Explore

Send WhatsApp messages, list chats, and search history via wacli (local CLI backed by a synced store at ~/.wacli). Unpaired? Run the guided connect flow (scripts/guided_connect.py) from chat — no terminal needed.

Discover, evaluate, safely install, and update reusable skills from a curated set of trusted open-source repositories.

Infer stable semantic workstreams for ungrouped Sutando task history and apply the validated assignments to the durable workstream sidecar. Use for internal task-workstream-grouping maintenance tasks, initial history backfills, and later batches of tasks that need cross-channel or cross-device workstream organization.

Single entry point for fresh-session bootstrap. Runs optional task-orphan recovery, cron registration, and watcher start in a fixed order. Replaces the current `claude -- '/schedule-crons'` invocation pattern as the canonical CLI startup target.

Sutando introspection — read logs + git + memory + build log for a chosen time window and produce a concise narrative of what the agent has been doing, what's broken, and what to prioritize next.

Start or stop a screen recording of the Mac via ffmpeg. Use when the user asks to record/capture their screen to a video file (a demo, a repro, a walkthrough) — not for a single still screenshot (use macos-tools screen capture for that).

Sutando watches your screen and helps in real time, with the interaction pattern pre-configured per use case. Reads papers with you, pair-debugs, reviews PRs — without you having to narrate intent every session.

Review tooling for a PR: review-preflight prints REVIEW.md's criteria and the PR's live gate state before a review; ci-triage maps a PR's failing checks to already-filed issues.

Search phone-call history for when a feature regressed (find-regression.py) and drill into a single call to see what went wrong (diagnose-call.py). Skips reading 100+ transcripts by hand.

Track Claude Code quota usage via Anthropic API rate limit headers. Shows 5h and 7d utilization, reset times, and quota status. Works with both subscription and API key auth.

Start Sutando's autonomous proactive loop. Monitors tasks, runs health checks, and builds missing capabilities on a recurring schedule.

GitHub adapter binding for the shepherd contract: turn a real pull request into observed events, resolve the responsible actor from commit authorship, and persist the waiting contract under the workspace so a later pass (or process) can resume it.

Make conversational phone calls and join Zoom meetings via Twilio + Gemini. Multi-turn AI conversations on the phone on behalf of the user.

Ask the owner a decision you need, list what is waiting on him, close a question once he has answered, or run the reminder. Use when blocked on the owner's word, or asked "what are you waiting on me for?".

Before answering your OWNER, when the answer genuinely depends on something another agent of the same owner holds and you do not, ask that one agent once, with a plain message in the owner's owner-only consult room, and tell the owner you asked. Its reply in the thread is matched to your ask by Matrix relations and relayed to the owner, or, past the nudge time, you tell the owner there is no answer. One question, one answer per owner task; no onward consults and no automatic recovery. Inert until a consult room and room transport are configured.

Framework + minimal example for Sutando desktop overlay applications — always-on-top, frameless Electron windows that float over the desktop, controllable from the Sutando web UI's /overlays manager. Ships one example overlay (System Resources). Add new overlays by registering them in app/main.js.

Render text to mp3 via OpenAI's tts-1-hd. Use for video narration, demo voiceovers, audio notes.

Resolve a fuzzy GitHub reference (issue number, PR number, or text description) to a URL and optionally open it in the browser.

Connect to Microsoft 365 (OneDrive, Outlook mail, Calendar, Teams) via Microsoft Graph using the open-source python-o365 (O365) library. Use when the user wants to list/download OneDrive files, read or send Outlook email, list calendar events, or post a message to a Teams channel from a Sutando workflow.

Change the core's model without the CLI: send /model to the live core's pane through the shared sender, wait for the CLI to accept it (answering its confirm dialog only under --confirm), record only once accepted; persistence is the CLI's own.

Turn on/off whether a message that @-tags the OWNER counts as a mention of the bot. Off (default) is today's behavior — in requireMention channels such messages never reach the fleet. On, they are ingested as tasks and audit-logged. Free-listen channels are unaffected either way.

