YouOS

v0.1.18

YouOS — local-first personal email copilot that learns your writing style from Gmail, Google Docs, and WhatsApp exports, then drafts replies in your voice. U...

0· 189·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for drbaher/youos.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "YouOS" (drbaher/youos) from ClawHub.
Skill page: https://clawhub.ai/drbaher/youos
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Required binaries: python3, gog
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install youos

ClawHub CLI

Package manager switcher

npx clawhub@latest install youos
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
Purpose & Capability
Name/description (personal email copilot) align with what the package requires and does: Python app that ingests Gmail/Google Docs/WhatsApp and builds local indices. Required binaries (python3, gog) and the documented dependency on the gog CLI for Gmail/Docs access are expected and appropriate.
Instruction Scope
SKILL.md and code instruct the agent/user to run local ingestion, build a local SQLite DB, run periodic 'improve' pipelines, and optionally install a background launchd service and a Gmail bookmarklet that injects drafts into the Gmail compose UI. Those are within the stated purpose but warrant caution because they touch user email data and may insert UI elements into Gmail. The SKILL.md explicitly warns about sensitive data ingestion and recommends local-only operation; instructions do not appear to instruct reading unrelated system files or exfiltrating secrets by default.
Install Mechanism
Install is manual via 'pip install -e .' which runs local package install code (SKILL.md warns to review source). No remote download URLs or package-install-from-untrusted-host steps are present in the instructions. One metadata inconsistency: the registry summary initially said 'No install spec — instruction-only skill', but SKILL.md contains explicit install instructions and the bundle includes many source files, so this is not instruction-only—review the supplied source before running pip install.
Credentials
Requested credentials/inputs are proportional: the skill needs the gog CLI authenticated to the user's Gmail/Docs accounts (documented). An optional external model (Claude) credential is only used if the user enables external fallback. There are no unrelated or inexplicably broad environment variables or secret requests declared.
Persistence & Privilege
The skill does not force permanent inclusion (always:false) and can operate locally. However the repository includes scripts to install a launchd background service (install_youos_launchd.sh / run_youos_launchd.sh) and various automation (nightly pipeline). Those are optional and only executed if the user runs them, but they grant persistent background execution when installed—review those scripts before enabling automation.
Assessment
This package largely does what it claims, but take these precautions before installing: - Review the source before running 'pip install -e .' (that executes package install code locally). - Confirm you trust the repository: it will ingest sensitive data (Gmail, Docs, WhatsApp) into a local SQLite DB. - Keep model.fallback set to 'none' if you want strictly local operation; enabling external fallback can send content to third-party model providers. - Inspect the launchd install scripts and bookmarklet code before running/ installing them—they enable background jobs and inject UI elements into Gmail when activated. - Ensure the gog CLI you authenticate is the intended account and that you understand what scope/permissions gog grants. If you want stronger assurance, run the app in an isolated environment (dedicated user/VM) and audit outgoing network activity while testing.

Like a lobster shell, security has layers — review code before you run it.

Runtime requirements

Binspython3, gog
latestvk971csh53a0ny3hd75xeskqqr5836smm
189downloads
0stars
12versions
Updated 1mo ago
v0.1.18
MIT-0

YouOS — Personal Email Copilot

YouOS is a full local Python app (not an instruction-only snippet). It drafts email replies in your style, grounded in your real past replies.

Safety / impact (read before install)

  • This app can ingest sensitive personal data (Gmail, Docs, WhatsApp exports) into a local SQLite database.
  • pip install -e . executes package install code locally; only run on trusted source.
  • Persistence is optional: background service install happens only if you explicitly run launchd install scripts.
  • Scheduled/nightly improvement runs are opt-in/manual unless you configure automation yourself.
  • External model fallback is optional; set model.fallback: none for local-only operation.

Install and runtime model

  • Install is manual via pip (pip install -e .) in a Python 3.11+ environment
  • Note: pip install -e . executes local package install code from this repository; review source before installing
  • Requires both:
    • python3 (3.11+)
    • gog CLI authenticated to the Gmail account(s) you want to ingest
  • Optional runtime path override: YOUOS_DATA_DIR

Credentials and configuration

  • Required: gog authentication for Gmail/Docs ingestion
  • Optional: Claude CLI/API credentials only if using external fallback generation
  • Recommended for local-only privacy: set model.fallback: none

