본문으로 건너뛰기
cccskills
무료GitHub에 공개

bdb-eventagency-skill

Use for event agency operations: client intake, project scoping, vendor management, crew coordination, production planning, pre-show logistics, and post-show wrap. Delegates technical execution to godmode-eventtech and bdbmediastorm.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(1)

  • SKILL.md15.6 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

BDB Event Agency Skill

The agency-ops layer for Hybridlabor Global — the business and production-coordination side of running an event company. This skill owns everything that happens before, around, and after the show. It does not own the technical execution itself.

Techniques adapted from talkvalue/event-agency-skills (Apache-2.0). Prose only — no upstream code, no scripts, no dependencies.


1. Routing Table

TaskRoute to
Signal flow, protocol binding, OSC/DMX, hardware limits, MCP orchestrationgodmode-eventtech
Live show architecture brainstorming, creative directionbdbmediastorm
Lighting cue execution, patch, playbackbdb-grandma3-mcp
Video clip, layer, and output controlbdb-resolume-mcp
3D scene, render, spatial buildgodmode-3d-creation
Client intake, scoping, budget, scheduling, vendor, crew, pre-show logistics, post-show wrapstay in this skill

Boundary rule: if the question is "what should this look like" or "what signal goes where", route out. If it is "who is bringing it, what did they promise, and what happens when they don't", it stays here. This skill never issues a technical execution instruction — it routes.


2. Client & Project Intake

An event request is not a booking until these are known. Anything guessed here becomes a change order later.

Required fields before committing

FieldWhy it is requiredIf missing
Date (and whether date or time is flexible)Drives every downstream deadline in this skillCannot scope. Ask first.
Venue (named, or shortlist with constraints)Venue holds all vendor access; determines power, load-in window, noise curfewCannot plan load-in or vendor sequencing
Brief — what must happen, who must be in the roomDistinguishes an event from a venue rentalAsk: "What does success look like when the last guest leaves?"
Budget range — a band, not a numberDrives the action policy: which classes are never vs criteriaDo not quote. Scope verbally only.
Technical riderAV, staging, power, network, FOH positionRoute to godmode-eventtech for feasibility
Attendance estimateDrives catering counts, staffing, registration, capacityCannot close catering or staffing
Client contact + decision authorityWho can actually say yesEvery later "we'll confirm" is a no

Scoping questions to ask before any commitment

  1. Who signs? Not "the client" — a name. The person who can approve a change order on show day.
  2. What is non-negotiable? Date, venue, budget, talent, or brand. Everything else is negotiable; know which one is not.
  3. What has been promised already, in writing, to anyone? Sponsors and speakers hold commitments that predate you.
  4. What is the failure mode we are least able to absorb? Late AV is a catastrophe; late florals are an inconvenience. Escalation urgency (§4) follows from this answer.
  5. Is there a technical rider yet? If not, that is a Phase 0 action, not a show-week action.

Scope statement rule

Produce a written scope before any deposit. It must name: date, venue, attendance band, the ten action classes with their modes (default-deny — see .aos/factory.yaml), what is explicitly out of scope, and the change-order trigger. A scope without an out-of-scope list will be expanded by the client by default.


3. Inbox Triage

Run this at the start of a working day on an active event, before a production meeting, or after time away. It produces a point-in-time snapshot, not a live feed.

Stakeholder classification

TypeKey signalsDefault tier
VendorAV, catering, security, decor, transport, staffing, rentals, print2
ClientContracting organisation, corporate domain, C-suite/VP2
SponsorActivation budget or in-kind contribution — distinct from the client even when the client also sponsors3
SpeakerBureau domain, rider, session, keynote, bio/headshot request3
VenueHotel, convention centre, outdoor site, venue coordinator, venue-employed catering manager2
InternalOwn domain, team aliases, automated notifications3

When a thread is ambiguous, default to the type with the higher production impact: Vendor > Venue > Speaker for operational threads.

Priority tiers

TierLabelResponse windowCriteria
1Immediate1–2 hoursBlocked deliverable, payment deadline within 48h, client waiting on confirmation, venue or vendor escalation, load-in unresolved within 72h of the event
2TodayBusiness hoursAdvancing information requests, assets for review, draft approvals, speaker logistics more than 72h out
3TrackingNoneFYIs, confirmations of receipt, vendor acknowledgements, threads you are CC'd on

Temporal override rules — apply before finalising tiers

These override the table above:

  • Any vendor email mentioning load-in, install, delivery window, or rider compliance is Tier 1 if the event is within 14 days.
  • Any client email containing a question in subject or body is Tier 1.
  • Any Tier 1 thread with no outbound reply in 24h+ escalates to flagged regardless of its original tier.

