Summary

FDEOps tackles AI coding agents' context loss by providing persistent, local "engagement memory." It targets Forward Deployed Engineers and similar roles, serving as a second brain and defensible record for client engagements, grounding AI interactions from initial contact to handoff.

How It Works

Core to FDEOps is a local "fieldbook" (.fde/ per client) storing engagement details in plain Markdown. Users interact naturally with AI agents (e.g., @fde), which manage memory writes, searches, and status updates via a local CLI. This ensures data privacy, ownership, and offline capability, allowing AI context access without sensitive data leaving the user's environment or consuming model tokens for basic memory operations. The fieldbook complements code memory for a holistic engagement record.

Quick Start & Requirements

Installation is primarily via npx skills add suboss87/fdeops or AI agent plugin marketplaces. A one-time workspace binding (npx fdeops resume --init <client_name>) links the current directory to a client's fieldbook. Verification uses npx fdeops resume. Requires Node.js >= 18 for CLI/adapters. Supports various AI agents (Cursor, Copilot, Gemini CLI, Claude Code) and local LLMs (Ollama, LM Studio). Manual/air-gapped installs are available. Links: docs/USAGE.md, docs/install.md.

Highlighted Details

• Engagement Lifecycle: Defines six phases: land, discover, plan, build, ship, close, guiding project progression.
• Structured Fieldbook: Uses plain Markdown files (e.g., context.md, brief.md, stakeholders.md) for a comprehensive, human-readable system of record.
• Local CLI Operations: Enables memory writes, search, and status checks without LLM tokens, enhancing efficiency.
• Data Sovereignty: Operates locally-first, ensuring user-owned, private data with no network calls for core memory and user confirmation for record updates.

Maintenance & Community

Built and maintained by Subash Natarajan. Feedback via GitHub Issues (CONTRIBUTING.md).

Licensing & Compatibility

MIT license permits broad use. Integrates with numerous AI agents and local LLMs via plain Markdown.

Limitations & Caveats

Emphasizes user confirmation for record updates, placing final judgment with the engineer. Tailored for client-site engagements; applicability may vary. npx fdeops scan provides heuristic leads, not definitive findings.