本页保存的是公开项目资料快照，阅读过程不需要连接 GitHub。

Shared AI Chats

An archive of public Claude, Grok, and Kimi conversations, exported from their share links as plain markdown, plus the three scripts that produce it. A handful of other conversations, added by hand from sources with no public share link, live alongside them.

1,334 conversations, 15,252 messages, ~38 MB.

Directory	Source	Conversations	Messages	Longest
claude_chats/	claude.ai/share	755	7,865	282
grok_chats/	grok.com/share	519	7,210	4,308
kimi_chats/	kimi.com/share	57	171	28
aistudio_chats/	Google AI Studio (manual)	3	6	4

Layout

scripts/claude_share_export.py     exporter for claude.ai shares
scripts/grok_share_export.py       exporter for grok.com shares
scripts/kimi_share_export.py       exporter for kimi.com shares
scripts/aistudio_export.py         converter for raw Google AI Studio page copies
link_lists/claudeaisharelinks.txt  the claude.ai links this archive was built from
link_lists/grokaisharelinks.txt    the grok.com links this archive was built from
link_lists/kimiaisharelinks.txt    the kimi.com links this archive was built from
raw_google_dumps/                  raw AI Studio page copies aistudio_export.py reads
claude_chats/                      one markdown file per Claude conversation
grok_chats/                        one markdown file per Grok conversation
kimi_chats/                        one markdown file per Kimi conversation
aistudio_chats/                    one markdown file per AI Studio conversation

claude_chats/.claude-share-export.json manifest for claude_chats/
grok_chats/.grok-share-export.json     manifest for grok_chats/
kimi_chats/.kimi-share-export.json     manifest for kimi_chats/
aistudio_chats/.aistudio-export.json   manifest for aistudio_chats/

Every conversation is one markdown file, named from its title:

Adding-Kerberos-Authentication-to-Hadoop-Cluster-on-EC2.md
AI-inflection-point-and-the-future-of-automation.md
네이버-AI-모델서비스-기획-직무-기업조사-리포트.md

Titles keep their original script rather than being transliterated, so non-Latin filenames are expected. A conversation with no usable title falls back to its snapshot UUID (18754235-198d-446b-afc6-26191ea62d27.md).

File format

All four converters write the same shape: a metadata header, then alternating turns separated by a horizontal rule.

# 네이버 AI 모델/서비스 기획 직무 기업조사 리포트

- Source: https://claude.ai/share/b76ffb8c-8a9d-43fb-b02d-0990e2e2fc7c
- Author: 쏠테크
- Created: 2026-02-04
- Messages: 2

---

## Human

당신은 자기소개서 작성을 돕기 위해…

---

## Assistant

# 조사 전략
…

The --- before each turn is load-bearing. Message bodies contain their own markdown headings, so without it a ## Assistant turn marker would be indistinguishable from a heading written inside a reply.

The headers differ slightly by source: Claude and Kimi transcripts carry Author, Grok and AI Studio transcripts carry Model. Grok transcripts may also include generated-image links, attachment names, and — when exported with the relevant flags — quoted thinking steps and cited web/X sources. Kimi transcripts may include quoted Tool: blocks — web search, code execution, or other tool calls the assistant made mid-turn, rendered inline rather than gated behind a flag. AI Studio transcripts may include [Attachment: name (size)] markers for uploaded files, and omit Source entirely when the raw dump didn’t record a LINK: line.

Manifests

claude_chats/.claude-share-export.json, grok_chats/.grok-share-export.json, kimi_chats/.kimi-share-export.json, and aistudio_chats/.aistudio-export.json map each conversation to the file written for it, plus title, message count, and export timestamp. Each converter reads its own manifest to skip conversations it already has, so re-running the same input is cheap. Deleting a manifest means the next run reconverts everything it covered.

For the three share-link exporters, a manifest is looked for in the output directory and then just above it, and a transcript that has since been moved into a subdirectory is recognised by name rather than downloaded again. aistudio_export.py’s manifest keys on a hash of the raw dump’s own bytes, since dumps have no id of their own.

Adding more

Claude and Grok need curl_cffi; Kimi needs requests:

pip install -r requirements.txt

Put share links in a file, one per line — blank lines and # comments are ignored:

# new batch
https://claude.ai/share/b76ffb8c-8a9d-43fb-b02d-0990e2e2fc7c
https://claude.ai/share/7a9b58a6-0903-4964-8207-cbe44de743ff

Then export into the matching directory:

python3 scripts/claude_share_export.py link_lists/claudeaisharelinks.txt -o claude_chats
python3 scripts/grok_share_export.py   link_lists/grokaisharelinks.txt   -o grok_chats
python3 scripts/kimi_share_export.py   link_lists/kimiaisharelinks.txt   -o kimi_chats

Any of the three also takes bare links or ids as positional arguments, -f FILE (repeatable) instead of a positional links file, and -o defaults to the current directory. Kimi links may carry a locale segment (kimi.com/share/en/) and tracking query strings (?ra=1); both are stripped down to the bare id before fetching.

Already-archived conversations are skipped, so it’s safe to keep appending to one links file and re-run it. Use --force to re-fetch a conversation that has gained messages since it was archived — the archive is a snapshot, not a live mirror.

Optional content flags, off by default (Claude/Grok only — Kimi has none, its tool-call blocks are always included):

• --include-thinking (both) — thinking blocks, where the snapshot exposes them
• --include-sources (Grok only) — web pages and X posts cited by each answer

AI Studio (manual)

Google AI Studio prompts (aistudio.google.com/app/prompts/...) sit behind a Google login — there’s no public share API to fetch, so there’s no -links.txt workflow here. Instead:

1. Select-all-copy the chat page’s text and paste it into a new .txt file under raw_google_dumps/. Optionally prepend a LINK: line so the transcript records where it came from.
2. Convert it:

   python3 scripts/aistudio_export.py raw_google_dumps -o aistudio_chats --model "Gemini 2.5 Pro"

   aistudio_export.py strips the page’s own UI chrome (ThinkingThoughts, chevron_right, the disclaimer footer, file thumbnail blocks) and splits on the User/Model speaker labels it leaves behind. --model is recorded on every transcript in the run; there’s no way to detect it from the dump.
3. If a copy missed a reply’s text (AI Studio sometimes only copies the collapsed thinking-toggle chrome, not the response under it), the script prints a warning and writes _[empty message -- not captured in the raw copy]_ — paste the missing text back into the dump and rerun.

Already-converted dumps are skipped on rerun (matched by content hash); use --force to reconvert one you’ve edited.

Caveats

Only public shares are reachable through the three *_share_export.py scripts; there is no authenticated path, so revoked or private conversations return “not found.” The links files have been pruned to what actually exported, so a fresh run of any of them should come back clean. aistudio_chats/ is the deliberate exception — those conversations were never public, so they’re transcribed by hand from a private, logged-in session instead of fetched.

Claude’s and Grok’s share pages are empty SPA shells, so those two exporters read the underlying JSON APIs directly. Those endpoints sit behind a Cloudflare check that fingerprints TLS rather than headers, which is why they use curl_cffi (browser TLS impersonation) instead of plain requests. Kimi’s share page is server-rendered instead — the conversation JSON is embedded in a window.HYDRATION_INIT_STATE = {...} script tag, no separate API call needed — and it isn’t behind a Cloudflare gate, so kimi_share_export.py uses plain requests.

Some fidelity is lost upstream. Claude’s own share API flattens tool calls and search widgets to a literal This block is not supported on your current device yet. fence, and that placeholder is what lands in these files. Nothing in the export pipeline can recover the original content.