| name | all-tags-maintenance |
| description | Rebuild ALL-TAGS.txt by scanning the vault for tag occurrences and writing an alphabetized `#Tag (count)` inventory. Use when the user asks to count tags, refresh the tag list, or rebuild ALL-TAGS.txt.
|
All Tags Maintenance
Use this skill when the tag inventory needs to be regenerated from the vault contents.
Purpose
Rebuild ALL-TAGS.txt so it reflects the current vault tag inventory.
Core Rules
- Read every note markdown file in the vault unless the user explicitly narrows the scope.
- Exclude infrastructure and tooling files unless the user asks otherwise:
.agents/, .obsidian/, .git/, skills/, scripts/, AGENTS.md, tasks.md, README.md, index.md, log.md, and ALL-TAGS.txt.
- Count literal tag occurrences, not unique pages.
- Preserve the exact tag spelling and case that appears in the source.
- Include tags from frontmatter and inline Markdown.
- Ignore false positives from code blocks, inline code, and URL fragments when possible.
- Write the results back to
ALL-TAGS.txt in the same #Tag (count) format already used.
- Sort the output alphabetically by tag text.
- Keep the file to one tag per line.
Workflow
- Scan the vault for tag occurrences, preferably by running
scripts/refresh_all_tags.py.
- Aggregate counts by exact tag text.
- Sort alphabetically.
- Rewrite
ALL-TAGS.txt in the existing format.
- Spot-check a few known tags for sanity.
- Update
tasks.md if the refresh is part of an active maintenance pass.
Counting Guidance
- Treat
#AI, #ai, and #AI-in-Education as distinct tags if they appear that way in source notes.
- Do not normalize casing unless the user explicitly asks for normalization.
- Do not rename tags just because a better controlled vocabulary exists.
- If the user wants taxonomy cleanup, use the tag-taxonomy workflow separately.
Output Expectations
When this skill runs, the expected output is:
- an updated
ALL-TAGS.txt
- a short note in
tasks.md if this was part of a broader maintenance pass
