| name | briefing |
| description | Write a one-page briefing or standup update — Headline, What's New, Why It Matters, What's Next / Asks — for a recurring audience that already holds context and needs the delta fast. Use for a terse status or situational update, not a narrative summary. Anti-trigger; for a longer standalone summary of a larger document written for a first-time reader use exec-summary, for a released-version history use changelog. |
| argument-hint | <the status update to write, and the period it covers> |
briefing
Produces a one-page briefing: a terse status or standup update for a
recurring audience — a standup, a sync, a regular stakeholder update — who
already hold the backdrop and need the delta and the next action, fast. It is
harness-native: no named external body (ISO/NISO/ANSI/APA/etc.) prescribes a
"briefing" format, so there is no domain standard to conform to here.
Pattern (one-page briefing / standup update)
- Headline — one line: the single most important thing to know right now.
- What's New — 2 to 4 bullets of the latest developments since the last
briefing. Favor the delta (changed or new items) over restating what's
already known.
- Why It Matters — one line per item: the implication or so-what. Every
"What's New" bullet is paired with one of these — there is no orphan update.
- What's Next / Asks — the next actions, owners, and any decisions or
inputs needed. Name something concrete; "continue monitoring" is not an ask.
Front matter carries a title, a date, and the period the briefing covers (since
the last update). No figures are required — this is a one-pager; add a single
status indicator only if it replaces a sentence, and render it as a Mermaid
xychart-beta, never ASCII art or an image. A footnote/source list is the only
back matter, and only when citation markers are used.
Rules
- Hard ceiling: one page. If it spills over, cut to the delta — the freshest,
most decision-relevant items win over restated context.
- Lead with the change; do not re-explain standing background the recurring
audience already holds.
- Every "What's New" bullet carries a paired "Why It Matters" line — an update
with no stated implication is incomplete.
- Exclude claims known to be false; flag uncertain or weakened ones inline
(e.g. "unconfirmed", "reportedly") rather than presenting them as settled.
- Account for everything the briefing is drawing from — compress to the delta,
but never silently drop an item; note thin or uncertain ones instead of
omitting them.
- Citations, when used, are inline markers (
[1], [2]) or bare links
resolving to a short footnote list — never raw internal identifiers in the
body prose.
Anti-triggers — do not use this genre for
- A longer, standalone summary of a larger document for a reader who does
not already hold context — that is
exec-summary: it re-establishes
background rather than assuming it, and isn't bound to a recurring cadence
or a "since last update" delta.
- A curated, versioned history of released changes — that is
changelog:
entries are anchored to release versions, not to a standup cadence.
- A single already-made technical decision with its rationale on record —
that is
adr: immutable and driver-and-outcome shaped, not a status delta.
MIF frontmatter
type: episodic — a briefing is a time-bound record tied to a specific
coverage period, not durable declarative knowledge. The L1 floor is id,
type, created + body; climb to L2 with namespace (briefing/<area>),
title, and tags, and to L3 with modified, temporal validity,
provenance, citations[] for any sourced claims, and typed
relationships[]. Gate every output with mif-validate (--level 1
minimum).
Why machine-readable
| Question an agent asks | Answered by (frontmatter) |
|---|
| What period does this update cover, and is it still current? | temporal.validFrom / temporal.ttl |
| Where did each "What's New" item come from? | citations[] |
| Who or what produced this briefing? | provenance (W3C-PROV) |
| What does this briefing follow on from or feed into? | typed relationships[] (relates-to, derived-from) |
The same document still reads as a terse human status update and projects
losslessly to JSON-LD and back — one artifact, two readers.
The L1 -> L3 climb
templates/good-l1.md — L1 floor: id, type, created + body. A
complete, valid briefing, but opaque to a machine consumer asking "is this
still current?" or "where did this come from?".templates/good.md — L3 (highest this genre supports): adds
namespace, modified, temporal validity for the coverage period,
W3C-PROV provenance, citations[] tied to the "What's New" items, and a
typed relationships[] graph (e.g. relates-to the prior period's
briefing). Validate with mif-validate --level 3.
Author at the highest level the drafting context supports (grade down
rather than fabricate). templates/bad.md shows the antipattern: "What's New"
bullets with no paired "Why It Matters" — orphan updates the reader has no way
to weigh.
