Agent skill

Wiki Curate

by flonat in flonat/flonat-research

Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report.

MITAuto-check: notesDocuments & Office

Install Wiki Curate

skills CLI
$ npx skills add flonat/flonat-research --skill wiki-curate -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install flonat/flonat-research wiki-curate --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/flonat/flonat-research.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/wiki-curate .claude/skills/wiki-curate && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
wiki-curate
GitHub stars
146
Token cost
~2k tokens
SKILL.md length
866 words
Files
1
Skills in repo
83
Repo updated
First seen
Licence
MIT

At a glance

Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report.

  • Works in 3 steps: Human-curated (auto_generated: false)… → Longer body beats shorter. → Alphabetically earliest slug breaks ties.
  • Assessing wiki quality before maintenance
  • SKILL.md covers When to Use, When NOT to Use, Architecture and What it checks, plus 4 more sections
  • Calls uv

What it does

Wiki Curate is an agent skill from flonat/flonat-research. Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report. Use when assessing wiki quality before maintenance. Not for promoting or merging concepts; use $wiki-grow or $wiki-merge.

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Documents & Office. The repository describes itself as: Shareable Claude Code + Codex infrastructure for PhD researchers — skills, agents, hooks, and rules for academic workflows. The licence is MIT.

When your agent uses it

  • Assessing wiki quality before maintenance

Example prompts

  • “/wiki-curate”

Requirements

  • Python 3
  • Pre-approved tools (allowed-tools): Bash, Read

Workflow steps

3 steps, taken from the first numbered list in SKILL.md.

  1. Human-curated (auto_generated: false) beats auto-generated.
  2. Longer body beats shorter.
  3. Alphabetically earliest slug breaks ties.

What it can do on your machine

Read from SKILL.md and the folder at commit da27600. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Bash
    • Read

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • uv

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use uv, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Wiki Curate loads about 2k tokens when it runs. Until then it costs about 75 tokens; SKILL.md has 866 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~75
When it runs · the whole SKILL.md, loaded when a task matches
~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Bash, Read

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from flonat/flonat-research at commit da27600, republished under its MIT licence (© flonat). 866 words, ~2,014 tokens.

Download SKILL.mdSave it as .claude/skills/wiki-curate/SKILL.md (or your agent's skills folder).
name
wiki-curate
description
Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report. Use when assessing wiki quality before maintenance. Not for promoting or merging concepts; use $wiki-grow or $wiki-merge.
allowed-tools
Bash, Read
argument-hint
(no arguments)
skill-dependencies
wiki-merge

Wiki Curate — Read-only Audit

Reads the vault wiki and produces a triage report. Never writes. Detects four classes of issue: overlap clusters (fragmentation), tag gaps (untagged concepts + orphan/near-duplicate tags), write-only concepts (zero backrefs from the corpus), and anatomy/lifecycle issues (drafts awaiting curation + curated pages that regressed the concept anatomy). Companion to wiki-grow (which adds new concepts) and wiki-merge (which collapses overlap clusters).

When to Use

  • After a wiki-grow run, to check what the new stubs introduced.
  • Before deciding which concepts to refine or merge.
  • Monthly, to keep the wiki's signal-to-noise ratio high.

When NOT to Use

  • For project knowledge folders — those are audited via knowledge-lint.
  • For atlas topics — that's research-atlas-audit.
  • For mechanical edits — this skill never writes. Use wiki-merge to collapse overlap clusters and the planned tag-inference pass in wiki-grow to backfill missing tags.

Architecture

scripts/wiki-curate-scan.py
  ├─ reads:  ~/vault/concepts/*.md
  │           ~/vault/atlas/<theme>/*.md         (for backref counts)
  │           ~/vault/books/<slug>/*.md         (for backref counts)
  │           <research-root>/<theme>/<proj>/CLAUDE.md   (for backref counts)
  └─ writes: /tmp/wiki-curate-report.md
             /tmp/wiki-curate.json

The skill is invoked by running the Python orchestrator. No Claude dispatch — the audit is pure deterministic analysis.


