What is DevArk?

DevArk-cli is an open-source command-line tool that analyzes your Claude Code sessions to extract productivity insights and generate reports. All analyses can run completely locally on your machine using Claude Code's capabilities. It also encompasses a statusline that help you get more from your sessions.

Getting Started fast

Just run:
``bash
npx devark-cli@latest
`

Main features

$3

New: Be prepared in 2-3 minutes for your daily standup meeting!

$3

Generate comprehensive productivity reports using Claude Code's sub-agents to analyze your sessions in parallel. No data leaves your machine.
See example of locally generated report here: https://devark.ai/example-daily-report

$3

Strategic product advisor in Claude Code. Your prompts are analyzed locally (via CC) to provide actionable guidance that pushes you to ship faster. Feedback appears in your Claude Code status line with concrete next steps.

Updates

$3

#### Today's Standup ✨
Get a personalized standup summary of your recent claude coding activity right from your terminal. Runs 100% locally using your own Claude Code installation
The standup feature analyzes your recent sessions and generates a concise summary of:
* What you worked on
* Key accomplishments
* Next steps

#### Improved Onboarding Experience 🚀
Getting started with devark is now smoother than ever:
Better first-run setup flow
Helpful prompts to guide you through configuration
Improved error messages when things go wrong

#### Improved Time Calculation ⏱️
More accurate session duration tracking with better handling of:
Massive sessions: Sessions with 10,000+ messages no longer cause performance issues
Smart truncation: Large sessions are intelligently truncated while preserving time data
Consistent timestamps: Better parsing and validation of session timing information
Edge case handling: Fixed issues with sessions that had missing or invalid timestamps

💬 More about Claude Status Line - Strategic Co-pilot / Advisor (Local)

🔒 Installation Safety: Automatically backs up your existing status line configuration. Uninstalling instantly restores your original setup - zero risk to your Claude Code configuration.

$3

- 🚀 Ship Faster: Get pushed to deliver
- 🎯 Stay Focused: Remembers your original mission and keeps you on track and forward ⏩︎
- ⚡ Concrete Actions: Specific next steps like:
- 📈 Strategic Thinking: Considers edge cases, user experience, and scaling at the right time
- 📊 Token Usage Display: Integrates with ccusage to show real-time token metrics
- ↩️ Easy Restore: Your original status line is backed up and can be restored anytime

Architecture

$3

1. Intercepts prompts submitted in Claude Code
1. Analyzes via the local Claude Code latest prompt with relevant session context.
4. Provides strategic guidance and pushes you to ship
5. Displays actionable feedback in the Claude Code status line

In more details:
`mermaid
flowchart LR
subgraph Input ["✳️ In Claude Code"]
User([👤 You type a
prompt]) --> Submit[Press Enter to
submit prompt]
Submit --> Hook[🪝 UserPromptSubmit
hook triggers]
end

subgraph Analysis ["🧠 Local Prompt Analysis via Claude Code SDK"]
Hook --> CLI[DevArk CLI
receives prompt]
CLI --> Check{Is new chat?}
Check -->|No| Context[📝 Include previous
conversation]
Check -->|Yes| Direct[💭 Analyze prompt
standalone]

subgraph Personality ["🎭 Infuse Coach Personality"]
Gordon[🧑‍🍳 Gordon
Tough love]
Vibe[💜 DevArk
Encouraging]
Custom[✨ Custom
Your style]
end

Context --> SDK[Claude SDK
analyzes prompt quality]
Direct --> SDK
Personality -.-> SDK
SDK --> Score[📊 Generate score
& suggestion]
end

subgraph Display ["💬 Status Line Feedback"]
Score --> Save[💾 Save to
~/.devark/analysis]
Save --> Status[Status bar
reads result]
Status --> Show[🟢 85/100
✨ Gordon says:
Add more context chef!]
end

Show --> Improve([📈 Better prompts
Better results])
`

$3

- Gordon - Sharp, pushy, business-focused. Creates urgency: "Ship by FRIDAY or you're fired!"
- DevArk - Supportive but pushy senior dev. Helps you ship: "MVP checklist: Auth works ✓ | Ship it!"
- Custom - User-defined personality with strategic focus

$3

$3

1. Run npx devark-cli
2. Select "Configure prompt coach status line"
3. Choose coach personality
4. (Optional) Enable ccusage token metrics display
5. Prompts will be analyzed locally in Claude Code

Note: Installation automatically backs up your current status line (if any). When you uninstall devark, your original configuration is instantly restored.

Local Report Generation Works

Generate comprehensive productivity reports using Claude Code's sub-agents to analyze your sessions in parallel. No data leaves your machine.
- Select timefrema
- Select projects

`mermaid
flowchart TD
Start([📝 Claude Code Sessions]) --> Select[devark-cli select time frame and projects]
Select --> Extract[Extracts & prepares session data]
Extract --> Launch[Launches Claude with instructions]
Launch --> Parallel{Parallel sub-agents
session analysis}
Parallel --> Gather[Gathers results &
Generates report]
Gather --> Output[📊 HTML Report in current folder]
style Start fill:#e1f5fe
style Output fill:#d4edda
`


Example of Local Report Output

☁️ Cloud Sync (Optional)

Optionally sync your sanitized session data to the devark dashboard for advanced analytics and tracking over time.

