Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a leader needs to assign work to a marketing team member or run a cross-functional meeting: a task brief with concrete deliverables, a deadline with a time and timezone, a definition of done, constraints, and check-in points, plus an async-first meeting template. Trigger on 'assign a task', 'brief for my team', 'team brief', '1:1 brief', 'definition of done', 'meeting agenda template', or 'marketing standup agenda'.
.claude/skills/minhnv0807-64-team-brief-global/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 158% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 139% | 0% |
> A good brief means the recipient knows exactly what to make, by when, and what "done" looks like. "Make some good content" is not a brief. "Write 3 Facebook captions on the pain angle, 200 words each, submitted Wednesday before 9:00 CET" is. Thirty minutes writing a proper brief saves three hours of rework.
Read .agents/product-marketing-context-global.md and the current campaign brief if available. If missing, ask up to 4 questions:
Check capacity before assigning: how many tasks do they hold, and which deadline comes first? If they are overloaded, choose between moving the deadline, cutting the scope, or reassigning — do not stack work and hope they absorb it.
Pick the role template below. Every brief needs: context (why), deliverable (what), done criteria (when it is finished), references (a good example and an anti-example), and constraints (hard limits).
For tasks over 3 days or with multiple deliverables, set a draft checkpoint for a mid-way review. Do not discover at the deadline that the work went in the wrong direction.
Ask "any questions?", get explicit confirmation of understanding and of the deadline, and set a follow-up reminder one day before it is due. On async teams, this confirmation happens in writing in the task thread so the record exists.
Check the output against the checklist in the brief. If it misses, say which criterion it missed — no vague reactions. To approve content or ads use 62-marketing-review-global; to approve visuals use 47-design-review-global.
markdown# Task Brief — [Task name] Assigned to: [name] · Assigned by: [leader] · Date: [date] Deadline: [day, date, time + timezone] Priority: Urgent / High / Normal ## Why this task matters [2-3 sentences: which campaign or objective it serves, why the deadline matters] ## What to produce | # | Deliverable | Detailed spec | Format | Quantity | |---|-------------|---------------|--------|----------| | 1 | | | | | | 2 | | | | | **Not in scope:** [list explicitly what is out of scope] ## Definition of done - [ ] [Criterion 1 — measurable, not vague] - [ ] [Criterion 2] - [ ] Submitted to: [folder / channel / method] ## Reference material - Brand voice / brand guideline: [link] - Campaign brief: [link] - Good reference: [link + why it is good] - Anti-reference, do NOT copy: [link + why] ## Constraints - [Hard limits: banned words, claims that cannot be made, mandatory format] ## Check-in points - [Date]: submit first draft for review - [Date]: final after feedback - If blocked or unsure: raise it immediately, do not guess
Writer: task (quantity + format + angle + platform) · tone (with a concrete reference) · one-sentence key message · must include (social proof / CTA / data) · length · deadline · where to submit.
Designer: task (quantity + format + platform) · pixel dimensions · exact copy to paste · brand reference · visual reference · mood in two adjectives · deadline · file format and submission folder. For a full asset breakdown, run 41-campaign-asset-list-global.
Media buyer: task (campaign name + platform) · objective · audience (link the audience research) · creative (link the approved files) · daily budget · KPI targets (max CPA / min ROAS) · launch date · reporting cadence.
Sales / support: task · lead source · response SLA · reference script · what must be logged in the CRM · criteria for handing a lead back to marketing.
Distributed teams should default to written updates and reserve live time for decisions only. Circulate the pre-read at least 12 hours before the call so people in other timezones can comment in writing.
markdown# Meeting notes — [Topic] — [Date] Attendees: [name + function] · Chair: [name] · Notetaker: [name] Duration: [30-45 minutes] · Timezone: [reference timezone + local equivalents] ## Pre-read (circulated [X] hours before) - Results vs target: [link to dashboard or report] - Written updates from anyone who cannot attend live: [link to thread] ## Agenda (live time = decisions only) | # | Topic | Presenter | Time | |---|-------|-----------|------| | 1 | Questions on the pre-read | | 5 min | | 2 | Blockers | | 15 min | | 3 | Decisions to make | | 20 min | | 4 | Assignments for next period | | 5 min | ## Decisions made | Decision | Basis (data/insight) | Accountable | Revisit when | |----------|----------------------|-------------|--------------| ## Action items | Task | Owner | Deadline (with timezone) | Done when | |------|-------|--------------------------|-----------| ## Open questions | Question | What is needed to decide | Owner | Due |
Meeting rules: do not meet to report numbers — send them in writing beforehand. Meet only to make decisions and clear blockers. Every meeting ends with action items that have owners and deadlines; if it does not, the meeting did not need to happen. Publish the notes in the shared channel within 24 hours so people in other timezones are not dependent on being in the room.
