weekly-podcast-brief
Skill0 starsChanged 5 months agoMITView on GitHub
Use to turn a weekly consolidation note into a ready-to-use NotebookLM podcast pack — a clean narrative source brief plus an Audio Overview instruction+prompt — so the user can generate a "week in review" audio to listen to on a walk. Auto-invoke right after a weekly consolidation is produced (e.g. a Monday review session), or on phrases like „podcast z tygodnia", „brief na spacer", „materiał do notebooklm z tygodnia", „audio z weekly", „zrób podcast z konsolidacji". This is the podcast step of a weekly review loop — it does NOT generate the weekly consolidation itself, it consumes it. Skip when no weekly consolidation exists yet (produce that first).
In this repository4 other files it gives its agents.
- ai-workshopSkill
- case-study-builderSkill
- kb-ragSkill
- repo-publish-auditSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Skip: if no weekly consolidation exists for the period yet → produce/locate that first.
- Very specific: line 33
Read references/templates.md for the exact brief structure, the NotebookLM customize-prompt template, the cleanliness NIE-list, and the …
- Distinctive: line 68
If a number is not in the source, omit it — never fabricate connective "spoiwo".
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Weekly Podcast Brief
Overview
Each week the user produces a dense weekly consolidation note (a Monday review session). It is hard to
cognitively hold 20–30 deliverables — they need to be re-shaped into a small number of interlocking
threads the user can absorb passively while walking. This skill distills the latest weekly consolidation
into a NotebookLM podcast pack: one clean, self-contained source brief plus a customize-prompt for
NotebookLM's Audio Overview. The output is for listening, not canon — insights return through a
human review gate.
This is a manual loop, by design. A true cron/scheduled job will NOT work here: (1) the distillation needs an LLM, not a deterministic script; (2) the notes may live in a cloud-synced location invisible to the OS scheduler. So the "weekly loop" = this skill firing inside the user's natural Monday session, right after the weekly consolidation is written. No new scheduled job.
When to use
- Primary (auto): immediately after a
weekly consolidationnote is created/updated in a Monday session — produce the podcast pack as the closing step of the weekly loop. - On-demand: the user says „podcast z tygodnia", „brief na spacer", „audio z weekly", „materiał do notebooklm z tygodnia", etc.
- Skip: if no weekly consolidation exists for the period yet → produce/locate that first. This skill consumes the consolidation, never replaces it.
Workflow
Read references/templates.md for the exact brief structure, the NotebookLM customize-prompt template,
the cleanliness NIE-list, and the file-naming convention. Keep that detail in the reference, not here.
- Locate the source. Find the newest weekly consolidation:
<vault>/AI-wiki/sessions/<DATE> — weekly consolidation.mdList the folder, pick the latest dated file, read it in full. Optionally skim any current live-status notes for framing (tone, what is "now"). - Distill into 4–6 causal threads, not a list. This is the whole point: cognitive compression.
Collapse the week's deliverables into a handful of threads that cause one another (typical spine:
funnel → portfolio → ladder/strategy → rigor-layer/tools → core tension). End on the real
tension/decision, not a tidy wrap-up. Use the thread skeleton in
references/templates.md. - Apply the cleanliness gate (hard, privacy). The source goes to NotebookLM = Google. Strip every
confidential client/pilot name and every cost figure; describe those threads generically. Public job
postings the user applied to are acceptable but used sparingly. See the NIE-list in
references/templates.md. - Write the source brief. Self-contained (NotebookLM does not know the source notes), narrative but
structured, ~2000–3500 words, third-person about the user, ending with 3 "pytania na spacer" + "cyfry
tygodnia". Save to:
<project>/prompts/<DATE> — Brief — co zrobiliśmy <okres> (źródło do podcast).md - Write the instruction + prompt file (separate file). Setup steps (paste-as-"Copied text"
workflow) + a customize prompt that walks the two hosts through exactly the threads from step 2 and
keeps the tension sharp. Save to:
<project>/prompts/<DATE> — NotebookLM podcast retro tygodnia (instrukcja + prompt).md - Report briefly. Give the user both file paths, the thread list (so they see the structure before listening), and the one-line cleanliness note. Remind: output = candidate; bring 2–3 insights back through the review gate.
Guardrails
- Do not invent facts or numbers. Everything traces to the weekly consolidation / live-status notes. If a number is not in the source, omit it — never fabricate connective "spoiwo".
- Two files, kept separate. The source brief is pasted into NotebookLM in full; the instruction/prompt lives in its own file so it is not ingested as a source.
- Reuse, do not duplicate. This skill is only the podcast step of the weekly loop. It never
regenerates the weekly consolidation and never writes to
sessions/— it writes only the two files inprompts/. - Output is candidate, not canon. NotebookLM audio is for learning on a walk; anything worth keeping re-enters the notes through the human review gate (fact-check vs the live-status notes).
References
references/templates.md— brief skeleton, customize-prompt template, cleanliness NIE-list, naming convention, and a worked example.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