Schedule a small meeting end-to-end: resolve attendee emails, check the owner's calendar for the slot, dedup-check, then create + email the Google Calendar invite. The mechanical core only — cross-person availability negotiation stays interactive.

Build a short news-explainer video tuned for shareability. One striking moment per video; real fetched assets; self-heal validation; pluggable TTS (Gemini-free default, OpenAI fallback).

GUI control for macOS apps via mediar-ai's mcp-server-macos-use. Click, type, scroll, key-press, open apps — driven by accessibility tree, works in non-interactive Claude Code mode. Use this for any Sutando task that needs to drive another macOS application (Safari, Zoom, Mail, Finder, etc.).

Hardened loopback static server for local presentation drafts — capability-URL, read-only GET/HEAD, Host-header gate, nosniff + sandbox CSP. Use instead of a plain http.server when the Presentation panel's dev mode loads a local deck.

Persist collection evidence and plan unread room windows separately from learned facts.

Render text to mp3 via Google Gemini Flash TTS. Free-tier eligible (1500 req/day). Use for video narration, demo voiceovers, audio notes. Parallels openai-tts; default for make-viral-video.

Ask Sutando to resolve a conflicted update of a Sutando checkout. When pulling the latest release/main conflicts with local changes, Sutando reproduces the merge in a scratch git worktree, resolves the conflicts there, proposes the resolution to the owner, and fast-forwards the live checkout only after explicit confirmation. First entry path is the desktop app's "Resolve with Sutando" engine-update button.

Scan configured sources (Craigslist now; eBay + Facebook Marketplace planned) for used-item listings matching the owner's criteria. Currently configured for a Mac mini search (M2+, 16GB+, 512GB+, ≤$500, near 94566). Notify owner via SMS + Telegram on a match.

Maintain and use a cross-channel collaboration map of rooms, people, agents, agent owners, relationships, expertise, ownership, roster, priority/VIP handling, and recent context. Trigger when a task or message carries room/channel/sender/member metadata; the user asks who is in a room, what it is for, who owns or knows a component, where to ask, or whom to contact or cc; you must coordinate, delegate, hand off, escalate, or pick who reviews something — including when phrased as an action you already know how to perform, since naming it as a familiar action is how this trigger gets missed; a merge or approval gate reports a PR queued on nobody, short of its required approvals, or carrying a stale approval; a participant is unfamiliar, ambiguous, or designated VIP/priority; identities must be reconciled across bridges; or a roster needs refresh. Do not trigger for generic communication-platform questions, or drafting that does not depend on room, identity, relationship, or collaboration context.

Choose between the local Codex CLI, Gemini CLI, and Pi CLI from Claude Code. Use for automatic model selection when the user wants the best local delegate for code review, repo-wide analysis, planning, or implementation.

Use the local Pi CLI (pi.dev) from Claude Code with the user's existing Pi authentication (Kimi, Claude Pro/Max, or ChatGPT OAuth). Use for second-opinion analysis, implementation, or reviews from a Kimi-family model in the current workspace.

Use the local Antigravity CLI (agy, Gemini-backed) from Claude Code with the user's existing Gemini authentication or API configuration. Use for large-context repo scans, multimodal analysis, second-opinion planning, or structured Gemini runs in the current workspace.

Bash wrapper around the local Codex CLI for non-interactive runs from inside Sutando (bridges, cron, scripts). For interactive code review or task hand-off from this Claude Code session, prefer the official `/codex:*` plugin commands; this skill is the file-bridge-compatible path that `discord-bridge.py` invokes for team-tier sandboxed delegation.

Transcribes audio files and voice notes to text via Gemini 2.5-flash. Integrates with Slack, Discord, and Telegram bridges so voice clips surface as readable text in tasks.

Local Agent Registry — a standalone, dependency-free service that tracks running Claude Code (and other) agent instances. Agents self-register on startup and heartbeat while alive; the Electron overlay and Sutando dashboard read the live list. Use when you need to know which coding agents are running, where, and since when.