Rationale: the tier table is a prior. The overrides exist because the failure mode of event email is a request that looked like an FYI and was not one.

Output shape

A dated digest containing: event name, timestamp, event phase, period covered, counts per tier, then the tier-1 and tier-2 threads with the specific action each needs and who owns it. Every extracted action item must be one of:

  • [OUR ACTION] — a commitment we made, with a date
  • [AWAITING] — a commitment someone else made, with a date
  • [APPROVAL NEEDED] — a decision only the client can make

Never mix these three. An action with no owner is not an action item.


4. Vendor Management

Track existing vendor commitments and deliverables. This skill does not source vendors, negotiate contracts, or process payments.

Vendor types, lead times, and failure impact

TypeTypical lead timeFailure impact
AV / Technical2–4 weeksCritical — no AV, no event
VenueContracted months aheadCritical — venue holds all vendor access
Catering / F&B2–3 weeks (final count 72h)High — dietary and count changes cascade
Transport / Logistics1–2 weeksHigh — gear and people movement
Security1–2 weeksHigh — compliance and safety
Entertainment / TalentRider-dependentHigh — contracted, non-fungible
Decor / Floral1–2 weeksMedium — visual, not operational
Staffing Agencies1–2 weeksMedium — replaceable if flagged early
Signage / Print5–10 business daysMedium — wayfinding and branding
Rentals1 weekMedium — tables, chairs, linens

Phase-based urgency

Urgency is relative to event proximity, never absolute time. The same 48-hour silence is fine in month one and a production risk in advance week.

PhaseWindowOverdue thresholdEscalation speed
NormalT-30 and out48h no responseStandard: email → wait → follow-up
PlanningT-30 to T-824h no responseAccelerated: email → same-day follow-up
AdvanceT-7 to T-212h no responseUrgent: email → phone within 4h
Load-inT-12h no responseEmergency: phone immediately, escalate to production lead
Show dayT-01h no responseEmergency: phone + contingency activation

Item status taxonomy

StatusMeaning
OVERDUEPast deadline, no confirmation
AT RISKDeadline approaching, last contact exceeds the phase threshold
ON TRACKConfirmed, or within the normal response window
BLOCKEDWaiting on us, not the vendor

BLOCKED items are the most important row in the table and are routinely under-reported. When a vendor is waiting on our headcount, floor plan, or approval, that is our team's problem. Surface it first.

Escalation path

AttemptAV / Venue / CateringDecor / Signage / RentalsStaffing / Transport
1stFollow-up email quoting the original requestFollow-up emailFollow-up email
2ndPhone the account manager within 4hPhone next business dayPhone next business day
3rdEscalate to production lead + contingency researchSource a backup vendorSource a backup vendor

On show day: skip email entirely. Phone or in-person only. Never send a follow-up email on T-0.

Drafting rules for vendor follow-ups

  • Reference the specific original request or deliverable — not "following up on my last email".
  • State the deadline explicitly.
  • Ask for a specific confirmation ("Can you confirm delivery by 14 March?"), not an open question.
  • No guilt, no pressure. Vendors respond better to clarity than to escalation tone.
  • For a phone call, prepare: vendor name, account manager, the specific item, the deadline and why it matters, and the fallback question — "If the original plan isn't possible, what's the alternative?"

Budget and invoice context

AgeStatusAction
Within termsCurrentNone
1–30 daysNudgeFriendly reminder — assume oversight
31–60 daysFirm requestDirect ask with date and amount
60+ daysEscalationAccount lead; consider late fee or hold

Variance alerts: Venue 5% over (largest single line, small % = big dollars), Catering 10% (per-head counts fluctuate), Marketing 15% (most flexible), Contingency 0% drawn before T-14 (should not be touched), all others 10%.


5. Production Coordination

5.1 Pre-production

Site survey — non-negotiable for any first-time venue. Record: power distribution and available amperage by area, network drops and their throughput, load-in vehicle access and dock dimensions, door widths for the largest scenic element, FOH position and sightlines, noise curfew, and rigging points with their load ratings. Route the technical read to godmode-eventtech — this skill records the findings and flags blockers, it does not validate the signal plan.

