Extraction prompts

You are helping me produce a developer handoff for an in-house operations and
project "command center" we're building for Leo Marshall Creative (LMC). I am
Jen — I run operations and I'm acting as a developer on this build. Everything I
ask for from here should be delivered in HANDOFF FORMAT so a developer can load
it directly into a system:
- Narrative / instructions → Markdown (.md)
- Anything list-like (projects, tasks, gear, contacts, subscriptions) → CSV,
  with a header row and one record per line.
Be exhaustive and literal. Pull from every file and message in this thread.
Where a field is unknown, write "UNKNOWN" rather than guessing. Do not summarize
away detail — this is source data, not a recap. Confirm you understand, then wait
for my first extraction request.

From everything in this thread, produce a CSV named lmc_projects_tasks.csv of
every project and every task you can find. One row per task. Columns:

project, task, category, status, owner, due_or_timeframe, priority, blocked_by,
next_action, notes, source_file

- category examples: construction, studio buildout, cabinetry, lighting, gear,
  TV show, client work, admin, vendor/manufacturing, marketing, other
- status: not_started / in_progress / blocked / waiting / done / UNKNOWN
- owner: Leo / Jen / Kemp / vendor / UNKNOWN
- priority: high / medium / low / UNKNOWN
Include finished items too, marked done. Be exhaustive — every construction line
item, every review, every follow-up. Output only the CSV.

Leo generates ideas and follow-ups at all hours (including late-night calls with
overseas vendors). From this thread, extract every IDEA, FOLLOW-UP, or unresolved
"we should…" / "need to…" / "waiting on…" into a CSV named
lmc_ideas_followups.csv. One row each. Columns:

item, type, related_project, who_owns_it, needs_response_from, time_sensitivity,
status, notes, source_file

- type: idea / follow_up / decision_needed / waiting_on / reminder
Capture even half-formed one-off thoughts. This is the "get it out of Leo's head"
inventory. Output only the CSV.

We currently track ~90% of gear in Cheqroom with SKUs and QR labels, but small
grip items are NOT tracked (C-stands, arms, knuckles, cardellinis, etc.) because
Cheqroom charges per SKU/QR. Those items walk away. We're considering building our
own inventory system to own it and drop the subscription.

Produce a Markdown file lmc_gear_notes.md covering, from anything in this thread
and your knowledge of our operation:
- Categories of gear we own (camera, lens, lighting, grip, audio, etc.)
- Which categories are currently tracked vs. untracked
- The specific small/untracked items most prone to walking away, with rough unit
  cost where known
- How we currently print labels in-house (workflow, hardware if mentioned)
- Kit structure: what "kits" exist and what pieces belong to each
Mark unknowns as UNKNOWN. This feeds an inventory module design.

We want to cut subscription costs — especially the many professional AI tools.
(Kemp, our editor, can fill gaps.) Produce a CSV named lmc_subscriptions.csv of
every paid software/AI/service subscription you can identify from this thread.
One row each. Columns:

service, category, plan_tier, monthly_cost, annual_cost, who_uses_it,
what_it_does, local_alternative_exists, keep_cut_or_review, notes

- category: AI / creative / storage / gear-mgmt / productivity / other
- keep_cut_or_review: keep / cut / review
Where cost is unknown, write UNKNOWN. Flag any that a local/owned tool could
replace. Output only the CSV.

Produce a CSV named lmc_contacts.csv of the people, vendors, manufacturers, and
clients referenced in this thread. One row each. Columns:

name, role_or_company, relationship, timezone_or_region, what_we_work_on_together,
best_contact_method, notes

Include overseas manufacturing/vendor reps (e.g. China calls), internal team
(Leo, Jen, Kemp, Tarifa), and any clients or partners. UNKNOWN where needed.
Output only the CSV.

Produce a Markdown file lmc_studio.md describing our physical studio from this
thread: the front studio (the part we rent out), the backspace (private), the
layout/grid, any remodel/construction changes since our PeerSpace listing, and
what still needs updated photos or documentation. Note anything relevant to
rentals, buildout status, or gear storage. Mark unknowns UNKNOWN.

Finally, write lmc_ecosystem_summary.md: a one-page executive summary of the
entire LMC ecosystem right now — the active projects, the state of the studio
buildout, the TV show, gear/inventory, vendor/manufacturing relationships, and
the biggest operational pain points. Write it for a developer who is about to
build a command center to run all of it. Plain, specific, no fluff.