RomanNum3ral
/
gstack
mirror of https://github.com/garrytan/gstack.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

docs: file scope-gate resolver-extraction TODO (eng-review D5 follow-up) 

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Garry Tan 2026-08-11 17:58:48 -07:00
parent deb80ddecc
commit 57c53e7913
No known key found for this signature in database
GPG Key ID: C1F69E85C74EFE1D
1 changed files with 28 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
TODOS.md
View File

@ -231,6 +231,34 @@ v1.47.0.0 baselines retained in `test/fixtures/` for the v1→v2 audit trail. Th
captured skill bytes match `origin/main` exactly (the rebasing branch left every
SKILL.md untouched). `bun test` is green again.
## Scope-gate follow-ups (filed via /plan-eng-review on the plan-mode auto-select-B change)
### P3: Extract the whole scope gate to a shared `{{SCOPE_GATE}}` resolver
**What:** Move the duplicated scope-gate prose (heading, intro sentence, the
plan-mode/named-target exceptions block, numbered items, the A/B/C menu, and the
Recommendation line) from `plan-eng-review/SKILL.md.tmpl` and
`plan-design-review/SKILL.md.tmpl` into a `scripts/resolvers/` module with 4-5
injected variant slots (preceded-by list, item-2 phrasing, option-C vocabulary,
recommendation tail, exceptions action tail).
**Why:** The two copies are hand-synced today. The drift-guard test in
`test/gen-skill-docs.test.ts` ("scope-gate exceptions drift-guard") makes the
duplication safe but is a stopgap — one source of truth is the real fix. Filed
as D5 of the eng review on the plan-mode auto-select-B change (2026-08-11).
**Pros:** Single source for a load-bearing gate; future gate changes (new
exceptions, wording tuning) land once.
**Cons:** Touches the resolver registry and its tests; must preserve the exact
generated bytes or re-baseline the carve/parity ceilings.
**Context / where to start:** structural-only diff, sequenced AFTER the
behavior change (refactor and behavior never together). The drift-guard test
becomes the migration's acceptance check: extract, regen, confirm byte-identical
output, then retire or simplify the guard. Effort: human ~half day / CC ~20 min.
**Depends on / blocked by:** the plan-mode auto-select-B PR landing on main.
## Token-reduction follow-ups (Phase B, filed via /plan-eng-review on the plan-ceo-review carve)
### P3: Carve the always-loaded `{{PREAMBLE}}` reference blocks into an on-demand doc