paper-ingest
Skill0 starsChanged 3 months agoMITView on GitHub
Normalize `_raw/` into extracted chunks, `index.md`, and `summary.md`, and register the paper in `_index/`. Use when the user asks to ingest a PDF, `sources/<id>/`, generate extracted chunks or a cited summary, or complete paper intake.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- Go-MinSeong/paper-review-servicetheir own version, first seen Nov 8, 2026
- shkodnik1917/claude-brainlabtheir own version, first seen Nov 8, 2026
- Vepricov/claude-brainlabtheir own version, first seen Nov 1, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- shkodnik1917/claude-brainlab★ 36same name
- Vepricov/claude-brainlab★ 35same name
- Go-MinSeong/paper-review-service★ 8same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- paper-extractSkill
- paper-studySkill
- paper-surveySkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 1 is distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
When a PDF exists under sources/<source-id>/original/, run the full set below.
- Very specific: line 23
If _raw/ exists, use it as the primary input
- Distinctive: line 24
If missing, ask the user to run Marker (full text).
- Very specific: line 25
Do not full-read the PDF to fill extracted/ (short checks and missing-page fills only)
- Very specific: line 28
Add block IDs per /paper-extract (one ID = one block.
- Very specific: line 30
Do not change meaning. Mark doubtful extractions with [extraction uncertain]
- Very specific: line 35
If mode is study/both, prepare study/ stubs from _templates/study-note.md and _templates/question-bank.md.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Paper ingest
When a PDF exists under sources/<source-id>/original/, run the full set below.
Extraction details: read /paper-extract when math or algorithms are broken, _raw/ is missing or the main task, or Marker / OCR / arXiv math is in play.
Templates: _templates/source-index.md, _templates/summary.md. Also _templates/study-note.md, _templates/question-bank.md, _templates/algorithm.md, _templates/repro-readme.md as mode requires.
Replace every {{...}} in copied templates (source-id, title, mode, date). Do not copy unused placeholder text into real notes.
Steps
- Confirm
source-idandmode(study/survey/both). Ask if unspecified - Confirm extraction inputs (
/paper-extract)- If
_raw/exists, use it as the primary input - If missing, ask the user to run Marker (full text); for math, use the reference priority (arXiv source first). Continue only after inputs are ready
- Do not full-read the PDF to fill
extracted/(short checks and missing-page fills only)
- If
- Normalize
_raw/into section chunksextracted/NN-<section-slug>.md- About 300–500 lines per file
- Add block IDs per
/paper-extract(one ID = one block; unique in the file) - Figures/tables: captions only; note "See original for figure"
- Do not change meaning. Mark doubtful extractions with
[extraction uncertain] - For survey/both, if math or algorithms are broken, follow
/paper-extract(arXiv source → Mathpix → page image + vision); on failure, follow that fallback and mark[extraction uncertain](never an arbitrary source)
- Generate
index.mdfrom_templates/source-index.md. Setmodefrom step 1. Keep only Artifact links that matchmode - Generate
summary.mdfrom_templates/summary.md(every claim must be cited perCLAUDE.md). Set the samemode - Register the paper in
_index/MOC.mdand_index/keywords.md. MOCstatusmust matchindex.mdfrontmatter. On first paper, delete the(none yet)row - If mode is study/both, prepare
study/stubs from_templates/study-note.mdand_templates/question-bank.md. If survey/both, preparesurvey/algorithm.mdfrom_templates/algorithm.mdandsurvey/repro/README.mdfrom_templates/repro-readme.md - Set
statustoreadyonindex.md,summary.md, and any stubs created in step 7 (unless extraction is still incomplete: thenextracting)
Done when
extracted/chunks exist via_raw/or an explicit exception path- Block IDs in each
extracted/file are unique index.md/summary.mdexist; every#^citation resolves to a block ID that exists in the target file- No
{{...}},…, or other template wording remains in generated files - Unused study/survey Artifact links were removed
- Frontmatter
modematches the confirmed mode;statusisready(orextractingif still incomplete);authors/year/venueare filled only from the paper - The paper is listed in
_index/with the samemodeandstatus - Citations point only to
extracted/(never_raw/)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

