| name | knowledge-brief-writing |
| description | Use when Adrian or another research-oriented teammate needs to turn accumulated findings, repo knowledge, or source-backed research into clearly structured, human-readable Markdown briefings, summaries, or analysis notes that teammates can skim and trust. |
Knowledge brief writing
Use this skill when the hard part is no longer finding information, but turning what is already known into a document that other people can read quickly, understand easily, and act on confidently.
This includes:
- turning research notes into a clean Markdown briefing
- summarizing accumulated knowledge from multiple findings without losing the thread
- writing analysis documents that are easy to skim, cite, and revisit later
- producing sectioned summaries for architecture, product, or ecosystem topics
- translating analyst-style notes into human-readable team documents
Read these first
- the source notes, URLs, or documents being summarized
analysis/ai-team-context-strategy.md
analysis/concepts/overview.md
- nearby
analysis/**/* documents on the same topic
- any architecture or product note the summary is meant to support
Workflow
1. Decide what kind of document this is
Before writing, identify the output shape:
- short topic explainer
- focused research note
- architecture briefing
- comparison summary
- decision-support memo
- knowledge handoff for another teammate
Do not start drafting before you know what the reader should walk away understanding.
1.5. Write the one-sentence takeaway first
Before drafting the body, write one internal sentence that answers:
- what did the research actually show?
- why does it matter?
- what should the reader remember five minutes later?
If you cannot write that sentence clearly, the document is still too muddy.
2. Extract the real signal first
Separate the material into:
- durable facts
- evidence or citations
- informed inference
- open questions
- recommended next moves
If everything feels equally important, the document is not ready to be written yet.
3. Choose a structure that helps humans skim
Prefer strong headings, short sections, and visible hierarchy.
A good default structure is:
- Scope — what this document covers
- Key points — the main takeaways up front
- What we know — durable facts or observations
- What it means — implications for
ai-team
- Open questions — uncertainty worth tracking
- Recommended next move — what to do with the knowledge
- Sources — the URLs, files, or notes behind the summary
Use this structure by default for research notes and analysis documents.
Use a different structure only when it clearly improves comprehension.
When useful, start from templates/analysis-brief-template.md and adapt it rather than improvising the whole structure from scratch.
3.5. Optimize for straight-to-the-point reading
Assume the reader will scan the top of the document before deciding whether to keep reading.
That means:
- the first section should tell them what the note is about
- the
Key points section should surface the answer early
- the most important implication should appear before background detail
- lower-value context should move downward, not upward
Do not make the reader traverse setup, chronology, or scene-setting before they reach the conclusion.
4. Write for the teammate who did not do the research
Assume the reader is smart but busy.
Write so they can understand:
- the topic
- the most important findings
- why those findings matter
- how certain or uncertain the conclusion is
- what follow-up is worth doing
Do not make the reader reconstruct the argument from a pile of bullets.
5. Keep the prose human-readable
Prefer:
- a short opening paragraph that says the main point plainly
- short paragraphs
- concrete wording
- direct claims
- explicit transitions between sections
- bullets where comparison or scanability matters
Avoid:
- generic analyst filler
- buzzword-heavy summaries
- giant undifferentiated paragraphs
- mixing facts, guesses, and recommendations without labels
- writing the document in the order the research happened instead of the order a reader needs to understand it
If a sentence does not help the reader understand the answer faster, simplify it or cut it.
5.5. Keep the brief focused
Every section should earn its place.
Trim or rewrite anything that does not help the reader do one of these:
- understand the subject
- grasp the key finding
- see why it matters
- decide what to do next
If a detail is interesting but does not change the conclusion, shorten it or move it lower.
5.6. Use a clear writing standard
Before finishing, check the draft against this standard:
- clear — a smart teammate can understand it on one read
- focused — the main point is visible early and repeated only when useful
- structured — headings and bullets help scanning instead of fragmenting the logic
- grounded — claims trace back to evidence, notes, or sources
- human — the prose sounds natural, not like stitched-together analyst paste
6. End with something useful
A strong summary should leave the team with one of these:
- a clearer shared understanding
- a concrete recommendation
- a reusable explainer
- a precise list of questions worth answering next
If the ending does not help someone decide, route, or learn, tighten it.
Working rules
- lead with the important point instead of burying it in the middle
- start with the conclusion, then support it
- use the same default section order unless there is a clear reason not to
- prefer clarity over completeness when the extra detail does not change the conclusion
- distinguish evidence from interpretation
- make headings informative enough that someone can skim the outline alone
- keep Markdown clean and reusable for future linking and citation
- when summarizing multiple sources, synthesize them into one coherent explanation instead of writing source-by-source sludge
- when the topic is strategic, include the implication for
ai-team explicitly rather than expecting the reader to infer it
- prefer one strong sentence over three vague ones
- rewrite until the document sounds like a thoughtful teammate, not an automated report generator
Successful outcome
- the document is easy to skim and easy to trust
- teammates can understand the subject without reading all original sources
- the structure makes the key points obvious
- the writing sounds human, clear, and intentional
- the main takeaway is visible early and remains stable throughout the document
- the knowledge becomes reusable instead of trapped in raw notes or chat history