agentleFS
olorehqolorehq/olore/.claude/skills/download-docs/SKILL.md

download-docs

Skill114 starsChanged 8 months agoMITView on GitHub

Download and filter documentation from source configs. Use standalone to re-download without rebuilding artifacts.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Check for --force flag

    line 30
  2. Very specific:

    If exists and no --force flag:

    line 49
  3. Distinctive:

    If download failed, return error:

    line 78
  4. Distinctive:

    Delete files that should be removed:

    line 131
  5. Distinctive:

    Update the lock file with final file count:

    line 143
  6. Very specific:

    Update files field in olore-lock.json if filtering occurred.

    line 150

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Download Docs

Download documentation files from a source config and filter non-useful files.

Usage

/download-docs prisma@latest         # Download specific version
/download-docs prisma@latest --force  # Re-download even if exists

Input

$ARGUMENTS format: {config_name}@{version} (e.g., prisma@latest, nextjs@16.1.3)

Optional flags:

  • --force - Re-download even if already built

Execution Steps

Step 1: Parse Arguments and Load Config

Parse $ARGUMENTS:

  • Extract config_name and version from {config_name}@{version}
  • Check for --force flag

Load config file:

cat vault/configs/{config_name}.json

Extract:

  • name - Package name
  • _source.type - Source type (github or url)
  • _source.repo - GitHub repo (if github type)
  • versions.{version}.ref - Git ref for the version

Step 2: Check if Already Built

test -f vault/packages/{config_name}/{version}/olore-lock.json && echo "EXISTS" || echo "NOT_FOUND"

If exists and no --force flag:

skip: {config_name}@{version} already built (use --force to rebuild)

Return early.

Step 3: Download Documentation

For GitHub sources:

bash -c 'source .claude/skills/build-docs/scripts/github.sh && download_from_github "vault/configs/{config_name}.json" "vault/packages" "{version}"'

For URL sources:

bash -c 'source .claude/skills/build-docs/scripts/url.sh && download_from_urls "vault/configs/{config_name}.json" "vault/packages" "{version}"'

With --force: pass "true" (or "--force") as the 4th argument, otherwise an already-built version is skipped:

bash -c 'source .claude/skills/build-docs/scripts/github.sh && download_from_github "vault/configs/{config_name}.json" "vault/packages" "{version}" true'

(Same for download_from_urls.)

Verify download succeeded:

test -f vault/packages/{config_name}/{version}/olore-lock.json && echo "OK" || echo "FAILED"

If download failed, return error:

error: {config_name}@{version} download failed

Step 4: Count Files and Determine Tier

file_count=$(find vault/packages/{config_name}/{version}/contents -type f \( -name "*.md" -o -name "*.mdx" \) | wc -l)
total_size=$(du -sk vault/packages/{config_name}/{version}/contents | cut -f1)
TierCriteria
1< 30 files AND < 500KB
230-100 files OR 500KB-2MB
3> 100 files OR > 2MB

Step 5: Filter Files (GitHub sources with >50 files only)

Skip this step if:

  • Source type is url
  • File count is <= 50

For GitHub sources with >50 files:

  1. For each file, read only the first 20-30 lines to make filtering decisions:

    • Frontmatter/metadata (YAML block)
    • Title and first heading
    • Opening paragraph describing the content

    This is sufficient to identify file type (changelog, RFC, API doc, tutorial, etc.). Do NOT read entire files for filtering - it wastes tokens on large repos.

  2. Evaluate each file against criteria:

KEEP - Files that help developers USE the library:

  • API reference documentation
  • Usage guides and tutorials
  • Configuration and setup docs
  • Code examples and patterns
  • Integration guides
  • Troubleshooting guides

DELETE - Files NOT useful for using the library:

  • Internal development docs
  • Contribution guidelines
  • Release notes and changelogs
  • Meeting notes, RFCs, proposals
  • Marketing and landing pages
  • Duplicate or stub files
  • Auto-generated index files with no content
  1. Delete files that should be removed:
rm "vault/packages/{config_name}/{version}/contents/{path_to_delete}"
  1. Remove empty directories:
find vault/packages/{config_name}/{version}/contents -type d -empty -delete

Step 6: Update olore-lock.json

Update the lock file with final file count:

# Get final count
final_count=$(find vault/packages/{config_name}/{version}/contents -type f \( -name "*.md" -o -name "*.mdx" \) | wc -l)

Update files field in olore-lock.json if filtering occurred.

Step 7: Return Summary

Return ONLY a brief summary:

done: {config_name}@{version} downloaded, {final_count} files{filtered_info}, tier {tier}

Examples:

  • done: prisma@latest downloaded, 312 files (126 filtered), tier 3
  • done: zod@latest downloaded, 17 files, tier 1

Outputs

  • vault/packages/{config_name}/{version}/contents/ - Downloaded documentation files
  • vault/packages/{config_name}/{version}/olore-lock.json - Build metadata

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.