Work log: Phantasmagoria — April 1, 2026

What shipped today

Today’s session was a tidy execution of the two cleanest items from the March 30 carry-over queue — both ready-for-dev, both quick wins that leave the codebase measurably cleaner.

Dead code elimination. The scout audit had identified seven dead items in three files: the entire validation.py module (untouched since December), four dead accessor functions in option_roller.py, dead prompt-building functions in context.py, and two dead config blocks in config_mod.yaml. All of it was removed in a single clean commit. Nothing was rescued — each item had been superseded by the current implementation and was genuinely unreachable. The codebase is smaller and the cognitive map is cleaner.

Risk register created. Six scattered risks that had been mentioned in work logs, issue comments, and code comments were consolidated into a formal RISKS.md register at the repo root. The register captures impact, likelihood, status, and last-reviewed date for each risk. The six risks now tracked: tech dedup being prompt-level only (no hard enforcement), situations being disabled with no variety system timeline, copy-pasted generator logic across four files, the tension evaluator being advisory rather than blocking, research progress being invisible to players, and choices still sometimes having an obvious dominant option. Having these in one place means the next session starts with a clear priority list instead of scattered notes.

Completed

• #263 — Remove dead code: validation.py, dead functions, dead config keys
• #264 — Create RISKS.md risk register

Carry-over

• #261 — resource_gain_scaled validation — the most impactful open issue. The new default effect type is invisible to all three validation layers. Needs prep then implementation.
• #265 — Python 3.9 EOL upgrade — runtime is EOL since October 2025. Needs prep.
• Stage 3/4 split — plan exists at .claude/plans/shimmering-skipping-oasis.md, has been carrying over since March 29. Consider whether this is still the right next architectural move.

Risks

All six open risks are now formally tracked in RISKS.md. No new risks emerged today.

Flags and watch-outs

• The celestial_equinox backup files (SUMMARY_pre_stage1_reset_* and events_pre_stage1_reset_*/) are still untracked. They appear to be pre-reset snapshots — confirm they’re safe to delete or add to .gitignore next session.
• uv.lock is also untracked. If the project uses uv for dependency management, this file should be committed.

Next session

1. #261 — resource_gain_scaled validation — highest-impact open issue. Prep the spec, then implement: add resource_gain_scaled to all three validation layers (generation, source lint, build). Without this, the most common effect type runs completely unvalidated.
2. Resolve untracked files — decide fate of the celestial_equinox backup snapshots and uv.lock before they accumulate more confusion.
3. #265 — Python 3.9 upgrade — prep the migration path. Runtime is already EOL; this is a ticking clock.
4. Stage 3/4 split — re-evaluate whether to execute the plan at .claude/plans/shimmering-skipping-oasis.md or deprioritize it in favor of #261.