What it checks

A. Overlap clusters (the headline feature)

Concept slugs are tokenised (kebab parts, stopwords removed, trailing s stripped for crude singularisation). Any token shared by ≥2 concept slugs forms a cluster.

Within each cluster, the canonical winner is suggested by:

  1. Human-curated (auto_generated: false) beats auto-generated.
  2. Longer body beats shorter.
  3. Alphabetically earliest slug breaks ties.

Pair-level title-Jaccard ≥ 0.40 is flagged as 🔥 high confidence (distinct from "shared keyword" — the slugs are substantively similar).

Example output from a recent run:

### `goodhart` (size 3)
- `goodhart-resistance`  — auto, 3533 chars
- `goodhart-taxonomy`    — auto, 3902 chars
- `goodharts-law`        — curated, 3111 chars  ← canonical (suggested)

Suggested action: wiki-merge goodharts-law goodhart-taxonomy goodhart-resistance
B. Tag coverage
  • Untagged concepts — frontmatter has no tags: field. Filters on /concepts?tag=X won't surface them.
  • Orphan tags — used by exactly one concept. Either expand or retire.
  • Near-duplicate tags — pairs differing only by case / hyphen / trailing s (e.g. ml vs machine-learning, model vs models).
C. Backref health

For each concept, count:

  • Incoming wikilinks — [[slug]] references from atlas topic bodies, book chapters, and project CLAUDE.md files.
  • Corpus mentions — case-insensitive whole-word slug occurrences in the same corpus.

Concepts with both counts zero are flagged as write-only — they don't surface anywhere. Two remedies:

  1. Insert [[slug]] references in the atlas topics that should reference the concept (mechanism A from this session — body wikilinks auto-surface as chips on the topic page).
  2. Add the slug to ~/vault/concepts/.denylist and delete the file — accept the concept isn't carrying weight.
D. Anatomy & lifecycle

The status: frontmatter field is the lifecycle axis: draft (no guarantees, may be project-voiced) → curated (conforms to the concept anatomy). Status is read from frontmatter with the same fallback as atlas-workspace (auto_generated: true → draft, else curated) so pages predating the field classify correctly. The anatomy is defined in packages/research-atlas-schema/schema.md → "Concept File Schema":

  • ## In one line — required definition block (machine-anchored).
  • ## In my portfolio — the boundary below which project-voiced prose ("the project", "this paper", "we propose", …) is allowed.

Two lists are reported:

  • Drafts awaiting curation — every status: draft concept, with a ✓/— table showing which anatomy pieces already exist (In one line, In my portfolio) and how many project-voice phrases sit above the boundary. Higher counts mean more restructuring to do.
  • Curated pages violating the anatomy — pages marked curated that regressed: missing the In one line definition, or project-voice prose above the In my portfolio boundary (including a portfolio section under a non-canonical heading). These are the priority fix — a curated page is supposed to conform.

The curation action itself is a Claude-driven restructuring edit (this skill is read-only); see the triage workflow below.


Show full SKILL.md (314 more words)Show less

Output

/tmp/wiki-curate-report.md with three sections (A/B/C) and a suggested triage order. Stdout from the script:

Concepts: 14 (6 curated, 8 draft)
Overlap clusters: 0
Drafts awaiting curation: 8
Curated pages violating anatomy: 1
Untagged: 8/14
Write-only (no backrefs): 0
Report: /tmp/wiki-curate-report.md
JSON:   /tmp/wiki-curate.json

The JSON sidecar feeds the planned wiki-merge skill so it can read the suggested winner/fold-in lists without re-running the detection.


Modes

ModeInvocationBehaviour
Defaultwiki-curate (or uv run python scripts/wiki-curate-scan.py)Run all three audits, write report + JSON, print summary to stdout. Read-only.

There's no --autonomous flag here — the skill is already pure-read and idempotent. Future cron integration: chained after the Saturday 06:45 wiki-grow run, the audit can land in log/audits/wiki-curate-YYYY-MM-DD.md for a weekly health snapshot.


