CtrlK
BlogDocsLog inGet started
Tessl Logo

uinaf/uinaf-intake

Save and publish faithful summaries of useful public sources to the shared uinaf/intake library.

—

—
Quality

—

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

SKILL.md

name:
uinaf-intake
description:
Save, capture, or intake a useful public article, video, tweet, podcast, paper, or research source into the shared uinaf/intake library. Use for explicit $uinaf-intake requests and natural requests such as "save this", "capture this source", or "intake this" when the target is public research material. Do not use for private documents, attachments, accounting intake, internal notes, or source material that cannot be retrieved well enough to summarize faithfully.

uinaf intake

Publish a compact agent-written summary to the public uinaf/intake repository. The Markdown entry is canonical; intake.uinaf.dev is generated from it.

Boundary

Only intake public source material. Never publish credentials, private context, personal data, paid/private source text, screenshots, media, full articles, or full transcripts.

If the source cannot be retrieved and the user has not supplied enough of it to write an accurate summary, stop. Do not create a hollow bookmark or infer the missing content.

Workflow

  1. Read the source completely enough to understand its claims and context.

  2. Find the registered uinaf/intake checkout. Prefer the workspace project registry, then ~/projects/uinaf/intake. Do not invent another permanent checkout.

  3. Run bun run intake:find -- "<source-url>" in that checkout.

    • Exit 0 prints an existing entry: improve that file instead of duplicating it.
    • Exit 1 means the normalized source is new.

  4. Read entry format, select existing tags where they fit, and write the proposed Markdown to a temporary file outside the checkout.

  5. Publish it with:

    skills/uinaf-intake/scripts/publish-entry.sh \
  "<registered-checkout>" \
  "<temporary-markdown-file>" \
  "entries/YYYY/YYYY-MM-DD-kebab-case.md"

  6. Report the resulting commit and public entry URL. If the push loses a race, the script rebases and retries without force-pushing. If rebasing reveals a duplicate source, merge the summaries deliberately and rerun.

Writing Rules

  • Summarize in your own words. Preserve important nuance and uncertainty.
  • Start with a short overview, followed by useful takeaways or structured notes.
  • Never repeat the title as a heading, and start sections at ##; the page already renders title as the only h1.
  • Attribute claims through the source URL; do not reproduce the source.
  • Use article, video, tweet, podcast, paper, or research.
  • Prefer the canonical public source URL. The site auto-embeds YouTube and shows a play-on-X control for video sources; tweets stay link-only. Agents only need the right type and source.
  • Use lowercase kebab-case tags and reuse existing tags before creating synonyms.
  • Keep the original saved date and file stem when improving an existing entry.
  • Do not add workflow fields, review states, or private annotations.

Normal single-source intake publishes directly to main. Bulk imports, schema changes, site changes, and skill changes require a pull request.

Verification

The publish script runs entry validation before committing and uses a signed commit. Do not bypass failed checks or disable signing. Never force-push main.

SKILL.md

tile.json