|
1 | 1 | --- |
2 | 2 | id: 192 |
3 | 3 | title: "Limit FO-built checklists to 3 items by default; allow workflow README override" |
4 | | -status: ideation |
| 4 | +status: implementation |
5 | 5 | source: "captain directive — FO dispatch checklists currently drift toward 8-14 items (observed throughout 2026-04-17/18 session). That enumerates mechanical steps instead of forcing the FO to name the 3 things that matter. Cognitive load on both ensign (reading) and FO (tracking) increases with length; prioritization erodes. A 3-item ceiling by default, overridable per workflow, forces the FO to compress." |
6 | 6 | started: 2026-04-18T04:40:02Z |
7 | 7 | completed: |
8 | 8 | verdict: |
9 | 9 | score: 0.6 |
10 | | -worktree: |
| 10 | +worktree: .worktrees/spacedock-ensign-limit-fo-checklist-to-three-items |
11 | 11 | issue: |
12 | 12 | pr: |
13 | 13 | mod-block: |
@@ -189,6 +189,34 @@ dispatch: |
189 | 189 |
|
190 | 190 | Precedence: per-workflow override (if present and valid) > default 3. Invalid values (non-integer, ≤0) rejected at `claude-team build` with clear error naming the file and key path. |
191 | 191 |
|
| 192 | +### Feedback Cycles — Cycle 1 (captain amendment at ideation gate) |
| 193 | + |
| 194 | +**2026-04-18.** Captain approved with four narrowing amendments that simplify implementation scope: |
| 195 | + |
| 196 | +1. **No mechanism enforcement at `claude-team build`.** Prose-only. The AC-1 claude-team validation + unit test is DROPPED. Rationale: at this point, FO discipline via prose is sufficient; mechanism can be added later if prose drift proves insufficient. |
| 197 | +2. **No workflow README frontmatter override.** The `dispatch.checklist.max-items` field and all supporting parsing logic is DROPPED. AC-3's adoption question becomes vacuous. Rationale: don't add configurability until multiple workflows actually need different caps. |
| 198 | +3. **"Up to 3" not "exactly 3".** The cap is an upper bound. 0, 1, 2, or 3 items are all valid. The FO must NOT pad to hit 3. Rationale: forcing 3 items would re-introduce the ceremonial-filler problem the original complaint was about. |
| 199 | +4. **Reframe as "linchpins" not "mandatory items".** The prose guidance should describe checklist items as the few things that demonstrate the job is done well — linchpins, not a complete work-breakdown. Suggested phrasing: *"Checklist items are linchpins — the few signals (at most 3) that demonstrate this dispatch's job is done well. Not a work breakdown; the ensign already knows how to read the entity body, commit, and write a stage report. Name what separates a good outcome from a ceremonial one."* |
| 200 | + |
| 201 | +### Revised scope for implementation (post-amendment) |
| 202 | + |
| 203 | +- **Single edit location:** `skills/first-officer/references/first-officer-shared-core.md` near line 60 (currently reads "Build a numbered checklist from stage outputs and entity acceptance criteria"). Add guidance that: |
| 204 | + - Checklists are capped at ≤ 3 items (upper bound, not a target) |
| 205 | + - Items should be linchpins — signals that demonstrate the dispatch's job is done well |
| 206 | + - The boilerplate (read entity body, commit before signaling, write stage report) is covered by structural conventions and MUST NOT appear in the checklist |
| 207 | +- **No `claude-team build` changes.** No workflow README changes. No entity template changes. |
| 208 | +- **Static test:** grep-based assertion in `tests/test_static_guardrails.py` (or equivalent) that the shared-core prose contains the ≤ 3 cap and the "linchpins" framing. |
| 209 | + |
| 210 | +### Revised acceptance criteria (post-amendment) |
| 211 | + |
| 212 | +**AC-1 (entity-level)** — `skills/first-officer/references/first-officer-shared-core.md` has a short paragraph (near line 60) articulating the ≤ 3 cap, the linchpin framing, and the excluded boilerplate. Verified by grep + file inspection. |
| 213 | + |
| 214 | +**AC-2 (entity-level)** — Static test (new or existing) asserts the prose is present with both the "3" cap and the "linchpin" framing. Verified by `make test-static` adding 1 passing test. |
| 215 | + |
| 216 | +**AC-3 (entity-level)** — `make test-static` green on main after the change. |
| 217 | + |
| 218 | +Previous AC-2 (mechanism enforcement), AC-3 (workflow adoption), AC-4 (commission skill), AC-5 (budget discipline) all DROPPED per amendments. |
| 219 | + |
192 | 220 | ## Stage Report (ideation) |
193 | 221 |
|
194 | 222 | ### Summary |
|
0 commit comments