Project Plan
Log project parent
Tech Stack
Log notes
I am discussing this project with Claude as we go along - he created a project management list - this came out of a discussion on the goals and process of the project, and generated the below planning document - to be honest the document is not that great. It does provide some guidelines but I post it here to demonstrate what he can produce based on a project conversation
You can see that it put the build log process as phase 4, and not as an early phase - it does have the skills document build before the actual episode generation (that's a plus)
Log Code
# The Wiki Waltz — Project Plan
An experiment in podcast-as-process: the show *and* the build log are both the deliverable. Build as time allows — no marathon sessions.
---
## Phase 0 — Decisions Before Building
These are cheap to decide now and expensive to un-decide later. Nail these down first.
- [ ] **Success / kill criteria** — define what "worked," "didn't work," and "interesting but not worth continuing" look like, before any episodes exist and before attachment sets in
- [ ] **Link-selection rule** — algorithmic pick, human pick, or hybrid; how many hops of lookahead to avoid dead-ending on a stub page
- [ ] **Walk method** — walk the link chain live while writing, or pre-walk it first and write it up as-if-live
- [ ] **Publishing cadence** — daily, weekly, irregular/whenever-there's-a-good-walk, or no fixed schedule at all
- [ ] **Wikipedia attribution policy** — decide once whether/how each episode credits Wikipedia as source material (show notes line, closing credit, etc.), even though narration is original wording, not reused text
- [ ] **Where the documentation lives** — redbuoy.media as a public build log, a separate page, or something else entirely
---
## Phase 1 — Voice & Format Definition
- [ ] **Define the episode voice character** — curious, discovering-in-real-time persona; explicitly *not* Harmonia; document identity, tone, what it may/may not say (mirroring how `harmonia-voice` skill is structured)
- [ ] **Write the Wiki Waltz skills document** — episode structure/format, workflow steps, opening/closing conventions, cold-open decision (starting on the "Wiki rabbit hole" page), category-list-as-menu mechanic
- [ ] **Decide episode structure** — tight sectioned format vs. loose discovery-driven format vs. hybrid
- [ ] **Decide on audio tags** — whether ElevenLabs performance-direction tags are part of the writing workflow from the start, or layered in later as a separate pass
---
## Phase 2 — ElevenLabs Voice Experiment
- [ ] Select/design a voice profile distinct from Harmonia's existing voice
- [ ] Write a short test script (cold open is the natural candidate) — clean version, no tags
- [ ] Do a tagging pass on the test script — sparse, only at real tonal pivots
- [ ] Generate test audio on Flash model first (cheap iteration), then Standard for final quality check
- [ ] Budget check against credit pool (~15,500 characters/episode-length ≈ 7 episodes/month on current plan, before test-script credits)
- [ ] Decide: is this voice/tag combination good enough to build the show around?
---
## Phase 3 — Technical Build
- [ ] **Extend `build_episode.py`** — add `--path` parameter (or equivalent) to support link-chain-based episodes instead of single-subject episodes
- [ ] **Extend `upload_episode.py`** as needed to match
- [ ] **Evaluate Pod Manager fit** — determine whether Wiki Waltz should be tracked as a new show inside Pod Manager or needs something separate; this was left open and should be resolved before building further
- [ ] **New intro/outro audio** — script the sting(s), generate via TTS, integrate into outroll system (separate from Harmonia's existing intro/outro)
- [ ] **Install/build a podcast feed** — stand up RSS feed on the target website
- [ ] **Show identity for the feed** — cover art, author/owner field, category, show description (blocking requirement before the feed can go live anywhere)
---
## Phase 4 — Documentation & Publish
- [ ] Set up the build-log format/location (depends on Phase 0 decision)
- [ ] Document decisions as they're made, not retroactively — the log is part of the experiment
- [ ] Publish first entry once Phase 0 decisions are locked
- [ ] Ongoing: log each subsequent phase as it's completed
---
## Phase 5 — First Episode
- [ ] Produce episode one using the locked format, voice, and walk method
- [ ] Publish to feed
- [ ] Log the episode + process notes
- [ ] Revisit Phase 0 success criteria — is this worth continuing?
---
## Open Questions Log
Running list of anything still undecided — append here rather than losing it in conversation:
-
Log entry date