Triage workflow

After running:

  1. Section D regressions first — any curated page flagged as violating the anatomy is a bug: it claims curated but doesn't conform. Fix in place (add ## In one line, or rename/add the ## In my portfolio section and move project-voiced prose under it).
  2. Section A — every overlap cluster fragments knowledge. Pick a canonical winner (often the suggestion) and run wiki-merge <winner> <fold-in-slugs> to collapse. Re-run wiki-curate to confirm the cluster is gone.
  3. Section D drafts — curate each draft to the anatomy:
    • Extract a 2-4 sentence project-agnostic definition into ## In one line.
    • Keep the cited, stable knowledge in encyclopedia voice.
    • Move all project-voiced prose under ## In my portfolio.
    • Verify references: keys are canonical Paperpile keys (rekey hand-minted lookalikes).
    • Flip status: draft → status: curated in frontmatter and set last_updated.
  4. Section B — untagged concepts won't surface in the tag filter. Until the planned wiki-grow tag-inference pass lands, add tags manually for the auto-promoted stubs you care about.
  5. Section C — write-only concepts. Decide per-concept: link from a topic, denylist, or accept.

Cross-References

SkillRelationship
wiki-growWrites the auto-promoted concepts this skill audits.
wiki-mergeActs on the overlap clusters this skill finds (planned — not yet built).
compile-knowledgeUpstream of wiki-grow; produces the project knowledge articles that get promoted.
research-atlas-auditAudits atlas topics; complementary lens.
knowledge-lintAudits per-project knowledge folders; complementary lens.

© flonat, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/wiki-curate of flonat/flonat-research.

Open the folder on GitHubat commit da27600

Compare with similar skills

Wiki Curate next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Wiki Curate compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Wiki Curate this skillflonat/flonat-research146—~2kAutomated safety check: NotesMIT
Markdown Article FormatterJimLiu/baoyu-skills26k6 repos~3.5kAutomated safety check: PassMIT
MarkitdownImCa0/just-laws78214 repos~3.2kAutomated safety check: NotesMIT
Obsidian MarkdownAtmosphere/atmosphere3.8k20 repos~1.3kAutomated safety check: PassApache-2.0
DOCXrvdbreemen/OTGW-firmware20733 repos~4.3kAutomated safety check: PassProprietary
Gzh Designisjiamu/gzh-design-skill3.9k1 repos~2.2kAutomated safety check: PassAGPL-3.0

Similar skills

  • Markdown Article Formatter

    JimLiu/baoyu-skills

    Reformats plain text or Markdown articles with frontmatter, a title, a summary, headings, bold, lists and code blocks, and saves a separate formatted copy.

    26k GitHub starsUsed in 6 repos~3.5k tokens
    Documents & OfficeAuto-check passed
  • Markitdown

    ImCa0/just-laws

    Convert files and office documents to Markdown. An agent skill from ImCa0/just-laws.

    782 GitHub starsUsed in 14 repos~3.2k tokens
    Documents & OfficeAuto-check: notes
  • Obsidian Markdown

    Atmosphere/atmosphere

    Create and edit Obsidian Flavored Markdown with wikilinks, embeds, callouts, properties, and other Obsidian-specific syntax.

    3.8k GitHub starsUsed in 20 repos~1.3k tokens
    Documents & OfficeAuto-check passed
  • DOCX

    rvdbreemen/OTGW-firmware

    A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx files).

    207 GitHub starsUsed in 33 repos~4.3k tokens
    Documents & OfficeAuto-check passed
  • Gzh Design

    isjiamu/gzh-design-skill

    微信公众号文章排版引擎,将 Markdown 转换为可直接粘贴到公众号编辑器的 HTML。主题风格从 references/theme-index.md 注册的自定义主题库中选取,自动章节编号、关键词下划线标记、引言卡片、目录导航、代码块、图片/GIF、作者签名。支持 Markdown / Word(.docx) / PDF / 纯文本输入(非 Markdown…

    3.9k GitHub starsUsed in 1 repo~2.2k tokens
    Documents & OfficeAuto-check passed
  • Reads, creates and edits Word .docx files with python-docx, and drops to raw OOXML for tracked changes, comments and byte-exact edits.

