$ loading_
Record and restore agent context with fan-out packs and signed receipts.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "io.github.NORTHTEKDevs/lossless-context-mcp" MCP server from askskill: Run: claude mcp add 'io-github-northtekdevs-lossless-context-mcp' -- npx -y lossless-context-mcp
Developers or researchers can use it to recover key task state after agent context compaction, reducing information loss after interruptions. It fits workflows that require continuous reasoning or multi-step execution.
When one task needs to be split into multiple subflows, it can package context into fan-out packs for different execution paths. This helps parallel work without repeatedly rebuilding shared background.
When agent output needs to be linked to repository changes, signed Git-bound receipts can preserve a record. This is useful for development workflows that care about traceability and trusted delivery.
It is a flight recorder for agent context. Based on the description, it supports restoring context after compaction, creating fan-out context packs, and producing signed receipts bound to Git.
The phrase 'signed git-bound receipts' suggests a clear focus on signed records and Git linkage. If you need traceability between agent execution and code changes, it may be helpful.
The provided material does not include installation steps, runtime requirements, or key information. Please see the source repository for prerequisites.
Compress context and persist checkpoints to cut AI agent token usage.
Monitor AI coding context usage and preserve state before compaction.
Monitor context pressure, compress history, and restore rules and files across sessions.
Manage agent context injection, retrieval, and layered storage for stable traceable workflows.
Build effective AI agents with MCP and simple workflow patterns.
Retrieve live task-specific context, constraints, and decisions for smarter agent execution.