edit
Skill9 starsChanged 5 months agoMITView on GitHub
Co-edit a paper or concept page in place — action inline review comments, follow chat instructions, or build on the user's own edits — grounded in the page's cited sources, preserving frontmatter and wikilinks
18 other projects use a version of this.
Adopted by
Most recently taken up by
- vincentventalon/claude-code-video-editing-skilltheir own version, first seen Nov 15, 2026
- vincentventalon/my-claude-skillstheir own version, first seen Nov 15, 2026
- Nbill27/academic-writing-skillstheir own version, first seen Nov 15, 2026
- EdwardAF-IT/claude-skillstheir own version, first seen Nov 1, 2026
- Erickuby/Social-Media-Autoposting-With-Claude-Codetheir own version, first seen Oct 25, 2026
- PKU-DAIR/AutoScitheir own version, first seen Oct 25, 2026
Used with
Projects with it also use daily-arxiv, exp-eval and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
LambenthanSkilldaily-arxiv
每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新
Also in 5 projects that use this
LambenthanSkillexp-eval
实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges
Also in 5 projects that use this
skyllwtSkillexp-pilot-eval
Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.
Also in 4 projects that use this
skyllwtSkillexp-pilot-run
Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.
Also in 4 projects that use this
phanghonghaoSkillposter
HTML/CSS A4 海报生成。从沙龙/活动 Markdown 文件生成科技风 A4 海报 PDF/PNG。Use when user mentions: poster, 海报, 海报生成, make poster, generate poster, 活动海报, 沙龙海报.
Also in 4 projects that use this
- +6 more from Lambenthan
- +1 more from skyllwt
Other versions18 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 18
- PrimeIntellect-ai/prime-agent★ 23ksame name
- DAIR-AutoResearch/AutoSci★ 1.7ksame name
- PKU-DAIR/AutoSci★ 1.7ksame name
- skyllwt/AutoSci★ 1.7ksame name
- skyllwt/OmegaWiki★ 1.7ksame name
- Lambenthan/empiricalwiki★ 85same name
- KingGyuSuh/codex-image-in-cc★ 35same name
- howells/fiction★ 23same name
- Erickuby/Social-Media-Autoposting-With-Claude-Code★ 17same name
- gagoar/iceberg★ 10same name
8 more not listed
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 repository14 other files it gives its agents.
- .claude/AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- alphaxivSkill
- askSkill
- autoexploreSkill
- collateSkill
- curriculumSkill
- goalSkill
- notionSkill
- obsidianSkill
- retrieveSkill
- tutorSkill
- webSkill
- zoteroSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 6 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 16 edited versions we read keep these.
6
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 21
If $ARGUMENTS is empty, ask the user which page to work on, then continue.
- Very specific: line 28
Load its sources: Read the page's sources: frontmatter list — these are the files it was synthesised from.
- Distinctive: line 44
If a request needs information not present in the sources, do not invent it.
- Distinctive: line 48
If a request is ambiguous, ask the user instead of guessing.
- Distinctive: line 54
Tutor re-test reminder: After confirming, check whether the edited page belongs to a learning course and, if so, respect the user's chosen …
- Distinctive: line 64
Preserve Obsidian/Markdown syntax exactly: Treat these as literal content — never reformat, normalise, or rewrite them unless a request …
- Very specific: line 72
Write math in LaTeX: When a request adds or revises math, use LaTeX (inline $ ...
- Distinctive: line 85
After a successful edit, check whether the page is covered by a learning course and remind the user to re-test at their chosen cadence:
“Common” is measured: kept by at least 6 in 10 of the 16 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Co-edit a knowledge-base page with the user, editing it in place. This is the
collaborative refinement step for a papers/ or concepts/ page: the user drives
and you make the changes. Any factual content must come only from the
sources the page was synthesised from — if the sources don't cover what's asked,
point out the gap instead of inventing it.
The user works in three (combinable) modes:
- Inline comments — they mark requests addressed to you inline in the file,
e.g.
@claude: phrase this more precisely and less bombastically. - Chat — they tell you in conversation what to change ("tighten the overview", "add a limitations paragraph from the InstructGPT source").
