read-claude-web-conversation

An agent skill by daymade, from daymade/claude-code-skills. Tags: analysis, api, documents, history, web.

What it does

Read or export the COMPLETE transcript of a Claude.ai web conversation — both private claude.ai/chat/... and public claude.ai/share/... links — by calling Claude.ai's internal API from inside the user's logged-in Chrome, and download its FILES too (uploads, deliverables). Use whenever the user pastes a claude.ai conversation or share link and asks to read, summarize, export, archive, or extract it — "read this Claude conversation", "导出这个网页版对话", "把这个对话拉到本地". Every naive approach fails SILENTLY: curl/WebFetch hit a Cloudflare challenge; get_page_text sees only the last message; the default API rendering collapses tool calls into placeholders (~5% of it); a share payload uses block shapes a /chat/-only renderer drops without error. Works even when the claude-in-chrome extension cannot pair (different account), via CDP or a macOS AppleScript fallback. Scope: ONLINE claude.ai. For LOCAL Claude Code sessions use read-claude-code-history; for an exported .txt/.json file use claude-export-txt-better.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add daymade/claude-code-skills --skill read-claude-web-conversation

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/daymade-claude-code/read-claude-web-conversation ~/.claude/skills/read-claude-web-conversation

Safety box score

Grade D, elevated risk: 54 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill exports full Claude.ai web conversations, including tool calls and downloadable files, by running code inside the user's logged-in Chrome and calling Claude's private API endpoints. The main risks are broad access to private conversation content and attachments, plus browser-based code execution and file writes on the user's machine.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	3 (high)	It evaluates JavaScript inside the user's page (`execute t javascript jsCode`, `Runtime.evaluate`) and runs native Python/AppleScript helpers, so it can execute code in the browser context as part of the export flow.
Network access	2 (moderate)	It makes authenticated requests to private Claude endpoints such as `/api/organizations/.../chat_conversations/...` and `/api/chat_snapshots/...`, and can download attachments from those APIs.
Filesystem reach	2 (moderate)	It reads Chrome profile state (`DevToolsActivePort` under `~/Library/Application Support/Google/Chrome` or similar) and writes exports/downloads to user-chosen paths (`open(args.output, 'w'...)`, `Path(args.out).write_text(...)`).
Secrets & credentials	3 (high)	The whole method depends on the user's logged-in Chrome session (`fetch` inherits the session cookie) and exposes private conversation history, tool outputs, and attachments from that authenticated account.
Privilege & persistence	1 (low)	It needs a one-time browser permission/setting (`Allow JavaScript from Apple Events`) and macOS Automation approval, but it does not install services, require sudo, or change OS settings itself.
Instruction hijack surface	1 (low)	The skill includes directives that shape or suppress user-facing explanation, e.g. `say nothing to the user about debugging ports` and `do not relay a toggle error you cannot attribute`, but it does not instruct the agent to follow external prompts or skip confirmations broadly.
Irreversible actions	1 (low)	It can automatically export a complete conversation and download all referenced files, but it does not send messages, publish content, or perform irreversible external actions.
Transparency	0 (none)	All shipped files are readable source text; there are no binaries, hidden blobs, or downloaded installers in the skill package.

Before installing

Read `scripts/cdp_channel.py` and `references/applescript_fallback_channel.md` if you are comfortable granting a skill access to your signed-in Chrome session.
Confirm you are OK with it reading private Claude conversations and attachments from the browser you use for Claude.ai.
If you may edit the renderer, run `uv run python scripts/selftest_fidelity.py` before trusting exports.
Try it first in a throwaway conversation and verify the output path before allowing transcript/file writes.

Rated 2026-10-19 by an automated review of 11 files from the skill's repository.

Source

Repository: daymade/claude-code-skills (all skills from this repository)
Path: daymade-claude-code/read-claude-web-conversation/SKILL.md
Branch: main
Updated: 2026-09-20

api-and-interface-design — Guides stable API and interface design. Use when designing APIs, module boundaries, or any public interface.
code-documenter — Generates, formats, and validates technical documentation — including docstrings, OpenAPI/Swagger specs, JSDoc annotations, doc portals, and user guides.
django-storages-s3 — Use when configuring Django to store static and media files on AWS S3 with django-storages.
source-driven-development — Grounds every implementation decision in official documentation.
aws-auth — Adds user authentication to web and mobile apps with Amazon Cognito (user pools and identity pools) and the AWS Amplify client auth libraries.
burpsuite-project-parser — Searches and explores Burp Suite project files (.burp) from the command line. Use when searching response headers or bodies with regex patterns.