Daily Briefing

daily-briefing

Portfolio-level triage surface for an LPM running two or more active matters. The skill reads correspondence, matter state, and calendar data across the portfolio and produces a morning briefing structured around four primary elements: meetings, deliverables, issues, and risks. Supporting context orients the reader enough to form questions of others — it does not try to answer every question itself.

The discipline the skill encodes is cross-matter ranking. An LPM running a portfolio does not read their matters sequentially. They rank across the book — a Green matter with a client call at 10:00 today beats a Red matter in steady state. This skill makes that ranking discipline explicit and reproducible.

When to use this skill

• The user pastes correspondence spanning two or more matters and asks for a briefing
• The user asks "what do I need to know today/this week across my portfolio"
• The user invokes a scheduled Monday sweep or weekend catchup
• The user asks for a pre-09:00 triage across their active book

For a single matter zoom-in, use matter-drill-down. For an audience-facing status report on one matter, use status-report-drafter.

Input Classification (apply before mode selection)

Classify the input before selecting a mode. The cost of misclassification (running the wrong mode) is higher than the cost of the classification step.

Step 1 — Matter count. Scan inputs for matter identifiers, client names, or matter-specific context.

• If inputs span two or more matters → this skill applies. Proceed to Step 2.
• If inputs are about a single matter → Hard gate. Stop. Do not produce any briefing content from this skill. Route to matter-drill-down using the exact handoff response specified in the Boundary section below. This is not a suggestion or a preference. Producing a portfolio-framed document on a single-matter input — including any hybrid labelled "Mode 1 — Portfolio briefing (single-matter input)" or similar — is a skill failure, not a graceful degradation. The pedantic response is the correct response.

Step 2 — Timeframe. Identify the timeframe the user has specified or implied.

• "Monday sweep" or "weekend catchup" → Mode 3
• "Last X hours/days" or a specific date range → Mode 4
• No timeframe specified → Mode 1 (default)
• "Drill into [matter]" or "focus on [matter]" → route to matter-drill-down

Step 3 — Input tagging. If the user has tagged inputs by source ([FROM LC], [INTERNAL], [CLIENT]), use those tags to route attribution in the output. Do not strip them.

Step 4 — Matter silence detection. After confirming multi-matter input, check each matter on the portfolio roster against correspondence provided.

• For each silent matter (on roster, no correspondence): note the matter as silent at this step.
• Do not treat roster entries as substantive input. A roster tells you the matter exists, its RAG, partner, LPM, fee model, and WIP. It does not tell you what is happening on the matter this week.
• For silent matters, the only permitted content in the briefing is: (a) one-line mini-briefing stating "no substantive update this period's inputs"; (b) the matter's roster facts (lead partner, LPM, RAG, phase) if explicitly needed for a handoff or partner attention reference; (c) inclusion in the Information Gap Flag in the Summary if three or more matters are silent.
• A calendar entry for a silent matter does not convert it to active status. Record the meeting in the MEETINGS section but hold the silence discipline everywhere else in the briefing.
• If three or more matters on the portfolio roster are silent, this is a portfolio-level signal — see Summary section instructions.

Before Starting Any Mode

Before producing any briefing, check:

Portfolio roster. Inputs must identify the matters in the portfolio. Look for a roster (matter names + identifiers + lead partner + lead LPM + RAG) at the top of the input, or infer from matter-specific correspondence. If no roster is provided and cannot be inferred, ask once: "What matters are in scope for this briefing?" Do not proceed without a roster.

Timeframe. State the timeframe explicitly in the output header. Inferred from user language or inputs, not assumed.

Owner. State the LPM's name in the output header. If not provided, use a placeholder [LPM NAME] — do not assume.

Identifiers. The briefing references multiple matters. Each matter mini-briefing includes matter name and matter number in its heading. Matter numbers use placeholder [M-NUMBER] if not provided in inputs.

Boundary

This skill produces a portfolio-level triage surface for the LPM's own consumption. It does not:

• Produce status reports for distribution (use status-report-drafter)
• Produce single-matter zoom-in (use matter-drill-down)
• Update RAID logs (use risk-and-issues-manager)
• Draft scope change memos (use scope-change-controller)
• Produce budget variance analysis (use budget-and-fee-manager)

The briefing surfaces triggers that invoke those skills. It does not do their work.

If the user pastes single-matter content and asks for a briefing, stop. Do not produce any content beyond the routing response. The response is literally:

> "Single-matter input detected (matter: [matter name]). The daily-briefing skill produces portfolio-level views across two or more active matters. Run matter-drill-down on [matter] for a focused single-matter working view. No briefing produced from daily-briefing."

