wiki-recall

wiki-recall is a Bun/TypeScript CLI for saving, recalling, and searching project context so a future coding session can resume with less re-explanation.

What it does

• Stores working scenarios: repositories, skills, current context, next steps, blockers, and notes.
• Recalls a scenario later by restoring saved context and optionally syncing repositories.
• Manages a local knowledge wiki and search index for reusable project knowledge.
• Can run as an MCP server over stdio for editor/agent integration.

Quickstart / install

Prerequisites:

• Bun 1.1 or later
• Git
• Python 3.11+ only if you use the Python engine scripts in engine/

git clone https://github.com/aviraldua93/wiki-recall.git
cd wiki-recall
bun install

Optional Python engine dependencies:

python -m pip install chromadb PyYAML

How to run

Run the CLI from source:

bun run dev --help
bun run dev init
bun run dev create my-project --template web-api
bun run dev save my-project --summary "Implemented retry handler" --next-step "Add tests"
bun run dev recall my-project

Set up a local ~/.grain knowledge base with the PowerShell wizard:

powershell -ExecutionPolicy Bypass -File scripts/setup.ps1 -Quick
powershell -ExecutionPolicy Bypass -File scripts/setup.ps1 -Interview

Quick Setup (-Quick) creates a minimal local brain from form prompts. Deep Interview (-Interview) uses Copilot CLI prompts to interview you and mine session history for richer project context. All data stays local unless you explicitly sync it.

Build a standalone executable:

bun run build
./wikirecall --help        # macOS/Linux
.\wikirecall.exe --help    # Windows

Other useful commands:

bun run dev list
bun run dev knowledge --help
bun run dev visualize --help
bun run dev mcp
bun run dev benchmark --help

How to test

bun run lint       # TypeScript type check
bun run test:unit  # unit tests
bun run test:e2e   # local end-to-end workflow tests
bun run test       # full Bun test suite

Current validation: bun run test runs 1,748 tests across 59 files. The e2e suite uses local temporary workspaces and does not require live services.

Project layout

• src/cli/ — CLI entry point and commands
• src/scenario/ — scenario create, recall, save, handoff, teardown, push, and pull logic
• src/knowledge/ — knowledge entity storage, search, and visualization
• src/memory/ — layered memory configuration and routing
• engine/ — optional Python indexing/harvest helpers
• skills/, templates/, schemas/ — packaged skills, starter files, and validation schemas
• tests/ — Bun unit and e2e tests

License

MIT