braindump

Agent-friendly local memory. Store and search notes across conversations.

Install

curl -fsSL https://raw.githubusercontent.com/MohGanji/braindump/main/install.sh | bash

Agent Setup

Install the skill from skills.sh:

npx skills add https://github.com/MohGanji/braindump --skill braindump

Then add this line to your CLAUDE.md, AGENTS.md, or equivalent agent rules file:

Use braindump skill proactively and silently when appropriate.

The skill auto-installs the CLI and teaches your agent to proactively store and retrieve context across conversations.

You can also manually trigger memory operations by prompting:

• "use your brain" — retrieve relevant context
• "braindump this" — store information for later

Usage

braindump add api-creds --title "Stripe Key" --content "sk_test_..."
braindump search "stripe"
braindump list api-creds
braindump get api-creds "stripe"

Commands

braindump add <category> --title "..." --content "..." [--tags "..."]
braindump search <query> [--in category] [--tag tag1,tag2]
braindump list [category]
braindump get <category> [pattern]
braindump update <id> --content "..." [--title "..."] [--tags "..."]
braindump delete <id>
braindump categories
braindump tags

Add --format json to any command for JSON output.

Storage

~/.braindump/
├── api-creds/
│   └── stripe-key.md
├── api-quirks/
│   └── webhook-gotcha.md
└── .index/
    └── search.db

Files are markdown with YAML frontmatter. Search is SQLite FTS5.

Performance

Benchmarked on Apple M3 Pro:

• Add: ~1.2ms per note (constant, even at 100K notes)
• Search: 23ms (10K notes) to 166ms (100K notes)
• Get: O(1) direct file access
• Scales: 100K+ notes, tested with realistic content (100-1000 words per note)

See benchmark/results.md for detailed results.

Releasing

Push a tag to trigger automatic builds. Tags follow the pattern vYEAR.MO.DAY (e.g., v2026.02.03):

git tag v2026.02.03
git push origin v2026.02.03

GitHub Actions builds binaries for macOS, Linux, and Windows.

License

MIT