    41k GitHub stars~2.5k tokensUpdated yesterday
    Documents & OfficeAuto-check passed

More from flonat/flonat-research

All 83 skills in this repo
  • Latex Posters

    flonat/flonat-research

    Create a large-format academic poster in LaTeX using beamerposter, tikzposter, or baposter.

    146 GitHub stars~1.5k tokensUpdated 9 days ago
    Auto-check: notes
  • Skill Creator

    flonat/flonat-research

    Create, revise, and evaluate reusable AI workflow skills, including trigger-quality tests.

    146 GitHub stars~4.4k tokensUpdated 9 days ago
    Auto-check passed
  • DOCX

    flonat/flonat-research

    Create, read, edit, or convert Microsoft Word documents while preserving professional document structure.

    146 GitHub stars~1.2k tokensUpdated 9 days ago
    Auto-check passed
  • PDF

    flonat/flonat-research

    Read, create, combine, split, rotate, OCR, watermark, secure, or extract content from PDF files.

    146 GitHub stars~488 tokensUpdated 9 days ago
    Auto-check passed
  • Init Project Orchestration

    flonat/flonat-research

    Create or migrate project-level agents, repeatable project workflows, and planning state from one client-neutral contract, then render repository-scoped adapters for both Claude Code and Codex.

    146 GitHub stars~1.6k tokensUpdated 9 days ago
    Auto-check passed
  • Pre Commit Audit

    flonat/flonat-research

    Deliver a fast pre-commit safety scan: file size, anonymity (author / affiliation strings in tex/bib), hardcoded secrets, and invisible-Unicode carriers.

    146 GitHub stars~2.8k tokensUpdated 9 days ago
    Auto-check: notes

Questions about Wiki Curate

What does Wiki Curate do?

Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report. Wiki Curate is an agent skill from flonat/flonat-research. Audit the Research Vault concept wiki read-only for fragmentation, missing tags, write-only concepts, and draft or anatomy conformance, producing a Markdown report.

When should I use Wiki Curate?

Wiki Curate fits situations like: assessing wiki quality before maintenance.

How do I install Wiki Curate in Claude Code?

Run `npx skills add flonat/flonat-research --skill wiki-curate -a claude-code`. Or copy the skill folder (skills/wiki-curate in flonat/flonat-research) into .claude/skills/wiki-curate in your project. Claude Code loads it when a task matches its description.

How do I install Wiki Curate in Codex?

Run `npx skills add flonat/flonat-research --skill wiki-curate -a codex`. Or copy the skill folder (skills/wiki-curate in flonat/flonat-research) into .agents/skills/wiki-curate in your project. Codex loads it when a task matches its description.

Can I use Wiki Curate in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add flonat/flonat-research --skill wiki-curate -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/wiki-curate, .gemini/skills/wiki-curate, .github/skills/wiki-curate and .opencode/skills/wiki-curate in your project.

What does Wiki Curate need to run?

Going by SKILL.md and its folder, Wiki Curate needs the command-line tools its instructions call (uv). Our summary lists: Python 3. Its frontmatter pre-approves these tools: Bash, Read.

Does Wiki Curate access the network?

SKILL.md contains no URLs. Its commands use uv, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Wiki Curate safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. Review the folder before installing.

What licence does Wiki Curate use?

Wiki Curate is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Wiki Curate use?

About 2k tokens (SKILL.md is roughly 8.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Wiki Curate?

Skills that share tags, products or a category with Wiki Curate: Markdown Article Formatter (JimLiu/baoyu-skills, 26k stars), Markitdown (ImCa0/just-laws, 782 stars), Obsidian Markdown (Atmosphere/atmosphere, 3.8k stars) and DOCX (rvdbreemen/OTGW-firmware, 207 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Wiki Curate?

flonat (a GitHub user) maintains it in flonat/flonat-research, which has 146 GitHub stars. The repository holds 83 skills in this directory. The repository was last updated on September 29, 2026.

Source: flonat/flonat-research on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.