Do not produce a "matter-level briefing structured to the template where possible". Do not produce a "Mode 1 — Portfolio briefing (single-matter input)" hybrid. Do not produce "a useful output rather than being pedantic". The pedantic response is the correct response. Producing a portfolio-framed document on a single matter is a failure mode, not a graceful alternative. The template is designed for cross-matter ranking; a single-matter input cannot be ranked, and cross-matter patterns cannot be assessed from one matter. Reshaping the template for single-matter use produces a document that looks like a briefing and fails at the primary job of the skill.

Operating Modes

Mode 1 — Portfolio briefing (default)

Fires when: User requests a briefing across the portfolio with no specific timeframe, or when inputs span multiple matters without timeframe context.

Input: Correspondence, call notes, matter baseline, and calendar data (if available) across the portfolio.

Output: Portfolio briefing using the standard template (below).

Mode 2 — Single-matter routing

Fires when: User requests a zoom-in from portfolio context — "drill into [matter]", "focus on [matter]" — but has not invoked matter-drill-down directly.

Handoff rule: If the user wants a working view on a single matter, route to matter-drill-down. Do not produce a single-matter briefing from this skill.

"You want the working view on [matter]. Invoke matter-drill-down for the full view. If you want a one-paragraph position summary of that matter within today's portfolio briefing, I can include it in a Mode 1 output."

Mode 3 — Monday sweep

Fires when: User asks for "Monday sweep", "weekend catchup", "what landed over the weekend", or the monday-sweep scheduled Routine invokes the skill.

Input: Correspondence spanning Friday 17:00 through Monday 07:00 (or equivalent weekend window), plus matter baseline. Calendar past window extended to Friday 17:00 to cover weekend meetings.

Output: Standard briefing template (below) with one addition:

• Weekend developments section (items that landed between Friday close and Monday open)

Mode 4 — Ad-hoc timeframe

Fires when: User specifies a timeframe — "last 48 hours", "since Wednesday", "this morning".

Input: Correspondence bounded to the specified window.

Output: Standard briefing template scoped to the timeframe. Calendar retrieval uses standard windows (past 7 days, week ahead) regardless of correspondence window. Flag explicitly where matter state precedes the timeframe and remains relevant.

Standard Template (Modes 1, 3, and 4)

Produce the template sections in this exact order. Do not reorder. Do not omit. If a section has no content, state "No items this period" and move on.

DAILY BRIEFING — [LPM Name]
Prepared: [Date and time]
Mode: [1 / 3 / 4 — name]
Portfolio: [Matter count — short names]
Timeframe: [Start through End]
LPM ownership: [Which matters user leads vs oversees]

SUMMARY
[3–5 lines. Cross-matter framing, not sequential matter summaries.
Name the portfolio-level signal in one sentence. Identify the one or
two items that shape the day. State what needs attention before 09:00.
Where a meeting today is the primary constraint on how the rest of
the briefing reads, name it here — e.g. "Client call on Meridian
at 10:00 is today's primary constraint."

Information Gap Flag when applicable: if three or more matters on
the portfolio roster are silent in the input, name this explicitly
in the Summary as a portfolio-level signal. Example:
"Three of six matters (Aldwych, Nexus, Hartwick) are silent in this
period's inputs — briefing covers the three matters with
correspondence received. Confirm status on the silent three before
acting on this briefing alone."
The purpose is to alert the reader that the briefing is partial,
not to fabricate content to fill the gap.]

[Mode 3 only:]
WEEKEND DEVELOPMENTS
- [Item with matter, source, and one-line summary]
- [...]

MEETINGS — TODAY AND THIS WEEK
Upcoming matter-relevant calls and meetings for the week ahead,
ranked by imminence. Past meetings (last 7 days; Mode 3: since
Friday 17:00) noted where no follow-up correspondence has been
identified — for the LPM's awareness, not as a directive.

Upcoming:
| # | When | Matter | Meeting | Attendees | Notes |
|---|---|---|---|---|---|
| 1 | [Day, time] | [...] | [...] | [...] | [...] |

Past — no follow-up identified (soft flag only):
| Matter | Meeting | Date | Note |
|---|---|---|---|

If no calendar data is available: state "Calendar not retrieved
this session — paste calendar entries or connect M365 Calendar
to populate." Do not invent meetings. Do not infer meetings
from correspondence references. If calendar is empty or
unavailable, this section states so and stops.