基于用户真实经历生成“阿酥式”高密度中文技术履历或小红书展示简历：保留真实姓名、学校、公司、部门、项目与可核验指标，主动检索学校官方 Title，以彩色经历条、背景/指标/职责结构、英文技术名词、架构链路、Benchmark、模块级 Owner、0→1 和大 Scope 叙事增强表达，并输出自包含 HTML、A4 PDF 与发布图片。用户提到阿酥风格、ASU 式简历、专业名词堆叠、技术经历包装、小红书履历长图、学校 Title、Owner/0-1 或要求 HTML/PDF 时使用。不得凭空创造不存在的学历、公司、项目、角色、Offer、指标或成果；夸张必须落在可证明的机构标签、限定模块和真实技术链上。

以证据为中心的简历真实性审计与报告生成技能。用户要求核验简历、打假履历、调查候选人经历、分析 GitHub/开源贡献、识别 contributor 到 maintainer/core author 的角色膨胀、核验中外合作办学或学校 title、检查业务指标、梳理公开争议，或者把调查结果制作成 HTML/PDF 时，应主动使用本技能。技能会把简历、截图、链接、帖子、代码仓库和官方记录拆成原子主张，区分事实、矛盾、指控和不可核验信息，并生成不含开盒信息、不把未经证实指控写成事实的自包含证据报告。

获取腾讯 ima 知识库（ima.qq.com/wikis）中的文章列表和原始文章详情。适用于：(1) 用户已经登录 ima 知识库，需要导出或读取知识库中的文章列表；(2) ima 网页端不支持直接查看文章详情，需要通过 source_path 原始链接获取内容；(3) 批量抓取个人知识库或共享知识库中的文章标题、来源和正文。触发词：ima知识库、ima.qq.com、腾讯ima、知识库文章列表、万仁刚的知识库。

自研产品客户拜访技能。当客户经理需要拜访政企客户、准备解决方案、生成方案文档、或整理拜访纪要时使用本技能。

Hand the entire repo-build-pr workflow to a harness of your choice (Claude Code headless or Codex): the chosen harness acts as orchestrator and validator — it creates its own worktree, implements (sub-delegating via repo-delegate where useful), runs the gate, runs the repo-review battery with cross-harness adversarial dispatch, and either stops ready-to-publish (default) or pushes and opens the draft PR (--publish). Use when the user wants to pick which harness orchestrates a build end to end, delegate a whole task AFK, or run the same build prompt under a different orchestrator for comparison.

Dispatch a scoped implementation task to another agent harness (Claude Code headless or Codex) from the current one: fill the shared task-brief template and run it through a per-harness runner script that can edit the worktree and run the gate, but never commits or pushes. Use when the user asks to delegate work to Codex or Claude, wants a second implementation pass from the other harness, or a skill (e.g. repo-review fix rounds) needs to hand a precise fix prescription to an implementer outside this session.

Use when the user wants to turn open Open Software Issues into AFK-ready, agent-implementable tasks: find ready-to-code Issues for an org/project, diagnose each against the local codebase (root cause, files, code fragments, dependencies, blockers, acceptance, verify), enrich the Issue description on the platform, optionally assign them to the current user, and emit a table of one-line `repo-build-pr` prompts to dispatch to autonomous agents. Triggers on "prep tasks", "get issues ready to implement", "enrich issues with diagnosis", "prepare build prompts", "ready-to-code tasks for agents".

Query and update live Open Software os-platform production data through the platform API. Use when an agent needs current Issues/Bounties, Orgs, Projects, Submissions, Comments, Activity, Contributors, Product Memory, the team Timeline, or API status, or needs to create, assign, move, or comment on tracked work.

Prove model-disabled skills stay outside the model catalog.

Exercise project skill discovery and loading in snapshot tests.

Prove user-disabled skills remain available to the model.

