skills/generic/create-digest/SKILL.md
Distill a document into a tight, journalist-style digest. Use when the user wants a summary, digest, or readable version of any document — specs, research notes, meeting transcripts, technical docs, PRDs, RFCs, or anything verbose that needs to be made concise and human. Also use when the user says /create-digest or asks to 'make this readable' or 'give me the highlights'.
npx skillsauth add ricky-yosh/agent-workflows create-digestInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Turn a document into something people actually want to read.
You take a source document — could be a spec, research notes, a long thread, meeting minutes, a technical RFC, anything — and produce a digest that captures everything important in a fraction of the words. The digest reads like a well-edited magazine piece: clear, direct, no filler.
Write like a journalist at a good publication. That means:
Read the source. Read the entire document. Don't skim — you need to understand what matters before you can decide what to cut.
Find the spine. What's the one thing this document is really about? Everything in your digest should connect back to that central thread. Side points get a sentence or get cut.
Write the digest. Structure it naturally — not as a mechanical extraction of sections from the original, but as a piece that flows. Use headers when they help the reader navigate, not as decoration. A 2-page spec might need zero headers. A 30-page RFC might need four.
Cut 20% more. Your first draft is too long. It always is. Read it again and find the sentences that repeat what an earlier sentence already implied. Find the qualifiers that add nothing ("somewhat", "relatively", "in most cases"). Remove them.
Save it. Write the digest to .aw/digests/ as a markdown file. Name it based on what was digested: spec-digest.md, research-digest.md, rfc-digest.md, etc. If there's already a file with that name, ask the user before overwriting.
One line in the terminal. Don't summarize the digest you just wrote — the user will read the file. Just confirm where it is:
Digest saved to
.aw/digests/[name]-digest.md
Always keep:
Usually cut:
Match the shape of the source. When the original is a sequence — tasks, steps, phases, a checklist — keep the items as a scannable list. Use the journalist voice for framing and context around the list, but don't melt individual items into paragraphs. A one-line description per item beats a paragraph that buries the same information.
The rule: if the reader will come back to this digest to check progress or find a specific item, structure wins over narrative. Lists stay lists. Tables stay tables. Sequences stay numbered.
No fixed rule. A good digest of a 1-page doc might be 3 sentences. A good digest of a 50-page RFC might be 2 pages. The right length is the shortest version that doesn't lose anything the reader needs. Aim for roughly 20-30% of the original length as a starting point, then let the content dictate.
development
Write tests for a batch of tasks BEFORE implementation. Detects the project's test framework and language, then writes tests that cover happy paths, edge cases, and failure modes. All tests should fail initially since there is no implementation yet. Part of the gf (greenfield) skill family. Use when the user says /gf-write-tests or wants to write tests before implementing a feature or task batch.
testing
Generate readable testing notes after a feature is complete. Compares spec against implementation to produce a concise list of what was built, how to test it, and any scope gaps. Writes testing-notes.md to .aw/greenfield-progress/. Part of the gf (greenfield) skill family.
testing
Save current session work to .aw/greenfield-progress/progress.txt. Captures decisions (why, not just what) that git diff cannot show. Updates tasks.json to mark completed tasks. Part of the gf (greenfield) skill family.
testing
Gate the greenfield workflow on user verification after an implementation step. Reads the implementation digest, shows verification steps, and waits for the user to confirm before advancing. Part of the gf (greenfield) skill family.