QuantumBFS/sci-brain

"name": "Jin-Guo Liu"   },   "homepage": "https://github.com/QuantumBFS/sci-brain",

},   "homepage": "https://github.com/QuantumBFS/sci-brain",   "repository": "https://github.com/QuantumBFS/sci-brain",

"homepage": "https://github.com/QuantumBFS/sci-brain",

"repository": "https://github.com/QuantumBFS/sci-brain",

#!/usr/bin/env python3

[Credential-related source omitted]

#!/usr/bin/env python3

#!/usr/bin/env python3

Executable file

Executable file

class MarkdownTests(unittest.TestCase):     def test_inline_and_blocks(self):         out = report.md_to_html("A **bold** `code` [link](http://x.example/) line.\n\n- one\n- two")

def test_inline_and_blocks(self):         out = report.md_to_html("A **bold** `code` [link](http://x.example/) line.\n\n- one\n- two")         self.assertIn("<strong>bold</strong>", out)

out = report.md_to_html("A **bold** `code` [link](http://x.example/) line.\n\n- one\n- two")

self.assertIn("<strong>bold</strong>", out)         self.assertIn("<code>code</code>", out)         self.assertIn('<a href="http://x.example/">link</a>', out)

self.assertIn("<code>code</code>", out)         self.assertIn('<a href="http://x.example/">link</a>', out)         self.assertIn("<li>one</li>", out)

self.assertIn('<a href="http://x.example/">link</a>', out)

#!/usr/bin/env python3

an agent interface that captures output may not deliver it to the terminal. The helper cannot detect whether the user heard it. See the [terminal BEL definition](https://invisible-island.net/xterm/ctlseqs/ctlseqs-contents.html).

[terminal BEL definition](https://invisible-island.net/xterm/ctlseqs/ctlseqs-contents.html).

sudo apt install pulseaudio-utils sound-theme-freedesktop

Ubuntu documents [paplay](https://manpages.ubuntu.com/manpages/jammy/man1/paplay.1.html)

in `pulseaudio-utils` and includes `complete.oga` in [sound-theme-freedesktop](https://packages.ubuntu.com/questing/all/sound-theme-freedesktop/filelist). Run it as the desktop user with access to the audio session.

[sound-theme-freedesktop](https://packages.ubuntu.com/questing/all/sound-theme-freedesktop/filelist).

Executable file

Executable file

Executable file

Primary reference: [Pi session format](https://github.com/badlogic/pi-mono/blob/main/packages/coding-agent/docs/session.md).

Primary reference: [OpenCode CLI: session and export](https://opencode.ai/docs/cli/).

Primary references: [Kimi CLI export](https://github.com/MoonshotAI/kimi-cli/blob/main/docs/en/reference/kimi-command.md),

[Kimi Code sessions](https://github.com/MoonshotAI/kimi-code/blob/main/docs/en/guides/sessions.md).

Useful primary entry points: - [Gemini CLI commands](https://github.com/google-gemini/gemini-cli/blob/main/docs/reference/commands.md)

- [Gemini CLI commands](https://github.com/google-gemini/gemini-cli/blob/main/docs/reference/commands.md)

- [GitHub Copilot CLI session data](https://docs.github.com/en/copilot/how-tos/copilot-cli/use-copilot-cli/chronicle)

#!/usr/bin/env python3

#!/usr/bin/env python3

The APS Harvest server (https://harvest.aps.org/docs/harvest-api) serves the

HARVEST = "https://harvest.aps.org/v2/journals/articles/{doi}" UA = "sci-brain-download-ref/1.0 (+https://github.com/QuantumBFS/sci-brain)"