Trigger phrases

  • "draft a reply to this email"
  • "write this email for me"
  • "how would I respond to this"
  • "what would I say to"
  • "help me reply"
  • "draft in my style"
  • "youos"
  • "my email copilot"
  • "email copilot"
  • "my copilot"
  • "generate a draft"
  • "reply draft" / "email draft" / "draft reply"
  • "compose reply"
  • "write a response"
  • "email response"
  • "how do I usually reply to"
  • "reply to this"
  • "help me write"
  • "write an email"
  • "compose a response"
  • "email assistant"
  • "my writing style"
  • "train on my emails"

Requirements

  • Apple Silicon Mac (M1/M2/M3/M4) with 8GB+ RAM (16GB recommended)
  • Python 3.11+
  • gog CLI configured with your Gmail account(s)
  • ~5GB free disk space
  • Run the UI locally by default (do not expose publicly unless intentionally secured)

Quick start

# Install
cd ~/Projects/youos
pip install -e .

# Check system requirements (Python, gog CLI, disk space, etc.)
youos doctor

# Run setup wizard (15 min, mostly ingestion)
youos setup

# Draft a reply
youos draft "paste inbound email here"
youos draft --sender john@company.com "email text"

# Open web UI
youos ui

# Check status
youos status

# Run nightly pipeline manually (add --verbose for step-by-step output)
youos improve
youos improve --verbose

# Run golden benchmark evaluation (8 curated test cases)
youos eval --golden

# Full corpus health report (pairs, quality scores, top senders)
youos corpus
youos corpus --json

# Ingest a WhatsApp chat export (optional — augments your corpus)
youos ingest --whatsapp ~/Downloads/WhatsApp-Chat.txt

# Add sender note (immediately rebuilds their profile)
youos note john@company.com "integration partner, prefers bullet points"

# Submit a feedback pair directly from the terminal
youos feedback --inbound "email text" --reply "your reply" --rating 4

# View stats
youos stats

# Teardown (remove all data, keep code)
youos teardown

Gmail Bookmarklet

Install from the Bookmarklet page in the web UI. Once installed:

  • Click bookmarklet on any Gmail thread → floating panel opens on the right
  • Click Generate Draft → draft appears in the panel
  • Click Insert into Gmail → draft injected into compose window
  • Rate the draft with stars and submit feedback — all without leaving Gmail
  • Click the bookmarklet again to close the panel

How it works

  1. Ingests Gmail, Google Docs, WhatsApp exports — plus organic pairs from emails you sent without YouOS
  2. Builds a retrieval index — BM25 + query expansion + semantic (LRU-cached) + multi-intent + per-account isolation + same-thread 2× + subject + topic signals + sender-type boosts + quality scores + relative confidence thresholds
  3. When you ask for a draft: detects multi-intent, retrieves score-ranked thread-deduplicated exemplars (reply preserved 600 chars, inbound trimmed 400), prompt token budget enforced; generates using per-mode persona with first-name greeting; local model empty/signature-only output falls back to Claude automatically
  4. Every email you review trains the model further — curriculum-ordered, quality-filtered, training pairs deduplicated by similarity, DPO pairs supported; nightly pipeline skips steps when data insufficient
  5. Nightly: ingests + organic pairs, incremental persona re-analysis (90-day weighted, EWMA avg words, p25/p75 confidence intervals), fine-tunes (with golden eval check), runs autoresearch on rotating benchmark sample
  6. Autoresearch benchmarks rotate weekly (seeded re-sample) — prevents overfitting to fixed test cases; golden eval composite tracked in pipeline log
  7. Style drift detection: Stats dashboard flags when your writing patterns shift significantly
  8. Your best-rated, least-edited replies surface higher in future retrievals via quality scoring
  9. Sender profiles track reply-time patterns and topics; youos note immediately rebuilds that contact's profile
  10. Submit feedback from terminal: youos feedback --inbound "..." --reply "..." --rating 4
  11. Setup wizard asks for internal domains — accurate sender classification from day one
  12. Facts store (/api/facts) — save context about contacts, projects, and preferences; facts are injected into generation prompts automatically for context-aware drafts
  13. Auto fact extraction — sender notes and feedback notes are parsed automatically on save using 15+ rule patterns (preferences, timezone, schedule, sign-offs, roles, relationships, project metadata); negation-aware with confidence scoring; LLM (Claude CLI) fallback for unstructured notes; fact deduplication/merging on upsert

Security & privacy notes

  • Gmail ingestion uses your local gog authentication; review connected accounts before ingestion
  • External LLM fallback is optional; if enabled (model.fallback: claude), inbound email/context can be sent to Claude for generation
  • For strict local-only operation, set model.fallback: none in youos_config.yaml
  • Data location defaults to local project paths (or YOUOS_DATA_DIR if set)
  • Review PRIVACY.md before first ingestion/deployment

Provenance

Comments

Loading comments...