02-campaign-brief-global: source of campaign context for writing task briefs.41-campaign-asset-list-global: the asset list handed to the designer.62-marketing-review-global: approval gate for content and ads after delivery.47-design-review-global: approval gate for visuals.65-team-performance-review-global: on-time rate and revision counts come from brief history.67-agency-vendor-brief-global: use instead when the recipient is an agency or freelancer rather than an employee.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | pass→pass | 10,384 | 7,496 | -28% | 1 | 1 | 0% | 1,644 | 3,072 | +87% | 0 | 0 | — |
case-01 | fail→pass | 19,628 | 21,571 | +10% | 1 | 1 | 0% | 3,169 | 4,808 | +52% | 0 | 0 | — |
case-02 | fail→pass | 22,989 | 20,354 | -11% | 1 | 1 | 0% | 3,237 | 4,575 | +41% | 0 | 0 | — |
case-03 | fail→pass | 29,400 | 18,467 | -37% | 1 | 1 | 0% | 4,242 | 5,074 | +20% | 0 | 0 | — |
case-04 | pass→pass | 10,355 | 20,930 | +102% | 1 | 1 | 0% | 1,777 | 4,688 | +164% | 0 | 0 | — |
case-05 | fail→fail | 21,142 | 22,886 | +8% | 1 | 1 | 0% | 3,840 | 5,926 | +54% | 0 | 0 | — |
case-06 | fail→fail | 17,650 | 19,420 | +10% | 1 | 1 | 0% | 2,954 | 5,121 | +73% | 0 | 0 | — |
case-07 | fail→pass | 8,649 | 12,471 | +44% | 1 | 1 | 0% | 1,530 | 3,954 | +158% | 0 | 0 | — |
case-08 | fail→pass | 11,634 | 14,212 | +22% | 1 | 1 | 0% | 1,826 | 4,359 | +139% | 0 | 0 | — |
case-09 | fail→pass | 9,654 | 11,386 | +18% | 1 | 1 | 0% | 1,697 | 3,831 | +126% | 0 | 0 | — |
case-10 | fail→fail | 11,839 | 13,313 | +12% | 1 | 1 | 0% | 2,037 | 4,190 | +106% | 0 | 0 | — |
case-11 | fail→pass | 14,968 | 14,291 | -5% | 1 | 1 | 0% | 2,274 | 4,249 | +87% | 0 | 0 | — |
case-12 | pass→pass | 14,401 | 13,425 | -7% | 1 | 1 | 0% | 2,251 | 4,155 | +85% | 0 | 0 | — |
case-14 | pass→pass | 10,517 | 4,909 | -53% | 1 | 1 | 0% | 1,816 | 2,749 | +51% | 0 | 0 | — |
case-15 | fail→pass | 12,659 | 9,860 | -22% | 1 | 1 | 0% | 2,011 | 3,371 | +68% | 0 | 0 | — |
case-16 | pass→pass | 9,019 | 5,150 | -43% | 1 | 1 | 0% | 1,491 | 2,689 | +80% | 0 | 0 | — |
case-17 | pass→pass | 11,231 | 9,337 | -17% | 1 | 1 | 0% | 1,929 | 3,487 | +81% | 0 | 0 | — |
case-18 | pass→pass | 16,101 | 12,200 | -24% | 1 | 1 | 0% | 2,631 | 3,922 | +49% | 0 | 0 | — |
case-19 | pass→pass | 10,644 | 12,350 | +16% | 1 | 1 | 0% | 1,853 | 4,017 | +117% | 0 | 0 | — |
case-20 | pass→pass | 8,251 | 4,205 | -49% | 1 | 1 | 0% | 1,322 | 2,673 | +102% | 0 | 0 | — |
case-21 | pass→pass | 7,524 | 4,750 | -37% | 1 | 1 | 0% | 1,230 | 2,636 | +114% | 0 | 0 | — |
case-22 | pass→pass | 14,744 | 10,937 | -26% | 1 | 1 | 0% | 2,540 | 3,759 | +48% | 0 | 0 | — |
case-23 | fail→fail | 8,553 | 8,379 | -2% | 1 | 1 | 0% | 1,562 | 3,317 | +112% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +35 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.