- Direct edits — they've already edited the page themselves; re-read it so your changes build on their current text, and never revert their wording.
If $ARGUMENTS is empty, ask the user which page to work on, then continue.
Steps
- Read the page.
Readthe file at $ARGUMENTS (always re-read so you see the user's latest direct edits). - Load its sources. Read the page's
sources:frontmatter list — these are the files it was synthesised from. Resolve each entry relative to the page; if that path doesn't exist, locate the file by its basename withGlob(e.g.sources/**/<name>.md,notes/**/<name>.md).Readevery resolved source. These are the only admissible basis for new or changed factual content. - Gather the requests. Collect both the
@claude: ...comments in the page (inline or on their own line; each refers to the text around it unless it says otherwise) and any instructions the user gave in chat. Comments without the@claude:marker are the user's private notes — leave them untouched. If there are no comments and no chat instructions, ask the user what they'd like changed and stop. - Action each request in document order, with a surgical
Editto the passage it refers to. For an actioned@claude:comment, remove that comment in the same edit. Follow the Editing rules below.- If a request needs information not present in the sources, do not invent it. Leave any related comment in place (so it stays outstanding) and note the gap for the wrap-up — e.g. "the cited sources don't state X". Make only the part of the change the sources do support.
- If a request is ambiguous, ask the user instead of guessing.
- Confirm completion. Report what you changed (one line each) and, separately, any gaps — requests left unactioned because the sources don't cover them — so the user can add a source or revise.
- Tutor re-test reminder. After confirming, check whether the edited page belongs to a learning course and, if so, respect the user's chosen cadence (see Tutor prompt trigger below). Skip entirely if no course covers the page.
Editing rules
- Ground edits in the sources only. New or revised factual claims must be
supported by the page's cited sources (step 2). Match the page's existing citation
style (inline relative links to
../sources/.../../notes/...). Pure wording/tone edits that add no new facts are always fine. - Preserve Obsidian/Markdown syntax exactly. Treat these as literal content —
never reformat, normalise, or rewrite them unless a request explicitly asks:
- YAML frontmatter (the
---block): keep every key, value, and ordering as-is unless asked to change a specific field. [[wikilinks]],[[link|alias]], embeds![[...]], block refs^block-id, heading links[[note#heading]],#tags, and callouts (> [!note]).- LaTeX math (
$ ... $,$$ ... $$) — leave existing math untouched unless a request targets it.
- YAML frontmatter (the
- Write math in LaTeX. When a request adds or revises math, use LaTeX (inline
$ ... $, display$$ ... $$, which render in Obsidian and VS Code) — never plain text or Unicode. If asked to make something more precise or rigorous, describe it mathematically and include the critical equations/derivations, grounded in the page's sources. - Only change what's asked. Edit the specific passage and nothing else. Leave other text, spacing, and line breaks untouched. Don't restructure headings or rewrap paragraphs unless asked.
- Edit in place. Change the original file at $ARGUMENTS — never create a new file, add a suffix, or change the location.
Tutor prompt trigger — re-test reminder
After a successful edit, check whether the page is covered by a learning course and remind the user to re-test at their chosen cadence:
- Glob
course/*/goal.md. For each, check whether the edited page's slug/title appears in that course's Knowledge Audit, or the page lives under the course's ownmodules/. If no course matches, do nothing. For a matching course
<slug>, read itstutor_prompt_mode(fromgoal.md):session-end— at the very end of the wrap-up (once per session, even if several pages in the course are edited), append:Course
<slug>covers this page — remember to run/tutor <slug>this session. Track with a flag filecourse/<slug>/.session_reminded(treated as created on the first reminder this session); don't repeat the line if it already exists.after-3-edits— read the integer incourse/<slug>/.edit_count(0 if absent), increment, write it back. When it reaches 3, promptYou've edited 3 pages covered by course
<slug>this session — run/tutor <slug>? and reset.edit_countto 0.neverorunset— no prompt.
These flag files are gitignored runtime state, not knowledge-base content.
Notes
- Works from the page and its sources — no API key. The user iterates by adding more
@claude: ...comments, chatting, or editing directly, and re-running the skill; each run actions the outstanding requests (leaving only those blocked by a source gap).
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.