DELIVERABLES — DUE TODAY OR THIS WEEK
Ranked across the portfolio by urgency and impact, not within matters.
Maximum 8 items. If more than 8 genuinely qualify, flag capacity
pressure in the SUMMARY and rank the top 8.

| # | Matter | Deliverable | Owner | By when | Why this matters |
|---|---|---|---|---|---|
| 1 | [...] | [...] | [...] | [...] | [...] |

ISSUES — CURRENT
Problems requiring action. Each with status and enough context for
the reader to act or to form questions.

| Matter | Issue | Status | Context for action |
|---|---|---|---|

RISKS — PLAUSIBLE FUTURE
Problems that may materialise. Each with probability/timing signal
and mitigation context.

| Matter | Risk | Signal | Mitigation context |
|---|---|---|---|

CROSS-MATTER PATTERNS
Signals that appear only when matters are viewed together. Name the
dynamic, not just the items. Examples: partner load pressure, LPM
load pressure, fee-structure exposure wave, shared-client overlap,
regulatory-driven disruption. If no patterns present, state
"No cross-matter patterns identified this period".

PARTNER ATTENTION — PORTFOLIO VIEW
Items requiring a lead partner's decision, response, or action.
Referenced by deliverables or issues section number, not duplicated.

| Matter | Partner | Item | Urgency | Reference |
|---|---|---|---|---|

MATTER MINI-BRIEFINGS
Length varies with matter weight. A matter in steady state gets
1–2 lines. A matter in active flux gets a paragraph. Do not
normalise length.

### [Matter 1 name] ([Matter number])
RAG · Phase · Lead Partner · Lead LPM · Fee model · WIP/Budget
[1–5 lines: current position, what is shaping the week, watch items.]

### [Matter 2 name] ([Matter number])
[...]

READING LIST — WHAT TO OPEN FIRST
Triage judgment. Max 5 items. Each item is something that would
most change the reader's day if read first. Not a summary of
content elsewhere in the briefing. Read times included.

| # | Source | Matter | Why this first | Read time |
|---|---|---|---|---|

HANDOFFS FLAGGED
Single-matter skills to invoke as follow-through. Each row names
the matter and the specific trigger.

| Skill | Matter | Trigger | Priority |
|---|---|---|---|

End of briefing.

Output rule: Produce from available information. Use placeholders for unknowns (e.g. [DATE TBC], [LPM NAME]). Do not withhold pending complete data. Do not ask questions before producing. Do not end with "want me to produce this as a .docx?" — produce the .docx.

Attribution rule: When a named individual in the inputs has produced substantive analysis or a recommendation (financial variance call, LC silence escalation judgment, scope framing recommendation), attribute it by name in the relevant section. Do not generalise to "the team" or "the matter".

Domain Knowledge

Cross-matter ranking discipline

Portfolio briefings are not matter reports stacked end to end. An LPM does not read matter by matter — they rank across the book. A Green matter with a client call at 10:00 today genuinely beats a Red matter in steady state. RAG is a trailing indicator of matter health; meeting imminence and deadline urgency are the leading indicators of today's action requirement. The two are different axes.

In the Deliverables section, rank across matters. Do not group by matter.

Calendar and meetings

Meeting imminence as a ranking signal. A matter with a client call or partner decision meeting today shapes how the rest of the briefing reads. Name the meeting in the SUMMARY when it is the day's primary constraint. The MEETINGS section surfaces it for planning; the SUMMARY contextualises it so the reader knows what is shaping the day before reading anything else.

Past meeting gap detection — judgment, not compliance. When a past meeting appears in the calendar record and no follow-up correspondence is identifiable, flag it softly in the past-meetings table. The question is not "was there a follow-up?" as a rule — it is "was this the type of meeting where a follow-up would be expected?" Apply this distinction:

• Client calls with substantive agenda → soft flag if no follow-up email, action note, or instruction identified
• Partner decision meetings → soft flag if no decision record or follow-on instruction identified
• LC calls → soft flag if no instruction email or written confirmation follows
• Internal team check-ins → judgment call; routine standups do not require follow-up; a review meeting where decisions were expected does
• Admin or scheduling calls → no flag expected

The flag is for the LPM's awareness only. Use the language "no follow-up correspondence identified" — not "follow-up overdue" or "action required". The LPM decides whether to chase. Do not escalate a past meeting to the ISSUES table solely because no follow-up was found; independent evidence of a problem is required to promote it beyond the soft flag.

Calendar matching to matters. Use the shortest distinctive keyword from each matter name as the search term — "Project Meridian" → search for "Meridian". Match calendar entries to matters by subject line keyword, meeting organiser, or attendee list (partner name, LC name, matter-specific contacts). Flag any entry that appears matter-relevant but cannot be matched to a specific matter — the LPM may know the context.

A calendar entry for a silent matter. Record the meeting in the MEETINGS upcoming table. A scheduled call on a quiet matter is a legitimate meeting requiring preparation. It does not promote the matter to active status in Deliverables, Issues, Risks, or Mini-Briefings. The silence discipline holds everywhere excep

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: legalopsconsulting
• Source: legalopsconsulting/lpm-skills
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.