeding 10k characters.
How to use
- 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
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 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
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 16 days ago
- Free with an Agensi account
Requires