UA = "sci-brain-download-ref/1.0 (+https://github.com/QuantumBFS/sci-brain)"

def _request(doi: str, accept: str, timeout: int) -> tuple[int, bytes]:     req = urllib.request.Request(

req = urllib.request.Request(

HARVEST.format(doi=doi), headers={"Accept": accept, "User-Agent": UA})     try:         with urllib.request.urlopen(req, timeout=timeout) as r:

try:         with urllib.request.urlopen(req, timeout=timeout) as r:             return r.status, r.read()

with urllib.request.urlopen(req, timeout=timeout) as r:

link = cit.find("ext-link")         if link is not None:             d["url"] = (link.get("{http://www.w3.org/1999/xlink}href")

if link is not None:             d["url"] = (link.get("{http://www.w3.org/1999/xlink}href")                         or _txt(link))

d["url"] = (link.get("{http://www.w3.org/1999/xlink}href")

links = []         if r.get("doi"):             links.append(f"[doi:{r['doi']}](https://doi.org/{r['doi']})")

if r.get("doi"):             links.append(f"[doi:{r['doi']}](https://doi.org/{r['doi']})")         if r.get("arxiv"):

links.append(f"[doi:{r['doi']}](https://doi.org/{r['doi']})")

if r.get("arxiv"):             links.append(f"[arXiv:{r['arxiv']}](https://arxiv.org/abs/{r['arxiv']})")         if r.get("url") and not links:

links.append(f"[arXiv:{r['arxiv']}](https://arxiv.org/abs/{r['arxiv']})")

#!/usr/bin/env python3

# Strip a leading https://doi.org/ if someone left one in

#!/usr/bin/env python3

S2_FIELDS = "title,abstract,authors,year,venue,publicationVenue,journal,externalIds,citationStyles,openAccessPdf" S2_BATCH_URL = f"https://api.semanticscholar.org/graph/v1/paper/batch?fields={S2_FIELDS}"

S2_BATCH_URL = f"https://api.semanticscholar.org/graph/v1/paper/batch?fields={S2_FIELDS}"

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

req = urllib.request.Request(

for attempt in range(6):         try:             with urllib.request.urlopen(req, timeout=60) as r:

try:             with urllib.request.urlopen(req, timeout=60) as r:                 return json.loads(r.read())

with urllib.request.urlopen(req, timeout=60) as r:

return True     try:         req = urllib.request.Request(url, headers={"User-Agent": ua})

try:         req = urllib.request.Request(url, headers={"User-Agent": ua})         with urllib.request.urlopen(req, timeout=120) as r:

req = urllib.request.Request(url, headers={"User-Agent": ua})

with urllib.request.urlopen(req, timeout=120) as r:

def get_json(url: str, email: str = "") -> dict:     ua = "sci-brain/0.5 (https://github.com/QuantumBFS/sci-brain)"

ua = "sci-brain/0.5 (https://github.com/QuantumBFS/sci-brain)"

if email:         ua += f" mailto:{email}"     req = urllib.request.Request(url, headers={"User-Agent": ua})

ua += f" mailto:{email}"     req = urllib.request.Request(url, headers={"User-Agent": ua})     with urllib.request.urlopen(req, timeout=30) as response:

req = urllib.request.Request(url, headers={"User-Agent": ua})

with urllib.request.urlopen(req, timeout=30) as response:

try:         url = "https://api.crossref.org/works/" + urllib.parse.quote(doi, safe="")

url = "https://api.crossref.org/works/" + urllib.parse.quote(doi, safe="")

return []     try:         url = "https://api.unpaywall.org/v2/" + urllib.parse.quote(doi, safe="")

try:         url = "https://api.unpaywall.org/v2/" + urllib.parse.quote(doi, safe="")         data = get_json(url + "?" + urllib.parse.urlencode({"email": email}), email)

url = "https://api.unpaywall.org/v2/" + urllib.parse.quote(doi, safe="")

return "unpaywall"     arxiv = (meta.get("externalIds") or {}).get("ArXiv")     if arxiv and fetch_pdf(f"https://arxiv.org/pdf/{arxiv}", out):

arxiv = (meta.get("externalIds") or {}).get("ArXiv")     if arxiv and fetch_pdf(f"https://arxiv.org/pdf/{arxiv}", out):         return "arxiv"

if arxiv and fetch_pdf(f"https://arxiv.org/pdf/{arxiv}", out):

[Credential-related source omitted]

print(f"  ok arxiv:{aid} (cached)")                 continue             if fetch_pdf(f"https://arxiv.org/pdf/{aid}", out):

continue             if fetch_pdf(f"https://arxiv.org/pdf/{aid}", out):                 print(f"  ok arxiv:{aid}")

if fetch_pdf(f"https://arxiv.org/pdf/{aid}", out):

