Web Access

Community

Ce-Legend

web-access

External-world access layer for Codex. MUST use when the user asks to look up, verify, open, browse, search, cite, inspect, download, upload, operate a website, check a current fact, read a specific URL, compare sources, interact with a dynamic page, or work inside a logged-in browser session. Also use whenever the answer may depend on information beyond model memory: current prices, policies, laws, schedules, product docs, API docs, company or people changes, news, releases, rankings, or page state. Do not under-trigger: answering these tasks from memory risks stale or unsupported output. This skill turns web work into evidence-driven escalation: search for discovery, open primary sources for proof, use browser automation for rendered or interactive pages, and use local Chrome CDP only when the user's real browser session is necessary.

Overview

Publisher	Ce-Legend
Repository	`ce-legend-skills`
Skill name	`web-access`
Stars	54
Forks	28
Bundled files	6
License	MIT
Links	Skill source Repository

Markdown instructions
A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.
Works with any LLM
AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.
6 bundled files
Scripts, templates, and references the model can read while it works. Files are read-only and never executed.
Open source
Published by Ce-Legend on GitHub. Read the source before you install it.

Installation

Install the Web Access AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
https://github.com/Ce-Legend/ce-legend-skills/tree/main/skills/web-access
Enable the skill in any chat where you want it available.

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills

git clone --depth 1 https://github.com/Ce-Legend/ce-legend-skills.git /tmp/ce-legend-skills
mkdir -p .claude/skills
cp -r /tmp/ce-legend-skills/skills/web-access .claude/skills/web-access

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Web Access in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Web Access on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Web Access is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

Loaded only when it is needed
The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.
Switch models mid-chat
Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Web Access

Use this skill as Codex's external-world access layer. It closes the gap between model memory and live reality: current facts change, pages render through JavaScript, important evidence hides behind interaction, and some workflows only exist inside a logged-in browser session.

Treat web work as an evidence pipeline. Define the needed result, choose the lightest credible access layer, preserve source trails, escalate only when the page demands it, and stop when the requested outcome is verified.

What This Skill Protects Against

stale answers from model memory when the facts may have changed
weak research based on search snippets instead of opened sources
losing source provenance, publication dates, timestamps, or page state
overusing browser automation when a direct source read is enough
underusing browser automation when content is rendered, interactive, or login-bound
treating a clicked button, upload, download, or form submission as successful before verifying the resulting state

Core Workflow

Define the success condition. Decide whether the task needs a current fact, a primary-source answer, a completed web action, a downloaded artifact, a screenshot, or browser state verification.
Start with the lightest reliable layer. Use search and direct page reads for factual lookup. Use browser automation only when content is dynamic, interactive, gated by UI state, or depends on upload/download flows.
Escalate by evidence. If a page is incomplete, blocked, login-sensitive, JavaScript-rendered, or behaving differently from raw HTML, move to a browser layer instead of repeating the same failed fetch.
Return verifiable output. For research, cite sources and prefer primary material. For browser actions, report the exact completed state and any remaining uncertainty.

Tool Selection

Prefer this order unless the user request clearly starts at a higher layer:

Built-in web/search tools: current facts, source discovery, page reading, finance/weather/sports lookups, and citation-backed answers.
Playwright or in-app browser tools: clicking, typing, DOM inspection, screenshots, uploads, downloads, and JavaScript-heavy pages.
Direct HTTP tools such as curl: status codes, headers, meta tags, raw HTML, JSON, JSON-LD, robots, sitemap, and minimal page probes.
Local Chrome CDP fallback: tasks requiring the user's Chrome login state, regular browser profile behavior, or pages that fail in isolated browser contexts.

Verification Standard

Treat search results as discovery, not proof.
Prefer official pages, primary documents, source announcements, original datasets, and directly opened URLs.
Use secondary reporting only when primary sources are unavailable or insufficient, and say so.
For unstable facts such as prices, laws, schedules, product specs, company roles, package docs, or recent events, verify freshness before answering.
Keep extracted evidence narrow: titles, dates, snippets, URLs, status, selectors, and the smallest relevant page passages.
When completing a web action, verify the post-action state through page text, URL, status, saved file, confirmation view, or another concrete signal.

Browser Strategy

Inspect visible state and DOM structure before acting.
Prefer extracting existing page data over replaying unnecessary UI steps.
Open separate tabs only for independent targets, and keep tab ownership clear.
For login-sensitive tasks, never export cookies, passwords, tokens, verification codes, or private session data.
For upload/download flows, verify the resulting page state or saved file instead of assuming success after a click.

Local Chrome CDP

Use the CDP fallback only when needed. First check dependencies:

bash
bash "${CODEX_HOME:-$HOME/.codex}/skills/web-access/scripts/check-deps.sh"

If CDP is required and not already running, start it in a long-running terminal session:

bash
bash "${CODEX_HOME:-$HOME/.codex}/skills/web-access/scripts/check-deps.sh" --start

Then read references/cdp-api.md for local endpoints and request patterns.

Site Patterns

Domain-specific notes may live under references/site-patterns/. When a target domain is clear, check for a matching file before exploring from scratch. To find aliases:

bash
bash "${CODEX_HOME:-$HOME/.codex}/skills/web-access/scripts/match-site.sh" "<domain or user request>"

Treat site notes as hints. Update them only with verified, repeatable behavior.

Resource Loading

Read references/cdp-api.md only when using local Chrome CDP.
Read references/site-patterns/<domain>.md only when working on that domain.
Run scripts from scripts/ instead of retyping their logic.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

agents/openai.yaml1 KB
references/cdp-api.md3 KB
references/site-patterns/cninfo.com.cn.md3 KB
scripts/cdp-proxy.mjs17 KB
scripts/check-deps.sh1 KB
scripts/match-site.sh1 KB

Frequently asked questions

What does the Web Access AI skill do?

Why use Web Access on TypingMind?

Because you install it once and use it with any model. Web Access is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Web Access in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Ce-Legend/ce-legend-skills/tree/main/skills/web-access. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Web Access?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Web Access?

As many as you like. As long as a model supports skills, you can use Web Access with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Web Access AI skill free?

Yes. It is published on GitHub by Ce-Legend under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Openclaw Repair Sweep

openclaw

OrganizationPopular

Run scoped OpenClaw issue/PR repair campaigns: coordinate workers, prove root causes, and land or close verified work under the requested authority.

Openclaw Test Performance

openclaw

OrganizationPopular

Benchmark, diagnose, and optimize OpenClaw test and plugin-suite runtime, import hotspots, CPU/RSS, heap growth, and slow coverage paths.

Openclaw Testing

Choose proportional OpenClaw tests and checks, diagnose failures, and route environment-sensitive or release proof to its owner.

Release Openclaw Ci

Run, watch, debug, and summarize OpenClaw full release CI, release checks, live provider gates, install/update proofs, and release-secret preflights.

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

Web Access

What This Skill Protects Against

Core Workflow

Tool Selection

Verification Standard

Browser Strategy

Local Chrome CDP

Site Patterns

Resource Loading

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now