eulinks

Personal link catalog organized by topic. Each subdirectory contains one or more *links.txt / *linki.txt files in a simple Markdown bullet format:

### Section Title
 - https://example.com/some-article
 - https://other.org/reference

Original mirrors: https://git.io/vATMf · http://bit.ly/eulinks

---

Tools

All three scripts run from inside eulinks/ and share the same file conventions:

• Only lines inside ### Header sections that have at least one URL bullet are touched.
• Lines inside ``` fenced code blocks are never modified.
• Bullet style (- vs *) is auto-detected from the section and preserved on insert.
• A trailing # 2026 marker (configurable) tags every automatically inserted line.

---

checklinks — verify URLs and remove dead ones

./checklinks                         # dry-run: report dead links across all files
./checklinks --only ai,bd_sql        # limit scan to specific subdirectories
./checklinks -a                      # apply: remove dead lines (asks for confirmation)
./checklinks -a -y                   # apply without confirmation
./checklinks --aggressive            # also treat 403/429/5xx/timeout as dead
./checklinks --report dead.txt       # write report to file instead of stdout

A line is removed only if every URL it contains is dead. Non-URL lines (headers, prose, markdown links with valid text) are never touched.

Options: --workers 24, --timeout 8.0, --retries 1.

---

insertlinks — batch-insert curated links from a JSON mapping

./insertlinks mapping.json
./insertlinks mapping.json --marker "# 2026" --dry-run

mapping.json format:

{
  "relative/path/to/file.txt": {
    "Section Header Title": [
      "https://example.com/article-1",
      "https://example.com/article-2"
    ]
  }
}

Each URL is inserted immediately after the matching header, before existing bullets. Header titles must match exactly (case-sensitive, stripped of leading/trailing whitespace). Headers not found in their file are reported as HEADERS NOT MATCHED to stderr and the script exits 1 — fix the title and re-run.

The script is not idempotent: running it twice on the same mapping inserts the URLs twice. Always run checklinks --only <dir> after inserting to verify new links are live.

---

enrichlinks — automatically find and insert new links via SearXNG

Requires the SearXNG instance from aistack to be running (docker ps | grep searxng). Check the actual host port: docker ps | grep searxng — it may be 8081, not 8080.

./enrichlinks DIR [DIR ...]                       enrich all unenriched sections
./enrichlinks DIR --dry-run                       show what would be inserted, don't write
./enrichlinks DIR --max 3                         up to 3 links per section (default: 2)
./enrichlinks DIR --context "Android Kotlin"      extra keywords appended to every query
./enrichlinks DIR --sections "Redis,Docker swarm" target specific sections (substring)
./enrichlinks DIR --no-skip-marker                re-enrich sections that already have the marker
./enrichlinks DIR --time-range none               broaden search (default: year)
./enrichlinks ai bd_sql prog_java                 loop over multiple dirs in one call

For each section that has URL bullets and no existing # 2026 marker, the script:

1. Builds a query "<section title> 2026 [--context]" and calls the SearXNG REST API.
2. Filters candidates: drops blacklisted domains (Instagram, TikTok, Facebook, Twitter/X, LinkedIn, Pinterest) and non-reference URL patterns (/shorts/, /reel/, IG post paths).
3. Checks up to 8 remaining candidates concurrently (6 workers) for HTTP 200.
4. Skips URLs already present anywhere in the file.
5. Inserts the top --max live results immediately after the header.

--context is important for directories with Polish or vague section titles — without it, SearXNG may return unrelated results. Use topic keywords that narrow the domain:

sw_android  →  --context "Android Kotlin development"
prog_python →  --context "Python programming tutorial"
ai          →  --context "AI machine learning deep learning"
sieci       →  --context "Linux networking sysadmin"

Options: --searxng URL, --timeout 8.0, --delay 1.2, --time-range year|month|week|day|none, --marker "# 2026".

Typical workflow for a new directory:

cd ~/Claude/eulinks
SEARXNG=http://localhost:8081   # adjust to actual port

# 1. Preview
./enrichlinks prog_python --searxng $SEARXNG --context "Python tutorial" --dry-run

# 2. Write
./enrichlinks prog_python --searxng $SEARXNG --context "Python tutorial"

# 3. Verify inserted links are live, remove dead ones
./checklinks --only prog_python
./checklinks --only prog_python -a -y

Manual workflow (insertlinks):

# 1. Create mapping.json with hand-picked URLs for specific sections
./insertlinks /tmp/my_mapping.json --dry-run
./insertlinks /tmp/my_mapping.json

# 2. Verify
./checklinks --only prog_java