Smithers prompt example

prompt.md

› Can you start over? I think it will just one shot this. Really? Yeah. So there is a lawyer and then his associate. and basically I want when they have meetings for the m4a file to fall in a centralized place, where then Claude or Codex can see those files and then create a meeting summary. The meeting summary should have a summary of what happened, next steps, key takeaways, et cetera. and then from there be able to split the next steps into notion where they're basically keeping all their project management stuff.

I want you to make a UI that shows me the progress of how we are building this

Ask me 10 clarifying questions about what we are building

Read smithers.sh/llms-full.txt to understand smithers first

We are building a smithers script but I want you to actually build this script using a smithers script to build it so we can buld it really fast in parallel

• I’ll use the smithers-docs-full skill first so I’m reading the local Smithers docs in the intended way, then I’ll ask the 10 clarifying questions before changing or scaffolding anything.

• Explored └ Read SKILL.md (smithers-docs-full skill)

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

• I’ve loaded the skill wrapper. Now I’m pulling the full Smithers docs from the Smithers source so the workflow/script structure is grounded in the current CLI behavior.

• Ran smithers docs-full --latest └ code: COMMAND_NOT_FOUND message: 'docs-full' is not a command for 'smithers'. … +3 lines (ctrl + t to view transcript) smithers --help,see all available commands smithers skills add,sync outdated skills

• Searching the web

• Searched the web for https://smithers.sh/llms-full.txt

• Searching the web

• Searched the web for '## CLI' in https://smithers.sh/llms-full.txt

• Searching the web

• Searched the web for https://smithers.sh/llms-full.txt

• Searching the web

• Searched the web for https://smithers.sh/llms-full.txt

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

• The local smithers docs-full wrapper is stale, so I used the live smithers.sh/llms-full.txt doc directly. Key constraints I’m carrying forward: use bunx smithers-orchestrator ..., put workflow code under .smithers/workflows, and use the Gateway/custom UI surface for live progress.

• Ran pwd && rg --files -g '!node_modules' -g '!.db' -g '!.sqlite' └ /Users/williamcory/trying

› use only codex medium so it runs fast

• Ran bunx smithers-orchestrator --help └ smithers@0.26.1 — Durable AI workflow orchestrator. Run, monitor, and manage workflow executions.

… +74 lines (ctrl + t to view transcript)
  --token-offset <n>                  Skip first n tokens of output
  --version                           Show version

• Ran bunx smithers-orchestrator --version └ 0.26.1

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────

• Read the Smithers docs at https://smithers.sh/llms-full.txt and checked local Smithers: bunx smithers-orchestrator is available at 0.26.1. I’ll use only Codex with medium effort for the Smithers agents.

10 clarifying questions:

1. Where will the .m4a meeting files land: a local folder, Google Drive, Dropbox, iCloud, S3, or something else?
2. Should the workflow poll/watch that folder automatically, or should someone manually trigger Smithers for a specific file?
3. Do you already have transcription handled, or should this workflow transcribe .m4a files too?
4. If transcription is needed, do you want OpenAI Whisper/API, local transcription, or another provider?
5. What should the meeting summary format include exactly: summary, attendees, decisions, next steps, deadlines, risks, key quotes, open questions?
6. How should next steps map into Notion: one task per action item, grouped under a meeting page, added to an existing database, or both?
7. Do you already have the Notion database schema and API integration token, or should I scaffold placeholders for those?
8. Should tasks be assigned to the lawyer, associate, or inferred from the meeting transcript?
9. What UI do you want first: a Smithers custom run UI showing build/run progress, or an app-style dashboard for ongoing meeting processing?
10. Should this be built as a reusable Smithers workflow only, or should I also include a small CLI/app wrapper for nontechnical use?