@lebronj/pi-suite
Slim Pi suite: vendored memory plus pi-web-access bootstrap companion.
Package details
Install @lebronj/pi-suite from npm and Pi will load the resources declared by the package manifest.
$ pi install npm:@lebronj/pi-suite- Package
@lebronj/pi-suite- Version
0.1.42- Published
- Sep 8, 2026
- Downloads
- 1,603/mo · 65/wk
- Author
- lebronj
- License
- MIT
- Types
- extension
- Size
- 360.6 KB
- Dependencies
- 0 dependencies · 5 peers
Pi manifest JSON
{
"extensions": [
"./vendor/pi-memory/index.ts"
]
}Security note
Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.
README
@lebronj/pi-suite
Slim Pi package for a minimal Pi setup: vendored memory in this package, plus pi-web-access as the bootstrap companion.
The npm package itself only loads the vendored memory extension. Benchmark tools, media tools, planning tools, safety-gate, autogoal, skill-creator, pi-lsp, and Playwright companions are not included in the default manifest.
Install
npm install -g --ignore-scripts @earendil-works/pi-coding-agent@0.84.3
pi install npm:@lebronj/pi-suite
pi install npm:pi-web-access
Or bootstrap Pi, write the default provider config, install this package, and install pi-web-access:
curl -fsSL https://registry.npmjs.org/@lebronj/pi-suite/-/pi-suite-0.1.42.tgz | tar -xzO package/scripts/bootstrap.sh | bash
The bootstrap script asks for the API key on the terminal; it does not ship a key.
Defaults written to ~/.pi/agent/models.json / settings.json:
- Provider:
lenovo-deepseek-v4-flash - Base URL:
https://modelfactory.lenovo.com/service-large-600-1777255649450/llm/v1 - API:
openai-completions - Default model:
DeepSeek-V4-Flash-0731
Override with TEAM_BASE_URL, TEAM_MODEL, or TEAM_API_KEY. Legacy endpoint remains available with TEAM_PROFILE=legacy through scripts/bootstrap.legacy.sh.
Zhizengzeng profile: GPT-5.5
TEAM_PROFILE=zhizengzeng ZHIZENGZENG_API_KEY=sk-... bash scripts/bootstrap.sh
Writes the zhizengzeng provider (https://api.zhizengzeng.com/v1) with gpt-5.5 / gpt-5.5-pro as main control models.
What is included
Vendored memory
Loaded from:
vendor/pi-memory/index.ts
Slim mode exposes the memory tools used by the harness:
memory_read— read long-term memory, user/state/review files, scratchpad, and daily logs.memory_search— search memory files; falls back to local lexical search when qmd/embeddings are unavailable.
The vendored @jhp/pi-memory source also contains write/edit/curate, review, skill-draft, and sync functionality. Set the relevant PI_MEMORY_* environment variables if you intentionally want those fuller modes.
pi-web-access companion
The bootstrap script installs:
npm:pi-web-access
It also writes ~/.pi/web-search.json with fetch/search fallback routing, including jina as a fallback for pages that fail plain HTTP/Readability because of 403, anti-bot, or JS-rendering issues.
Set SERPER_API_KEY / JINA_API_KEY in that config file if you want the paid/keyed fallback tiers.
Not included by default
These are intentionally not loaded or packed by the npm manifest:
extensions/update-plan.tsextensions/bench-control.tsextensions/media-tools.tsextensions/bench.tsextensions/safety-gate.tsextensions/autogoal.tsskills/skill-creatorpi-lsppi-playwrightpi-subagentspi-mcp-adapterpi-mono-figma
Memory notes
memory_search works without qmd or embeddings. Optional BM25/semantic/deep search support requires qmd:
bun install -g https://github.com/tobi/qmd
qmd collection add ~/.pi/agent/memory --name pi-memory
qmd embed
Memory versioning is local-only by default. If you want backup sync, set PI_EVOLUTION_REMOTE to your own private repo before bootstrap/setup, or add a personal remote later:
git -C ~/.pi/agent/evolution remote add origin <url>
Set PI_EVOLUTION_AUTO_PUSH=1 only if automatic remote sync is desired.
Update
Update Pi itself:
pi update --self
Update installed Pi packages:
pi update --extensions
Publish
cd /home/jianghp3/.pi/packages/pi-suite
npm pack --dry-run
npm publish --access public