#!/usr/bin/env python3

#!/usr/bin/env python3

Executable file

#!/usr/bin/env python3

[Credential-related source omitted]

cached = valid_pdf(pdf)             all_cached &= cached             status = ("cached" if cached else "ok" if fetch_pdf(f"https://arxiv.org/pdf/{value}.pdf", pdf) else "miss") if kind == "arxiv" else fetch_doi_pdf(value, meta, pdf, args.em

all_cached &= cached             status = ("cached" if cached else "ok" if fetch_pdf(f"https://arxiv.org/pdf/{value}.pdf", pdf) else "miss") if kind == "arxiv" else fetch_doi_pdf(value, meta, pdf, args.email)             if status == "miss"

status = ("cached" if cached else "ok" if fetch_pdf(f"https://arxiv.org/pdf/{value}.pdf", pdf) else "miss") if kind == "arxiv" else fetch_doi_pdf(value, meta, pdf, args.email)

#!/usr/bin/env python3

arXiv papers are submitted as https://arxiv.org/pdf/<id> URL sources, so the

Service: the remote https://mineru.net/api by default, so no local GPU or

[Credential-related source omitted]

DEFAULT_API_URL = "https://mineru.net/api"

[Credential-related source omitted]

[Credential-related source omitted]

data = json.dumps(body).encode() if body is not None else None         headers = self._headers({"Content-Type": "application/json"} if data else None)         req = urllib.request.Request(self.base + path, data=data, method=method, headers=

headers = self._headers({"Content-Type": "application/json"} if data else None)         req = urllib.request.Request(self.base + path, data=data, method=method, headers=headers)         try:

req = urllib.request.Request(self.base + path, data=data, method=method, headers=headers)

try:             with urllib.request.urlopen(req, timeout=timeout) as r:                 return json.loads(r.read() or b"{}")

with urllib.request.urlopen(req, timeout=timeout) as r:

def download(self, file_id: str) -> bytes:         req = urllib.request.Request(f"{self.base}/v1/files/{file_id}/content", headers=self._headers())

req = urllib.request.Request(f"{self.base}/v1/files/{file_id}/content", headers=self._headers())

with urllib.request.urlopen(req, timeout=300) as r:

headers["Authorization"] = f"Bearer {self.key}"         req = urllib.request.Request(url, data=data, method="PUT", headers=headers)

req = urllib.request.Request(url, data=data, method="PUT", headers=headers)

with urllib.request.urlopen(req, timeout=max(120, len(data) // 20_000)):

return {"type": "local", "path": str(pdf.resolve())}         if kind == "arxiv" and "url" in self.sources():             return {"type": "url", "url": f"https://arxiv.org/pdf/{ident}"}

if kind == "arxiv" and "url" in self.sources():             return {"type": "url", "url": f"https://arxiv.org/pdf/{ident}"}         return {"type": "file_id", "file_id": self.upload(pdf)}

return {"type": "url", "url": f"https://arxiv.org/pdf/{ident}"}

[Credential-related source omitted]

[Credential-related source omitted]

#!/usr/bin/env python3

continue         meta = {             "source": f"https://arxiv.org/abs/{arxiv_id}",

meta = {             "source": f"https://arxiv.org/abs/{arxiv_id}",             "type": "arxiv",

"source": f"https://arxiv.org/abs/{arxiv_id}",

f"**Authors:** {meta['authors']}", "",                 f"**Citation:** {venue_line(s2)}", "",                 f"**arXiv:** [{arxiv_id}](https://arxiv.org/abs/{arxiv_id})"]

f"**Citation:** {venue_line(s2)}", "",                 f"**arXiv:** [{arxiv_id}](https://arxiv.org/abs/{arxiv_id})"]         if meta.get("doi"):

f"**arXiv:** [{arxiv_id}](https://arxiv.org/abs/{arxiv_id})"]

if meta.get("doi"):             body += ["", f"**DOI:** [{meta['doi']}](https://doi.org/{meta['doi']})"]         body += ["", "## Abstract", "", s2.get("abstract") or "_(abstract unavailable)_"]

body += ["", f"**DOI:** [{meta['doi']}](https://doi.org/{meta['doi']})"]

ext = s2.get("externalIds") or {}         meta = {             "source": f"https://doi.org/{doi_canon}",

meta = {             "source": f"https://doi.org/{doi_canon}",             "type": "doi",

