Version: 1.6 | Date: January 2026
Persistent memory for AI coding assistants through structured documentation.
Every time you start a new session with an AI coding assistant:
"Continue working on the authentication feature we discussed Friday."
"I don't have any context about previous conversations. Could you tell me about the authentication feature?"
You spend 15-30 minutes re-explaining your project, decisions, and progress. Every. Single. Time.
WITHOUT CxMS WITH CxMS
Session 1 Session 2 Session 3 Session 1 Session 2 Session 3
│ │ │ │ │ │
▼ ▼ ▼ ▼ ▼ ▼
┌───────┐ ┌───────┐ ┌───────┐ ┌───────┐ ┌───────┐ ┌───────┐
│ FULL │ │ FULL │ │ FULL │ │ FULL │ │ FULL │ │ FULL │
│CONTEXT│ │CONTEXT│ │CONTEXT│ │CONTEXT│ │CONTEXT│ │CONTEXT│
│ │ │ │ │ │ │ ↑ │ │ ↑ │ │ ↑ │
│ ▲ │ │ ▲ │ │ ▲ │ │ │ │ │ │ │ │ │ │
│ │ │ │ │ │ │ │ │ │ 5 sec │ │ 5 sec │ │ 5 sec │
│15-30 │ │15-30 │ │15-30 │ │ read │ │ read │ │ read │
│ min │ │ min │ │ min │ └───┬───┘ └───┬───┘ └───┬───┘
│rebuild│ │rebuild│ │rebuild│ │ │ │
└───────┘ └───────┘ └───────┘ ▼ ▼ ▼
┌─────────────────────────────┐
════════════════════ │ Session.md persists │
~60-90 min WASTED │ context between sessions │
per week └─────────────────────────────┘
~0 min wasted
CxMS gives AI assistants structured, persistent, user-controlled memory through markdown files.
| Without CxMS | With CxMS |
|---|---|
| Start from zero each time | Pick up exactly where you left off |
| Re-explain everything | AI reads context files automatically |
| Decisions forgotten | Decisions documented with rationale |
| Inconsistent approaches | Consistent patterns maintained |
CxMS is AI-agnostic. It works with any coding assistant that can read files:
| AI Assistant | Compatible | Notes |
|---|---|---|
| Claude Code | ✅ | Auto-reads CLAUDE.md |
| Gemini CLI | ✅ | Auto-reads GEMINI.md |
| GitHub Copilot | ✅ | Reads .github/copilot-instructions.md |
| Cursor | ✅ | Reads .cursorrules |
| Aider | ✅ | Reads CONVENTIONS.md via config |
| ChatGPT | ✅ | Upload or paste file contents |
| Any file-aware AI | ✅ | Same methodology applies |
Why "CLAUDE.md"? Claude Code auto-reads files named
CLAUDE.md. You can rename it toCOPILOT.md,AI_CONTEXT.md, or anything else - the methodology works the same.
View Live Dashboard - Anonymous community telemetry
See how the CxMS community uses the system: file adoption, deployment levels, feature requests, and more.
| Date | Enhancement | Description |
|---|---|---|
| 2026-01-27 | E20: Multi-Tool Export | Export profiles to Cursor, Copilot, Windsurf, Aider |
| 2026-01-27 | E19: Role-Based Profiles | web-developer, project-manager, data-engineer, devops, technical-writer |
| 2026-01-27 | E21: Context Lifecycle | Unified context management (consolidates E5+E6+E11) |
| 2026-01-27 | E8: Communication Efficiency | Concise AI output guidelines in templates |
| 2026-01-26 | Community Dashboard | Live telemetry visualization |
| 2026-01-25 | E17: Pre-Approved Operations | Skip permission prompts with Approvals file |
| 2026-01-25 | E16: Parent-Child Inheritance | Child projects reference parent CxMS conventions |
| 2026-01-25 | Context Monitoring | Self-monitor context % via statusline script |
See: Product Roadmap for all 21 enhancements (9 implemented)
| Level | Files | Best For |
|---|---|---|
| Lite | 3 | Quick experiments, solo projects |
| Standard | 9 | Team projects, ongoing development |
| Max | 17+ | Enterprise, complex multi-phase projects |
See: templates/DEPLOYMENT.md for full details
Lite (minimum viable):
your-project/
├── CLAUDE.md # From templates/core/CLAUDE.md.template
├── MyApp_Session.md # From templates/core/PROJECT_Session.md.template
└── MyApp_Tasks.md # From templates/core/PROJECT_Tasks.md.template
Adding to existing project? See CLAUDE.md.existing-project.template
One Prompt to Rule Them All (Max deployment):
Read MyApp_Startup.md and follow its instructions.
This single prompt reads all context files, applies approvals, checks context %, and provides a full session summary.
Or use level-appropriate prompts:
| Level | Startup Prompt |
|---|---|
| Lite | Read CLAUDE.md and MyApp_Session.md. Summarize current state and await instructions. |
| Standard | Read CLAUDE.md, MyApp_Session.md, and MyApp_Tasks.md. Summarize state, active tasks, and suggest next action. |
| Max | Read MyApp_Startup.md and follow its instructions. |
The AI will automatically check for CxMS updates and apply pre-approved permissions.