$3

- 🎁 Free Forever: Up to 1,000 session analyses per month
- 📈 Track Over Time: Monitor your prompt quality and productivity trends
- 🔄 Auto-sync: Configure hooks for automatic background sync
- 🔒 Privacy First: All code removed before upload, only patterns synced
- 📊 Web Dashboard: Learn and improve your AI coding sessions.

$3

`mermaid
flowchart TD
subgraph Local ["🏠 Your Machine"]
Sessions[Claude Code Sessions] --> Select[Select sessions]
Select --> Privacy[🔒 Privacy Layer
Removes code & secrets
Keeps only patterns]
end

subgraph Cloud ["☁️ DevArk Cloud"]
Privacy --> Upload[Upload patterns]
Upload --> Verify[Server verification]
Verify --> Analysis[AI analysis]
Analysis --> Dashboard[📊 Web Dashboard]
end

subgraph Features ["Dashboard Features"]
Dashboard --> Track[📈 Track prompt improvement over time]
Dashboard --> Metrics[⚡ Deeper productivity insights]
Dashboard --> Streaks[🔥 Peak times/Low times]
Dashboard --> Prompt[💬 User Prompt Analysis]
end

style Local fill:#e1f5fe
style Cloud fill:#f3e5f5
style Features fill:#d4edda
`


Auto-Sync

$3

Claude Code Hooks allow DevArk to automatically sync your Claude Code sessions.

What are Claude Code Hooks?

• Small commands that run at specific moments in Claude Code

• They work silently in the background (you won't notice them)

Which hooks do we use?

📍 SessionStart - Syncs previous sessions when you start/resume work
(Triggers: startup, resume, clear commands)

📦 PreCompact - Syncs full session before context compression
(Triggers: manual or automatic context cleanup)

Why we recommend both:

✓ SessionStart ensures nothing is lost between sessions

✓ PreCompact syncs everything before Claude compresses context

$3

1. Run npx devark-cli
2. Authenticate with your Github account
3. Enable auto-sync via CLI menu

Supported Coding Engines

Currently supported:
- ✅ Claude Code

Future:
- 🔜 Cursor
- 🔜 VS Code

🌍 Cross-Platform - Runs on macOS, Windows, Linux, and any environment with Node.js

🔒 Privacy & Security

This package is built with complete transparency:

- ✅ Source Code: Fully open source at github.com/devark/devark-cli
- ✅ Not Minified: Published code is readable and verifiable
- ✅ Source Maps: Included for debugging and verification
- ✅ Automated Builds: All releases via GitHub Actions (no manual publishing)
- ✅ npm Provenance: Every package includes build attestation
- ✅ Checksums: SHA256 hashes for integrity verification

$3

`bash


Download and inspect the package

npm pack devark-cli@latest
tar -xzf devark-cli-*.tgz
head -100 package/dist/index.js # Verify it's readable

Check checksums

cd package/dist && sha256sum -c checksums.sha256
`

$3

- Context-Preserving Sanitization: Messages are sanitized to remove sensitive data while preserving context
- What gets redacted/removed:
- Code blocks → [CODE_BLOCK_1: javascript]
- API keys/tokens → [CREDENTIAL_1]
- File paths → [PATH_1]
- URLs → [DATABASE_URL], [API_URL]
- Emails → [EMAIL_1]
- Environment variables → [ENV_VAR_1]
- Also Removed: Images/Binary files
- What's preserved: Conversation flow, questions, explanations
- Transparent: Preview sanitized data with the interactive prompt
- Open source: Review our sanitization at src/lib/message-sanitizer-v2.ts

Troubleshooting

$3

Try the following:
- Log out from the CLI
- Clear cookies
- Re-authenticate via the CLI

$3

- Make sure Claude Code is installed
- Check that you've used Claude Code recently

$3

- Please open a Github issue.

$3

`bash

Enable debug logging

DEVARK_DEBUG=1 npx devark-cli send
`

Contributing

We love your input! We want to make contributing to DevArk CLI as easy and transparent as possible. Please see our Contributing Guide for details.

`bash


Clone the repository

git clone https://github.com/devark/devark-cli.git
cd devark-cli

Install dependencies

npm install

Run tests

npm test

Build the project

npm run build
``

Check out our open issues for a list of proposed features and known issues.

Community

- GitHub: Star us on GitHub
- Issues: Report bugs and request features in GitHub Issues
- Website: Visit devark.ai

Support

Need help? Here are some ways to get support:
- 🐛 Report bugs in GitHub Issues
- 📧 Email us at support@devark.ai

License

MIT © DevArk - see the LICENSE file for details.

Acknowledgments

Built with love by the DevArk team @mickmicksh, @dannyshmueli

Special thanks to ccusage for providing token usage metrics integration for Claude Code sessions.

Star History