"source": f"https://doi.org/{doi_canon}",

f"**Authors:** {meta['authors']}", "",                 f"**Citation:** {venue_line(s2)}", "",                 f"**DOI:** [{doi_canon}](https://doi.org/{doi_canon})"]

f"**Citation:** {venue_line(s2)}", "",                 f"**DOI:** [{doi_canon}](https://doi.org/{doi_canon})"]         if meta.get("arxiv_id"):

f"**DOI:** [{doi_canon}](https://doi.org/{doi_canon})"]

if meta.get("arxiv_id"):             body += ["", f"**arXiv preprint:** [{meta['arxiv_id']}](https://arxiv.org/abs/{meta['arxiv_id']})"]         body += ["", "## Abstract", "", s2.get("abstract") or "_(abstract unavailable)_"]

body += ["", f"**arXiv preprint:** [{meta['arxiv_id']}](https://arxiv.org/abs/{meta['arxiv_id']})"]

canonical = f"{owner}/{repo}"         meta = {             "source": f"https://github.com/{canonical}",

meta = {             "source": f"https://github.com/{canonical}",             "type": "github",

"source": f"https://github.com/{canonical}",

}         body = [render_frontmatter(meta), "", f"# {canonical}", "",                 f"**Source:** https://github.com/{canonical}"]

body = [render_frontmatter(meta), "", f"# {canonical}", "",                 f"**Source:** https://github.com/{canonical}"]         for cand in ("README.md", "Readme.md", "readme.md", "README.rst"):

f"**Source:** https://github.com/{canonical}"]

#!/usr/bin/env python3

[Credential-related source omitted]

[Credential-related source omitted]

#!/usr/bin/env python3

Replaces the old `sci-hub-server` MCP. A real (headless) browser is required because the mirrors serve their PDFs from behind a DDoS-Guard JavaScript challenge that plain HTTP clients (curl / urllib) cannot solve.

challenge that plain HTTP clients (curl / urllib) cannot solve.

[Credential-related source omitted]

for dom in domains:         base = f"https://{dom}"

base = f"https://{dom}"

#!/usr/bin/env python3

#!/usr/bin/env python3

Used by fetch_metadata.py (--download-arxiv-source). Downloads https://arxiv.org/e-print/<id>, detects the payload (tar / gzip / PDF-only),

https://arxiv.org/e-print/<id>, detects the payload (tar / gzip / PDF-only),

return "miss"     try:         req = urllib.request.Request(

try:         req = urllib.request.Request(             f"https://arxiv.org/e-print/{arxiv_id}",

req = urllib.request.Request(

f"https://arxiv.org/e-print/{arxiv_id}",             headers={"User-Agent": ua})         with urllib.request.urlopen(req, timeout=120) as r:

headers={"User-Agent": ua})         with urllib.request.urlopen(req, timeout=120) as r:             data = r.read()

with urllib.request.urlopen(req, timeout=120) as r:

#!/usr/bin/env python3

[Credential-related source omitted]

S2_TITLE_URL = "https://api.semanticscholar.org/graph/v1/paper/search/match"

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

[Credential-related source omitted]

request = urllib.request.Request(f"{S2_TITLE_URL}?{query}", headers=headers)

for attempt in range(6):         try:             with urllib.request.urlopen(request, timeout=60) as response:

try:             with urllib.request.urlopen(request, timeout=60) as response:                 data = json.loads(response.read())

with urllib.request.urlopen(request, timeout=60) as response:

**APS DOIs are handled automatically.** For any `10.1103/*` DOI the helper first calls the [APS Harvest API](https://harvest.aps.org/docs/harvest-api), which serves

calls the [APS Harvest API](https://harvest.aps.org/docs/harvest-api), which serves

skipped. API errors and HTML landing pages fall through to the next source. PDFs must have both a `%PDF` header and `%%EOF` trailer. A DOI miss continues to the DOI fallback below. Service contracts: [Crossref](https://www.crossref.org/docu

PDFs must have both a `%PDF` header and `%%EOF` trailer. A DOI miss continues to the DOI fallback below. Service contracts: [Crossref](https://www.crossref.org/documentation/retrieve-metadata/rest-api/), [Unpaywall](https://unpaywall.org/pr

the DOI fallback below. Service contracts: [Crossref](https://www.crossref.org/documentation/retrieve-metadata/rest-api/),

[Unpaywall](https://unpaywall.org/products/api).