Update MyApp_Session.md with this session's work before we end.
See: SESSION_END_CHECKLIST.md for complete wrap-up workflow
That's it. You now have AI session continuity.
AI context is temporary; files are permanent.
Everything the AI needs to know must exist in files it can read.
This simple principle drives the entire system. Instead of relying on AI memory (which resets), we store context in markdown files that persist between sessions.
┌─────────────────┐
│ │
│ CLAUDE.md │◄──── AI reads FIRST
│ (The Brain) │ Project overview
│ │ Mandatory requirements
└────────┬────────┘
│
┌──────────────┼──────────────┐
│ │ │
▼ ▼ ▼
┌──────────────┐ ┌──────────────┐ ┌──────────────┐
│ │ │ │ │ │
│ Session.md │ │ Tasks.md │ │ Context.md │
│ (The Memory) │ │ (The Todo) │ │ (The Index) │
│ │ │ │ │ │
│ Current state│ │ What to do │ │ Where to look│
│ Last session │ │ Priorities │ │ Doc map │
└──────────────┘ └──────────────┘ └──────────────┘
│ │ │
└──────────────┼──────────────┘
▼
┌─────────────────────────────────────────────────┐
│ OPTIONAL LOGS │
├─────────────┬─────────────┬─────────────────────┤
│Activity_Log │Decision_Log │ Issue_Log │
│ What done │ Why decided │ What broke │
└─────────────┴─────────────┴─────────────────────┘
| Document | Purpose |
|---|---|
| CxMS_Introduction_and_Guide.md | START HERE - Full training guide, history, examples |
| CxMS_Practical_Implementation_Guide.md | Implementation details, templates, patterns |
| templates/DEPLOYMENT.md | Deployment levels (Lite/Standard/Max) |
| templates/MIGRATION.md | Upgrade guide - Fresh install or update existing |
See: templates/DEPLOYMENT.md for deployment levels | templates/MIGRATION.md for upgrades
| Template | Purpose |
|---|---|
CLAUDE.md.template |
Project overview with mandatory AI requirements |
CLAUDE.md.existing-project.template |
Adding CxMS to existing projects |
PROJECT_Startup.md.template |
One prompt to initialize full session context |
PROJECT_Approvals.md.template |
Pre-approved operations (skip permission prompts) |
PROJECT_Context.md.template |
Documentation index and reading order |
PROJECT_Session.md.template |
Current state - update every session |
PROJECT_Tasks.md.template |
Task tracker with status |
PROJECT_Prompt_History.md.template |
Audit trail of prompts and actions |
SESSION_START_PROMPTS.md |
Copy-paste prompts for session starts |
SESSION_END_CHECKLIST.md |
Session wrap-up workflow |
| Template | Purpose |
|---|---|
PROJECT_Activity_Log.md.template |
Record of significant changes |
PROJECT_Decision_Log.md.template |
Decisions with rationale |
PROJECT_Issue_Log.md.template |
Issue tracking |
PROJECT_Session_Summary.md.template |
One-line per session quick reference |
PROJECT_Deployment.md.template |
Deployment tracking across environments |
PROJECT_Compaction_Log.md.template |
Context loss tracking |
PROJECT_Performance_Log.md.template |
CxMS effectiveness metrics |
| Template | Purpose |
|---|---|
PROJECT_Plan.md.template |
Multi-phase project planning |
PROJECT_Inventory.md.template |
Component and endpoint inventory |
PROJECT_Strategy.md.template |
AI workflow strategy |
PROJECT_Exceptions.md.template |
Technical workarounds documentation |
PROJECT_Prompt_Library.md.template |
Curated prompts with improvement analysis |
| Template | Tool | Purpose |
|---|---|---|
GEMINI.md.template |
Gemini CLI | CxMS config for Gemini |
copilot-instructions.md.template |
GitHub Copilot | CxMS config for Copilot |
cursorrules.template |
Cursor | CxMS config for Cursor |
CONVENTIONS.md.template |
Aider | CxMS config for Aider |
MULTI-TOOL-DEPLOYMENT.md |
All | Multi-tool deployment guide |
| Profile | Tools Installed | Use Case |
|---|---|---|
web-developer |
Playwright, Prettier, ESLint | Frontend/backend web development |
project-manager |
GitHub CLI, documentation tools | Multi-agent coordination, planning |
data-engineer |
DuckDB, SQL tools | Data pipelines, ETL, analytics |
devops |
Docker, Terraform, kubectl | Infrastructure, CI/CD, deployment |
technical-writer |
Vale, markdownlint | Documentation, content creation |
Profile CLI:
node tools/cxms-profile.mjs list # List available profiles
node tools/cxms-profile.mjs install web-developer # Install profile globally
node tools/cxms-profile.mjs export web-developer --format all # Export to other AI tools┌─────────────────────────────────────────────────────────────┐
│ SESSION LIFECYCLE │
├─────────────────────────────────────────────────────────────┤
│ │
│ START │
│ │ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Read CLAUDE.md │ ← Project overview │
│ └────────┬────────┘ │
│ │ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Read Session.md │ ← Current state │
│ └────────┬────────┘ │
│ │ │
│ ▼ │
│ [ WORK ] │
│ │ │
│ ▼ │
│ ┌─────────────────┐ │
│ │Update Session.md│ ← CRITICAL! │
│ └────────┬────────┘ │
│ │ │
│ ▼ │
│ END │
│ │
└─────────────────────────────────────────────────────────────┘
Remember: Always update Session.md before ending a session or when context compacts!
