screen-companion

screen-companion is a skill for Claude Code, Codex from sonichi/sutando. It costs 50 tokens per session (1,977 once invoked), scanned A, original, MIT.

A set of prepared screen-and-voice interaction modes for the coding agent. Each mode gives the agent a specific purpose, such as reading a paper, pair-debugging, or reviewing a code change.

In plain words
What is it for?
Use it for guided paper reading, debugging with an IDE, or walking through a GitHub pull request. It is intended for active collaboration, not silent screen monitoring.
Why use it?
Without a preset mode, you must explain the same goal and interaction style at the start of every screen-sharing session.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it for guided paper reading, debugging with an IDE, or walking through a GitHub pull request. It is intended for active collaboration, not silent screen monitoring.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/sonichi/sutando/screen-companion
Install

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.

Any agent
npx skills add sonichi/sutando --skill screen-companion
Clone the repo
git clone --depth 1 https://github.com/sonichi/sutando

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for screen-companion

README.md
[![agentmods](https://agentmods.dev/badge/skills/sonichi/sutando/screen-companion.svg)](https://agentmods.dev/skills/sonichi/sutando/screen-companion)
Your own site
<a href="https://agentmods.dev/skills/sonichi/sutando/screen-companion"><img src="https://agentmods.dev/badge/skills/sonichi/sutando/screen-companion.svg" alt="Measured on agentmods" height="20"></a>
Per session 50 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,977 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 2 findings, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium MCP Rug Pull · line 74
    npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.
    Fix: Pin the version: npx @scope/[email protected]
  • medium MCP Rug Pull · line 83
    npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.
    Fix: Pin the version: npx @scope/[email protected]
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00050 $0.01977
Opus 5 $0.00025 $0.00988
Sonnet 5 $0.00010 $0.00395
Haiku 4.5 $0.00005 $0.00198

Measured 8d ago against content hash 2476ad9aafb8, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

screen-companion 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 8d ago.

The scan reads SKILL.md. This mod also ships 4 executable files (scripts/activate.ts, scripts/load-config.ts, scripts/read-selection.ts, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

skills/screen-companion/SKILL.md · 130 lines

How it starts

The opening of the file, as written. The whole thing — 130 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Screen Companion

Sutando is already capable of watching your screen + voice-chatting in real time (vision push-mode from PR #735 + bodhi VoiceSession). The gap this skill closes: Sutando doesn't know what you're trying to do. Every session you have to re-narrate intent ("I'm reading a paper, ask about figures"; "I'm debugging, suggest hypotheses").

This skill ships pre-baked interaction-pattern configs — each one encodes a purpose, the right system-prompt overlay, the right tool subset, and the right vision cadence. Owner activates a config by name; the skill handles the rest.

When to Use

  • "Read this paper with me" — paper-reading mode (PDF / arXiv / blog).
  • "Debug this with me" — stack-trace + IDE pair-debug.
  • "Review this PR with me" — GitHub PR diff walk-through.
  • Or any other use-case that has a config in configs/.

NOT for: silent screen-watching (use the voice agent directly), one-shot questions about a screenshot (use the look_at_screen inline tool).

Architecture

configs/<name>.yaml             # the interaction pattern, declarative
        ↓
scripts/activate.ts             # entry: --config <name> [--goal "..."]
        ↓ loads config, builds:
        ↓
voice-agent's VoiceSession      # gets a system-prompt overlay + tool subset
        ↓                       # + vision_mode + cadence_ms config
Push-mode vision frames flow    # at the configured cadence
        ↓
Owner asks questions in voice   # answers grounded in what's on screen

The skill itself is small — most of the value lives in the configs. New use case = drop a YAML into configs/. No code change required.

Configs ship with v0

Config Activation Vision mode Shape
guided-setup --config guided-setup --goal "..." or "guide me through this" push, 700ms proactive + goal-directed (Sutando narrates next steps for a configuration task)

v0 demo angle: "Sutando helps you set up something you've never done before" (e.g., a Discord dev portal bot config). User shares screen, names the goal at activation, Sutando narrates the next concrete step in real time. More visceral demo than paper-reading; matches a near-universal user pain (everybody fights some dev portal once a quarter).

Read the full file on GitHub · 130 lines

Files

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.

Changes

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.

  1. 8d ago First seen · 130 lines · 50 tokens per session scan A 2476ad9aafb8

Subscribe to this mod's changes

screen-companion is a skill published in the GitHub repository sonichi/sutando (393 stars, last pushed yesterday), licensed MIT. It adds 50 tokens to every session and 1,977 once invoked, about $0.0003 per session on Opus 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 2026-08-30.