Paperwork gate. Nothing loads in without: certificate of insurance (COI) from every vendor on site, venue access permits, and crew credentials. These are the most common cause of a T-1 access failure.

Run-of-show (ROS) is the single coordination artefact. One document, versioned, owned by the production lead, distributed to every vendor. It defines: load-in window and sequence, soundcheck, doors, show start, each cue block, and load-out. A conflict discovered in the ROS is cheap. A conflict discovered at the dock is not.

Speaker materials — deadline framework:

MilestoneDueDeliverables
D-30Bio (25 / 50 / 100-word variants, third person) + headshot (min 300×300px)
D-21Session abstract (75 words, attendee-focused), learning outcomes, AV requirements
D-14Final slide deck, event template applied
D-7Travel, hotel, ground transport, dietary and accessibility requirements
D-1Attendance and schedule confirmed

Learning outcomes use action verbs only — implement, apply, build, create, deploy, evaluate, identify, master. Never: explore, discuss, learn about, understand, discover, dive into, unpack.

Banned in any bio or session description: thought leader, visionary, guru, passionate about, world-class, cutting-edge, revolutionary, groundbreaking, leverage, synergize, transformative, innovative. Replace with specific numbers, named clients, measurable outcomes, concrete credentials.

5.2 Day-of flow

PhaseAnchorWhat must be true to advance
Load-inT-1 or earlierCOI and permits verified per vendor, power patched and tested, ROS sequence confirmed with the dock
SoundcheckFixed slot, never "when ready"Line check done, cue stack loaded, FOH and monitors positioned, walk-through with the client if contracted
DoorsPublished startHouse open, accessibility routes clear, front-of-house briefed on the ROS
ShowPer cue blockProduction lead confirms each block against the ROS; deviations logged, not improvised
StrikeAfter clearanceEquipment accounted for against the load-in manifest, rentals checked out before the truck leaves

The strike manifest is the load-in manifest. Count on the way in, count on the way out. Rentals left on site are invoiced at loss.

5.3 Post-show wrap

Within 72 hours of strike:

  • Debrief — what worked, what did not, what nearly failed. Blameless, specific, and it feeds the next ROS.
  • Invoice triggers — headcount actuals vs. quoted, rental shortfalls, overtime hours, damage claims. Capture while the evidence exists.
  • Thank-you and forward look — personalised, referencing what the attendee actually did, not a broadcast.
  • Asset archiving — recorded content, photography, final ROS, run sheets, and the vendor contact list with performance notes, filed where the next production can find them.
  • Vendor performance note per vendor — this is the input to §4's escalation tiers for the next event. A vendor who delivered three times running is not escalated on the first silence.

5.4 Post-event reporting

Performance tiers: Exceeded (beat goal by 10%+ on primary metrics), Met (within 10%), Missed (more than 10% below on one or more), Mixed (hit some, missed others — say which and why).

Report structure: performance summary → what worked (with evidence) → what did not (with likely cause) → key insights (each tied to a data point) → specific recommendations for the next event → sponsor ROI if applicable.

Channel attribution is inherently imperfect — most attendees meet several touchpoints before registering. Report it honestly, acknowledge mixed sources, and never over-credit a single channel.


6. Verification

  • Every event is classified into a phase (Normal / Planning / Advance / Load-In / Show Day) and that classification is stated, not assumed.
  • Urgency thresholds are applied relative to event proximity, not absolute days.
  • Every BLOCKED item names our team as owner — no BLOCKED item is left attributed to a vendor.
  • Technical execution questions were routed to godmode-eventtech / bdbmediastorm rather than answered from this skill.
  • No signal-flow, protocol, hardware-limit, or cue-execution instruction was issued from this skill.
  • COI, permits, and crew credentials are confirmed before load-in, per vendor.
  • The run-of-show is versioned, distributed, and owned by the production lead.
  • Speaker materials are checked against the D-30 / D-21 / D-14 / D-7 / D-1 framework.
  • Banned fluff words are absent from every bio, abstract, and session description.
  • On show day, no vendor follow-up was sent by email.
  • The strike manifest reconciles against the load-in manifest.
  • Invoice triggers captured within 72 hours of strike.
  • Attribution figures are presented with their acknowledged uncertainty.
  • No upstream script, Python file, or Composio tool dependency was added.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

hybridlabor-api/aos62026년 10월 16일 업데이트

hybridlabor-api의 스킬 모두 보기

이 스킬의 문제 신고