An MCP server for developers managing tool-output context consumption and session continuity in AI coding agents. Raw tool data stays in sandboxed execution, while SQLite supports session state. The project reports up to 98% context savings; verify that figure on your workload.
Project overview
The reported context reduction can be tested by comparing the same coding task with and without the tool.
Project type
MCP · AI Agent · AI Coding
Use cases
Coding & Development
Deployment
Refer to project documentation
License
License pending
Best for
Developers already using AI coding agents and encountering raw tool-output context consumption or lost continuity after compaction.
Teams evaluating local session processing without required cloud sync and with telemetry recorded as disabled by default.
CLI users willing to let progressive throttling constrain repetitive tool calls.
Key capabilities
Keep raw data in sandboxed execution and return execution stdout to the context, reducing the amount of tool-result data entering the conversation. This core capability is available.
Use Levenshtein distance to correct query spelling before searching again. This supporting capability is available.
Reduce results from repeated tool calls, then block continuing repetitions and direct the agent toward batch execution. This supporting capability is available.
Limitations and risks
The documented Elastic License 2.0 restriction prohibits offering the project as a hosted or managed service. Check the applicable terms before planning an external service or competing closed-source SaaS.
GPU requirements, minimum hardware and supported operating systems are unspecified. The facts also leave self-hosted and managed-service status unspecified; verify the target environment.
Sandbox tools return execution stdout rather than all raw tool data. Confirm suitability when the agent needs direct access to complete original results.
Getting started
The validated first-use path is installation through NPM or a marketplace. It does not establish one complete command sequence, platform requirement or configuration procedure for every environment.
After installation, use a representative coding task to inspect tool execution, returned stdout and session continuity through the documented CLI interaction.
Evidence and sources
GitHub project description: Context window optimization for AI coding agents. Sandboxes tool output (98% reduction), persists session memory, and enforces routing across 17 platforms via MCP + hooks.
README: Every MCP tool call dumps raw data into your context window. A Playwright snapshot costs 56 KB. Twenty GitHub issues cost 59 KB. One access log — 45 KB. After 30 minutes, 40% of y…
README: Context Mode is an MCP server that solves all four sides of this problem:
README: Platforms are grouped by install complexity. Hook-capable platforms get automatic routing enforcement. Non-hook platforms need a one-time routing file copy.
README: context-mode insight # opens the hosted Insight dashboard in browser