- Developers using Claude Code, GitHub Copilot, ChatGPT, or similar AI assistants
- Teams wanting consistent AI-assisted development across members
- Anyone tired of re-explaining their project to AI every session
- A text editor
- Markdown files (
.md) - An AI coding assistant that can read files
No installation. No dependencies. No API keys. Just markdown files.
CxMS was developed and validated through real production use:
- Before: 15-30 minutes of context rebuilding per session
- After: Instant context restoration with a single prompt
- Context loss incidents: Reduced by ~80%
- Session productivity: Significantly improved
Ongoing Development:
- 21 enhancements documented, 9 implemented (E8, E9, E10, E13, E16, E17, E18, E19, E20)
- 5 role-based deployment profiles with CLI tool
- CxMS uses itself for development tracking (dogfooding)
- Active metrics collection for validation
See: LPR LandTools™ Case Study - Real-world implementation details
Context_Management_System/
├── README.md # This file
├── CLAUDE.md # Repository guidance
├── CxMS_Introduction_and_Guide.md # START HERE
├── CxMS_Practical_Implementation_Guide.md # Implementation details
├── CxMS_Product_Roadmap.md # 21 enhancements (9 implemented)
│
├── templates/ # 27+ templates organized by category
│ ├── DEPLOYMENT.md # Deployment guide (Lite/Standard/Max)
│ ├── MIGRATION.md # Fresh install & upgrade guide
│ ├── VERSIONS.md # Template version manifest
│ ├── core/ # Required templates (10)
│ │ ├── CLAUDE.md.template
│ │ ├── CLAUDE.md.existing-project.template
│ │ ├── PROJECT_Session.md.template
│ │ ├── SESSION_END_CHECKLIST.md
│ │ └── ...
│ ├── logs/ # Optional logging (7)
│ │ ├── PROJECT_Activity_Log.md.template
│ │ ├── PROJECT_Decision_Log.md.template
│ │ └── ...
│ ├── docs/ # Optional documentation (5)
│ │ ├── PROJECT_Plan.md.template
│ │ ├── PROJECT_Prompt_Library.md.template
│ │ └── ...
│ ├── multi-tool/ # Tool-specific configs (5)
│ │ ├── GEMINI.md.template
│ │ ├── copilot-instructions.md.template
│ │ ├── cursorrules.template
│ │ ├── CONVENTIONS.md.template
│ │ └── MULTI-TOOL-DEPLOYMENT.md
│ └── profiles/ # Role-based profiles (5)
│ ├── MANIFEST.json # Profile registry
│ ├── web-developer/ # Playwright, Prettier, ESLint
│ ├── project-manager/ # GitHub CLI, docs tools
│ ├── data-engineer/ # DuckDB, SQL tools
│ ├── devops/ # Docker, Terraform, kubectl
│ └── technical-writer/ # Vale, markdownlint
│
├── docs/ # GitHub Pages
│ └── dashboard/ # Community telemetry dashboard
│
├── tools/ # CxMS utilities
│ ├── cxms-report.mjs # Telemetry reporter (anonymous metrics)
│ ├── cxms-profile.mjs # Profile manager CLI (install, export)
│ ├── statusline-command.sh # Context monitoring (Mac/Linux)
│ └── statusline-command.ps1 # Context monitoring (Windows)
│
├── case-studies/ # Real-world implementations
│ └── LPR_LandTools_Case_Study.md
│
├── # CxMS Self-Tracking (dogfooding)
├── CxMS_Session.md # CxMS development state
├── CxMS_Tasks.md # CxMS task tracker
├── CxMS_Prompt_Library.md # Curated prompts with analysis
├── CxMS_Performance_Log.md # CxMS metrics
├── CxMS_Activity_Log.md # CxMS activities
├── CxMS_Decision_Log.md # CxMS design decisions
├── CxMS_Prompt_History.md # CxMS audit trail
│
└── Archive/ # Historical reference materials
Dogfooding: CxMS uses its own system to track its development - proving the methodology works.
Contributions welcome:
- Issues/PRs: Bug reports, feature ideas, documentation improvements
- Case Studies: Submit your CxMS experience to help validate and improve the system
MIT License - Feel free to use, modify, and distribute.
Developed through extensive real-world use with Claude Code (Anthropic). The methodology emerged from solving actual context management challenges in production software development.
Stop re-explaining your project. Start where you left off.
"AI context temporary is. Files permanent are. Remember this, you must."
— Master Yoda (CxMS Code Name since v1.3. Speak this way, required it is not. But fun, it is.)