Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add sonichi/sutando --skill ag2space-voicegit clone --depth 1 https://github.com/sonichi/sutandoWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sonichi/sutando/ag2space-voice)<a href="https://agentmods.dev/skills/sonichi/sutando/ag2space-voice"><img src="https://agentmods.dev/badge/skills/sonichi/sutando/ag2space-voice/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sonichi/sutando/ag2space-voice"><img src="https://agentmods.dev/badge/skills/sonichi/sutando/ag2space-voice.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00052 | $0.01124 |
| Opus 5.5 | $0.00021 | $0.00450 |
| Sonnet 5.5 | $0.00010 | $0.00225 |
| Haiku 4.5 | $0.00005 | $0.00112 |
Grade A, and why
ag2space-voice scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AG2 Space voice
Optional plugin for installs that talk to an AG2 Space client (desktop app or webapp).
The host engine carries a product-neutral voice session origin (VoiceSessionOrigin,
src/task-bridge.ts) and offers client frames to skills (setup(ctx), see
skills/MANIFEST.md). This skill owns everything AG2 Space specific on
top of that. Remove the directory and the engine boots, types and tests unchanged.
Wire contract (client side: webapp features/room/voice/)
| Direction | Frame |
|---|---|
| client → engine | {type:'session.context', version:1, room_id, room_name, surface:'room'|'dm', capabilities?} — after session.config and on every room change; a DM frame has room_id: null |
| engine → client | {type:'session.context.ack', version:1, room_id, bound, surface:'dm'|'room'|'refused', reason?} |
| engine → client | {type:'ui.navigate', version:1, request_id, target:'dm'|'room'|'home', query?} |
| client → engine | {type:'ui.navigated', version:1, request_id, ok, room_id?, room_name?, error?:'not_found'|'ambiguous'|'unsupported', candidates?} |
What it does
session-context.ts— parses and bounds the frame (Matrix room id grammar, name cap, capability list) and builds the ack.room-binding.ts— the room a client names is a claim. It becomes the session's origin only when the gateway bridge answersstate/voice-room-checks/<key>.request.jsonwith a verified<key>.verdict.json(owner and agent both joined). A different room releases the bound one before the wait; a newer frame supersedes a pending one; a refusal, a throw or silence leaves the session on the owner DM.- Origin handed to the core: channel
ag2space, target = the room id, header keyschannel_kind: room+source_room_id, theroom_context:body line, the[dm-only]delivery note, and averifythat re-asks the gateway bridge when the result is delivered. Results therefore land asresults/proactive-result-<task>-<ts>.to-ag2space.txtwith[channel: <room>]first. - In-flight work keeps the origin and recent context from when the
workcall began (or the turn the relay agent's turn check sends), including across the relay agent's wait for the user to finish, its request check, a room switch or disconnect/reconnect. A pending transcription is omitted if the room changes, the turns provider is replaced, or the client disconnects; the new session's words never fill that old task. Re-announcing the room already bound, and a spoken farewell ("research X, bye"), change none of these, so the owner's words for that task are kept. Delivery still re-checks the original room's membership and honors[dm-only]. - Prompt: a
ROOM:context line while docked, and one system notice per actual room change. navigate.ts/navigate-protocol.ts— thenavigate_uitool ("let's talk in my DM", "take me to GTM in Investors", "go home"). It is contributed throughvoiceSurface(), so it is on the web voice session only, never the phone tool table, and only when this install has the gateway channel (REMOTE_TASK_TOKEN/AG2_REMOTE_TOKENin the environment or inchannels/ag2space/.envunder the Claude config dir). Its NAVIGATION prompt rule is present under the same condition. The frame goes only to a client that announced theui.navigatecapability; any other attached client gets an immediate "update the app" answer.
What ships with it
8 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago Changed · +9 lines a41d44b92221
- 25d ago First seen · 60 lines · 52 tokens per session scan A d40524cd997c
ag2space-voice is a skill published in the GitHub repository sonichi/sutando (397 stars, last pushed today), licensed MIT. It adds 52 tokens to every session and 1,124 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-22.
Other skills, from other repositories
app-builder
Build and edit small, personal visual tools and artifacts — dashboards, trackers, calculators, data visualizations, charts, simple landing pages, and slide decks the user wants for THEMSELVES. This is the right skill whenever the user asks to "visualize this," "make a chart," or "build an artifact" for their own use…
gmail
Manage Gmail email — drafting, sending, organizing, filters, vacation replies, and inbox analysis.
assistant-migration
Migrate from ChatGPT, Claude, OpenClaw, Hermes, Manus, and other AI assistants into Vellum by inspecting their data exports, conversation archives, files, prompts, custom instructions, memory, saved memories, tools, GPTs, workflows, integrations, and relationships, then mapping as much as safely possible into Vellum…
guardian-verify-setup
Set up channel verification for phone, Telegram, Slack, Discord, or email channels via outbound verification flow.
stripe-link-wallet
Agent wallet on the Stripe Link CLI (link-cli). Connect the wallet, then buy, purchase and pay for things on the user's behalf via approved spend requests, one-time-use cards, and 402 / Machine Payment Protocol (MPP) payments.
media-processing
Ingest and process media files (video, audio, image).