[Credential-related source omitted]

[Credential-related source omitted]

Both scripts also carry [PEP 723](https://peps.python.org/pep-0723/) inline

dependency metadata, so if you happen to have [uv](https://docs.astral.sh/uv/),

`mineru_parse.py` is stdlib-only: it talks to MinerU's v1 parse API over HTTPS, so nothing needs installing and no local GPU is used. By default it calls the hosted service at `https://mineru.net/api`, which accepts anonymous jobs

so nothing needs installing and no local GPU is used. By default it calls the hosted service at `https://mineru.net/api`, which accepts anonymous jobs (about 200 pages per file; results are kept 30 days on the service). A key

hosted service at `https://mineru.net/api`, which accepts anonymous jobs

(about 200 pages per file; results are kept 30 days on the service). A key from <https://mineru.net/apiManage/token> raises the quota; the user exports it themselves, and it must never be written into the KB or the repo:

from <https://mineru.net/apiManage/token> raises the quota; the user exports it

[Credential-related source omitted]

arXiv papers are submitted as `https://arxiv.org/pdf/<id>` URLs, so the service

decides, which on mineru.net is its VLM). To run MinerU on your own hardware instead, start `mineru-api --allow-local-source` (from `uv tool install mineru`) and point the helper at it with `--api-url http://127.0.0.1:8000` or

instead, start `mineru-api --allow-local-source` (from `uv tool install mineru`) and point the helper at it with `--api-url http://127.0.0.1:8000` or `MINERU_API_URL`; it then reads the PDFs straight from `.raw/`.

and point the helper at it with `--api-url http://127.0.0.1:8000` or

A deep-thinker that **conquers one hard problem by autonomous search**, modeled on a [CDCL](https://en.wikipedia.org/wiki/Conflict-driven_clause_learning)/DPLL SAT solver. Given a

[CDCL](https://en.wikipedia.org/wiki/Conflict-driven_clause_learning)/DPLL SAT solver. Given a

#!/usr/bin/env python3

rm -f "$LOG"

> **Note:** Google Scholar actively blocks automated access — automated fetching may hit CAPTCHAs or rate limits. If scraping fails, suggest alternatives: export BibTeX manually from the Scholar profile page (Scholar → select all → export B

#!/usr/bin/env python3

- `sources/NotesOnWritingPaper12.pdf` — the original Martinis notes. - `sources/1807.01815_Ho2019_quantum-scars.md` — the model paper: Ho, Choi, Pichler & Lukin, *Periodic orbits, entanglement and quantum many-body scars in constrained mode

- von Delft's *Style Guide* online: <https://homepages.physik.uni-muenchen.de/~vondelft/JansStyleGuide.html>

LMU theoretical-physics group guide. Online: <https://homepages.physik.uni-muenchen.de/~vondelft/JansStyleGuide.html>

The slide template, gallery, layouts, and style documentation come from [GiggleLiu/sci-brain-slides](https://github.com/GiggleLiu/sci-brain-slides).

[GiggleLiu/sci-brain-slides](https://github.com/GiggleLiu/sci-brain-slides).

Read the v0.1.0 [layout guide](https://github.com/GiggleLiu/sci-brain-slides/blob/v0.1.0/docs/layout-patterns.md)

and [style tokens](https://github.com/GiggleLiu/sci-brain-slides/blob/v0.1.0/docs/style-tokens.md)

for the API and design examples matching this release. Browse the [gallery PDF](https://github.com/GiggleLiu/sci-brain-slides/blob/v0.1.0/docs/gallery.pdf) for component examples; the gallery is maintained upstream and is not included

[gallery PDF](https://github.com/GiggleLiu/sci-brain-slides/blob/v0.1.0/docs/gallery.pdf)

Report template defects and propose component changes in [the upstream repository](https://github.com/GiggleLiu/sci-brain-slides).

[the upstream repository](https://github.com/GiggleLiu/sci-brain-slides).