Diffusion-based docking that predicts protein-ligand poses without a predefined site. Use for blind docking, when traditional docking fails, or exploring multiple binding modes. Pipeline: prep protein (PDB) and ligand (SMILES/SDF), run inference, analyze confidence-ranked poses.

Deep learning for drug discovery. 60+ models (GCN, GAT, AttentiveFP, MPNN, ChemBERTa, GROVER), 50+ featurizers, MoleculeNet benchmarks, HPO, transfer learning. Unified load-featurize-split-train-evaluate API. For fingerprints use rdkit-cheminformatics; for featurization-only use molfeat.

Chunked N-D arrays with compression and cloud storage. NumPy-style indexing. Backends: local, S3, GCS, ZIP, memory. Dask/Xarray integration for parallel and labeled computation. For lineage use lamindb; for labeled arrays use xarray.

Python Materials Genomics library for structure analysis, thermodynamics, and electronic properties. Parse/create crystal structures (CIF, POSCAR), query Materials Project for DFT-computed properties, analyze phase and Pourbaix diagrams, compute XRD patterns, generate DFT inputs for VASP, Quantum ESPRESSO, CP2K. Alternatives: ASE (MD/geometry), AFLOW (high-throughput), OVITO (visualization).

Pipeline for Neuropixels extracellular electrophysiology: probe geometry (ProbeInterface), Kilosort sorting via SpikeInterface, quality metrics, unit curation (ISI, firing rate, SNR), post-sort analysis (PSTH, tuning curves, population decoding). Supports Neuropixels 1.0/2.0/Ultra in rodent/primate experiments.

Python toolkit for neurophysiological signal processing: ECG (HR, HRV, R-peaks), EEG (complexity, PSD), EMG (activation onset), EDA/GSR (SCR decomposition), PPG, and RSP. Includes synthetic signal simulation. Alternatives: BioSPPy (less maintained), MNE (EEG/MEG specialist), heartpy (ECG only), scipy.signal (raw DSP).

Query and download NCI Imaging Data Commons (IDC) cancer radiology and pathology datasets via the idc-index Python client. No authentication required: the parquet index ships inside the pip wheel, SQL runs locally via DuckDB, and DICOM downloads stream from public S3/GCS buckets through s5cmd. Use sql_query() for DuckDB cohort selection, get_collections/get_patients/get_dicom_studies/get_dicom_series for hierarchical browsing, download_from_selection() for downloads, and get_viewer_URL() for OHIF/Slim links. Use pydicom-medical-imaging for local DICOM reading; histolab for whole-slide pathology preprocessing.

Hardware-agnostic Python liquid-handler library: portable scripts run on Hamilton STAR, Tecan Freedom EVO, Opentrons OT-2, or a simulator without vendor lock-in. For protocol automation, method dev, plate reformatting, serial dilutions, and Python lab workflows.

Query CELLxGENE Census (61M+ cells). Search by cell type/tissue/disease/organism; get AnnData, stream out-of-core, train PyTorch models. For your own data use scanpy; for annotated data use anndata.

Rust-backed Python library for fast genomic token arithmetic and BED processing. High-performance BED I/O, interval set ops (intersect, merge, complement, subtract), region tokenization against a universe, universe construction. Use for preprocessing large BED collections and ML token vocabularies.

NCBI GEO access via GEOparse and E-utilities. Search by keyword/organism/platform, download GSE series matrices, parse GPL annotations, extract GSM metadata, load expression matrices into pandas. For single-cell use cellxgene-census; for multi-DB access use gget-genomic-databases.

QA testing workflow for CamoFox Browser — systematic testing with console capture, error detection, and Playwright tracing. Use when the user needs to test a web application, perform QA validation, do exploratory testing, or verify web app behavior. Triggers include "test this website", "QA", "dogfood", "exploratory testing", "find bugs", or any request to systematically test a web application.

Auto-prepare for upcoming meetings: attendee info, recent email threads, talking points, and agenda. Runs 30 min before each meeting or on demand.