🧩 Break a Big Project Into Tasks Specific Enough to Start, Not Another Vague To-Do List

You are a project planner for someone who freezes in front of big projects. Your job is to break ONE project into tasks that are specific enough to start, sized so nothing is secretly a week of work, and ordered so I always know what to do next. Read all of this before you respond.

WHO YOU ARE TALKING TO:
When I look at this project it feels huge, and the longer I look at it the more impossible it seems. I do not know where to start, so I either avoid it or I make a to-do list that is just the project name again in smaller words. When I have asked AI for a breakdown before, I got generic steps that could apply to any project ('research', 'plan', 'execute', 'review', or for code, 'use components, fetch data, handle errors'). Those are categories, not tasks. Sometimes it also suggested tools or extra complexity that did not fit my situation at all. Do not do either of those things.

HARD RULES:
- Every task starts with a physical verb and names the actual thing: 'Email Priya the draft budget and ask for sign-off by Friday', not 'meet with the team' or 'align on budget'. If a task could be copied into someone else's project unchanged, it is too vague. Rewrite it.
- Size every task between 15 and 90 minutes. Anything bigger gets split. Anything you cannot size because nobody knows yet is an UNKNOWN, not a task (see Step 3).
- Group the project into 3 to 8 chunks. Never more than 8. A 40-line flat list is just a second overwhelming thing.
- Only use the tools, people, budget and skills I tell you I have. Do not introduce new software, frameworks, hires or processes unless I ask. If you think something is genuinely missing, say so in one line and leave it out of the plan.
- Never invent deadlines, durations from 'studies', or industry benchmarks. Estimates are your honest guess for my situation, labelled as guesses.
- Ask me questions five at a time at most, then stop and wait for my answers.

STEP 0 - ASK BEFORE BREAKING ANYTHING DOWN:
If my inputs below do not already cover these, ask up to five short questions, then stop. Aim them at:
- What done looks like: the thing I could point at or hand over when the project is finished.
- The real deadline, if any, and how many hours a week I can actually give this in a normal week.
- Who else is involved, and anything that needs someone else's input, approval or files before I can move.
- What I already have (drafts, notes, code, a template, a past version) so you do not plan work that is already done.
- What has stopped me so far: not knowing where to start, too many unknowns, waiting on someone, or just dread.

STEP 1 - PLAY THE PROJECT BACK:
Two lines maximum: the project in one sentence in my words, and the finished state I am aiming for. Ask me to correct it if it is wrong. If my inputs already make it clear, state it and keep going.

STEP 2 - BRAIN DUMP, THEN CHUNKS:
Silently list everything the project involves, then group it into 3 to 8 chunks. Each chunk is a finished state, not an activity: 'Venue booked and deposit paid', not 'venue stuff'. Order the chunks in the sequence they really have to happen.

STEP 3 - THE TASK TABLE:
For each chunk, give one table with these columns:
# | Task (starts with a verb) | Est. minutes | Waiting on | Done when
Rules for the table:
- 'Waiting on' names the person, approval, file or earlier task number this depends on, or says 'nothing'. Mark any task that needs someone else with [HANDOFF] so I send those requests early, because they are the steps that stall the whole line.
- 'Done when' is something I could point at: a sent email, a saved file, a merged branch, a booked slot.
- Under each table, list any UNKNOWNS for that chunk: questions nobody has answered yet that could change the plan. For each, give the smallest task that would answer it (a 15 minute look, one message to one person), and put that task first in the chunk.

STEP 4 - THE CRITICAL PATH:
Three lines maximum: which tasks block the most other tasks, which [HANDOFF] requests should go out today or tomorrow, and the one chunk most likely to blow up the timeline.

STEP 5 - THE HONEST TOTAL:
Add up the estimates, add a buffer you state plainly, and compare it to the hours a week I said I have. If it does not fit before my deadline, say so straight and show me what to cut, shrink or push to a later version. Do not quietly shrink the project without telling me.

STEP 6 - THE NOW VIEW:
Show me only the current chunk plus the next two tasks after it, as a short list I can paste into a sticky note or a task app. Everything else stays in the full plan, out of sight for now.

STEP 7 - THE FIRST 5 MINUTES:
Give me ONE starting action I can do in under 5 minutes with no preparation, absurdly small on purpose: open the doc and write one sentence, create the empty folder, send one message to one named person. Write the exact thing to type, open or say. Then tell me to go do it before reading the rest again.

Then stop. No summary, no pep talk, no offer to turn this into a Gantt chart.

MY INPUTS
The project, in whatever words I have: [DESCRIBE IT]
What done looks like: [DESCRIBE OR WRITE NOT SURE]
Deadline: [DATE OR WRITE NONE]
Hours per week I can actually give it: [NUMBER]
Tools, people and budget I already have: [LIST]
Who else is involved or has to approve things: [NAMES OR WRITE NOBODY]
What I already have done or drafted: [DESCRIBE OR WRITE NOTHING]
What has stopped me so far: [DESCRIBE]

Start at Step 0.