agentmods.dev

katashiro-web-task

katashiro-web-task is a skill for Claude Code, Codex from brettchien/katashiro. It costs 167 tokens per session (2,097 once invoked), scanned A, original, MIT.

A browser-task guide for Katashiro, which lets an agent read and operate pages in the user's real Chrome session, including logged-in pages.

In plain words
What is it for?
Use it to read pages, fill forms, click through workflows, switch tabs, and paste screenshots into tools such as Jira.
Why use it?
It reduces the risk of acting on the wrong tab or changing a page without first checking what is open.

Skill for Claude CodeCodex

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

not rated 10repo 3d ago A scan 167 tokens original MIT

Good fit Use it to read pages, fill forms, click through workflows, switch tabs, and paste screenshots into tools such as Jira.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/brettchien/katashiro/katashiro-web-task
View source ↗ brettchien/katashiro
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 brettchien/katashiro --skill katashiro-web-task
Clone the repo
git clone --depth 1 https://github.com/brettchien/katashiro

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 katashiro-web-task

README.md
[![agentmods](https://agentmods.dev/badge/skills/brettchien/katashiro/katashiro-web-task/github.svg)](https://agentmods.dev/skills/brettchien/katashiro/katashiro-web-task)
Your own site
<a href="https://agentmods.dev/skills/brettchien/katashiro/katashiro-web-task"><img src="https://agentmods.dev/badge/skills/brettchien/katashiro/katashiro-web-task/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.

agentmods 80×15 button for katashiro-web-task

Your own site · 80×15
<a href="https://agentmods.dev/skills/brettchien/katashiro/katashiro-web-task"><img src="https://agentmods.dev/badge/skills/brettchien/katashiro/katashiro-web-task.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 167 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,097 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.
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 $0.00167 $0.02097
Opus 5.5 $0.00067 $0.00839
Sonnet 5.5 $0.00033 $0.00419
Haiku 4.5 $0.00017 $0.00210

Measured 4d ago against content hash 8d623dccf91c, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

katashiro-web-task 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 4d 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.

skills/katashiro-web-task/SKILL.md · 122 lines

How it starts

The opening of the file, as written. The whole thing — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.

katashiro-web-task

Katashiro's page tools (katashiro.*) act in the user's real Chrome, as the user, with whatever they are logged into. This skill is the order of operations that gets a web task done in few calls and without surprises. The tool descriptions have the details of each tool; this is how to combine them.

Requirements

  • The current chat is a Katashiro chat (its browser tunnel is attached). Not in Discord/Slack threads or cron turns.
  • Write tools (click, type, fill_form, navigate, new_tab, switch_tab, close_tab, …) need act mode on. With act mode off you can still read, but only the active tab: switching tabs is a write.

1. See the page: snapshot first

  1. Which tab? If the user says "this page", the target is the active tab of their window. If you need another tab, katashiro.tabs, then katashiro.switch_tab by url substring (more stable than index, which shifts as tabs open and close). switch_tab only reaches tabs in the side panel's window (your window in tabs); for a tab in another window, new_tab its URL. When your conversation is in the background (§5), the user's tab is out of reach: tell them, and ask them to switch back to this conversation.
  2. katashiro.snapshot — the accessibility tree with a ref on each interactive element and a snapshotId. This is the primary way to see a page: cheap, and it gives you what to click. Scope it with selector to re-check one region.
  3. Text only? katashiro.get_text (optionally scoped by selector) is cheaper than a snapshot.
  4. "This", "what I selected" → katashiro.get_selection.
  5. katashiro.screenshot only for what text cannot tell: layout, images, charts, a canvas. It is expensive; never use it to read text or to confirm an action worked.
  6. read_dom only when you need literal markup of one element.

Action tools (click, type, navigate, fill_form, press_key, scroll, …) return the updated snapshot. Do not call snapshot again right after them.

Read the full file on GitHub · 122 lines

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. 4d ago First seen · 122 lines · 167 tokens per session scan A 8d623dccf91c

Subscribe to this mod's changes

katashiro-web-task is a skill published in the GitHub repository brettchien/katashiro (10 stars, last pushed 3d ago), licensed MIT. It adds 167 tokens to every session and 2,097 once invoked, about $0.0007 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 2026-11-02.

Browser automationacpai-agentsbrowser-automationchrome-extensionmanifest-v3mcp
Related

katashiro-web-task compared with skills from other repositories

katashiro-web-task next to 6 skills from other repositories
ModTypeStarsLast pushLicenceScanTokensWorks with
katashiro-web-taskbrettchien/katashiroSkill103d agoMITA167Claude Code, Codex
browser-automationCrawlio-app/crawlio-browserSkill82mo agoApache-2.0A68Claude Code
extractCrawlio-app/crawlio-browserSkill82mo agoApache-2.0A27Claude Code
robot-trainingCrawlio-app/crawlio-browserSkill82mo agoApache-2.0A56Claude Code
monitorCrawlio-app/crawlio-browserSkill82mo agoApache-2.0A21Claude Code
canonical-recordingCrawlio-app/crawlio-browserSkill82mo agoApache-2.0A28Claude Code
agent-browser-clisleepinginsummer/agent-browser-cliSkill6464d agoMITA32Claude Code, Codex