Skill65 repo starsupdated yesterday

link-ingest

Link Ingest is a CLI-based workflow for importing raw files into Link knowledge bases, refreshing source pages, proposing memories, and validating wiki updates without requiring MCP. Use this skill when users need to systematically process raw data files through the ingest pipeline, check data quality, generate wiki documentation from sources, and ensure consistency across concept, entity, memory, and backlink indexes.

View source Repository: link

Install in Claude Code

Copy

git clone --depth 1 https://github.com/gowtham0992/link /tmp/link-ingest && cp -r /tmp/link-ingest/skills/link-ingest ~/.claude/skills/link-ingest

Then start a new Claude Code session; the skill loads automatically.

Definition

SKILL.md

# Link Ingest

Use `lnk ingest-status` as the source of truth. In a source checkout, replace `lnk` with `python3 link.py`. The command tells you which raw files need work and which checks must run next.

1. Inspect the ingest plan:
   ```bash
   lnk ingest-status [link-root]
   ```
2. If Link reports secret-looking values, unreadable files, or unsafe paths, stop and ask the user to fix or redact them.
3. Read only the pending raw files named by the ingest plan. Create or update one `wiki/sources/...` page per raw file, and update existing concept/entity/exploration/memory pages before creating thin duplicates.
4. Keep durable memory proposal-only until the user approves it:
   ```bash
   lnk propose-memories raw/<file> [link-root]
   ```
5. After writing wiki pages, rebuild generated indexes and validate:
   ```bash
   lnk rebuild-index [link-root]
   lnk rebuild-backlinks [link-root]
   lnk validate [link-root]
   lnk health [link-root]
   ```

Do not put raw source contents into chat unless needed for the current ingest task. Preserve source paths and provenance on generated pages.