web
Skill9 starsChanged 5 months agoMITView on GitHub
Ingest a vetted web resource (Wikipedia article, blog post, tutorial, or docs page) into a credibility-scored source file — curated registries first, then WebSearch/WebFetch
13 other projects use a version of this.
Adopted by
Most recently taken up by
- G4bar/web-clitheir own version, first seen Nov 15, 2026
- tpc-pascal/skillsmiththeir own version, first seen Oct 20, 2026
- MysterionRise/ctf-kittheir own version, first seen Oct 13, 2026
- arch3rPro/ark-spacetheir own version, first seen Oct 11, 2026
- kjsa7412/claude_code_newslettertheir own version, first seen Oct 9, 2026
- jircdev/crew-plugintheir own version, first seen Oct 8, 2026
Other versions13 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 13
- MysterionRise/ctf-kit★ 12same name
- LocalKinAI/kinclaw★ 9same name
- Shad0wMazt3r/The-Scaffolding★ 3same name
- tyler-berggren/tb-claude-kit★ 2same name
- arch3rPro/ark-space★ 1same name
- Boxline-dev/examples★ 1same name
- jircdev/crew-plugin★ 1same name
- kjsa7412/claude_code_newsletter★ 1same name
- toannhu96/korean-interview-coach-skill★ 1same name
- tpc-pascal/skillsmith★ 1same name
3 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
- editSkill
- goalSkill
- notionSkill
- obsidianSkill
- retrieveSkill
- tutorSkill
- zoteroSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.
0
Common
Most of the 12 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 15
If $ARGUMENTS is empty, ask the user for a URL, a wiki "<topic>", or a search "<query>", then continue.
- Distinctive: line 56
If dropped, tell the user why (one line) and stop.
“Common” is measured: kept by at least 6 in 10 of the 12 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
Ingest a factual, verifiable, widely-used web resource into
sources/web_<slug>.md, recording a credibility score so every downstream
citation is auditable. Prefers curated registries (Wikipedia, official docs,
canonical blogs) and falls back to WebSearch/WebFetch, scoring each candidate
with resources_cli.py (a deterministic domain-tier prior) plus your own judgment
for borderline cases. Normally invoked by /curriculum as a pull step, but can be
run directly.
If $ARGUMENTS is empty, ask the user for a URL, a wiki "<topic>", or a
search "<query>", then continue.
Layout
Output goes to the personal sources/ by default; when invoked by /curriculum,
write to that course's per-course library course/<slug>/library/ instead (the
caller passes the destination directory). Either way the filename is web_<slug>.md.
Template:
.claude/Templates/web_source.md. The engine is .claude/src/resources_cli.py (stdlib; needs
no API key). The credibility keep-threshold is 0.60 by default.
Mode A — wiki "<topic>" (curated registry, preferred)
- Search Wikipedia and review the scored hits:
Each JSONL line hassh .claude/src/pyrun .claude/src/resources_cli.py --wikipedia "<topic>" -k 5title, url, snippet, score, tier. Pick the best-matching article (confirm with the user if ambiguous). - Fetch the full article content:
Returnssh .claude/src/pyrun .claude/src/resources_cli.py --wiki-page "<exact title>"title, url, summary, extract(full plain text),references,score. - Format & write (Step "Write" below) using
extractas## Content. Keep the substantive sections; you may drop "See also"/"External links" tails (note the omission). Wikipedia iscanonicaltier — it clears the threshold.
Mode B — <url> (a specific page)
- Score the domain first:
sh .claude/src/pyrun .claude/src/resources_cli.py --score "<url>" - Fetch the page with
WebFetch(returns clean Markdown). While reading, note the signals that refine credibility: a named author/byline, a references/citations section, and the publication year. - Decide (see Credibility decision below). If kept, format & write. If dropped, tell the user why (one line) and stop.
Mode C — search "<query>" (discovery)
- Search with
WebSearchfor the query (optionally seed from curated hubs:sh .claude/src/pyrun .claude/src/resources_cli.py --hubs "<area>"). - Score each candidate's domain with
--score <url> <url> .... - Present the top 3–5 as a list labelled
tier · score · why(highest first), and ask which to ingest. For each chosen URL, continue as Mode B from step 2.
Credibility decision
For each candidate combine the deterministic score with the signals you observed:
tier == "blocked"→ never ingest.- score ≥ 0.60 → keep.
- score in [0.45, 0.60) → escalate: judge from the fetched content whether it's authoritative, factually grounded, and widely-recommended (named expert author, cites primary sources, official/maintained). Keep only if yes; record your one-line reason.
- score < 0.45 → drop unless the user explicitly insists.
Record the final score, tier, and a one-line rationale in frontmatter. Never
ingest content you can't attribute to a real, identifiable source.
Write
- Get today's date:
date +%Y-%m-%d. - Slugify the title to snake_case →
<dest>/web_<slug>.md(<dest>=sources/by default, orcourse/<slug>/library/when/curriculuminvoked this skill). If it exists, ask before overwriting. - Write following
.claude/Templates/web_source.md:type: web_source,title,url,retrieved: <today>.credibility:block withscore,tier,rationale.concepts_mentioned:the key concepts the page covers (for retrieval/links).source_links:the canonical URL.## Content: the cleaned objective text (headings, lists, code, LaTeX preserved; nav/ads/boilerplate stripped). Faithful — no personal input, no invented facts; note any large omissions.
Wrap up
Report: the file written, its credibility score/tier, concepts extracted, and
any candidates dropped (one-line reason each). If invoked directly, offer to run
/collate <topic> to fold this source into a paper/concept page.
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.

