docs: upstream spec-kit improvements to consider (2026-07-20)

[pylot-app]

Summary

Automated scan of github/spec-kit v0.8.1 → v0.13.0 (our fork sync point → upstream HEAD).

Issues are disabled on this repo, so tracking this as a PR for discussion. Do not merge — this is a reference document only.

Key findings

• No bulk sync recommended: upstream v0.9–v0.13 made deep CLI/workflow/integration changes out of scope for our Claude-native fork
• 5 cherry-pickable command-template improvements identified (see docs/upstream-improvements-2026-07-20.md)

Top picks

1. 🆕 speckit.converge — new command to assess codebase vs spec/plan/tasks and append remaining tasks (high value)
2. 🐛 speckit.taskstoissues dedup — skip tasks that already have a GitHub issue (prevents re-run duplicates)
3. 🔧 Hook enforcement directive — agents must actually invoke hooks, not just emit the block (all 9 templates)
4. 🔧 speckit.implement constitution load — enforce governance constraints during code generation
5. 🔧 speckit.constitution command scanning — scan .claude/commands/speckit.* not upstream path

Consumer repo status

Repo	Status
fellowship-dev/booster-pack	✅ Up to date (synced 2026-07-06)
fellowship-dev/inbox-angel-worker	⏳ PR github#97 open (checklist update, unmerged)
fellowship-dev/mtg-lotr	✅ Up to date
fellowship-dev/rails-backend	⚠️ Not accessible (App not installed)
fellowship-dev/clapes	⚠️ Not accessible (App not installed)

🤖 Generated by automated spec-kit sync (2026-07-20)

[pylot-app]

PR Review: #8 — docs: upstream spec-kit improvements to consider (2026-07-20)

Branch: docs/upstream-improvements-2026-07-20 → main
Size: +113 / -0 across 1 file

Summary

This PR adds a single new documentation file (docs/upstream-improvements-2026-07-20.md) — a bot-generated reference report cataloguing cherry-pickable improvements from upstream github/spec-kit v0.8.1 → v0.13.0. It is explicitly marked "Do not merge" and exists purely for team discussion and tracking. The approach is sound: structured, prioritized, and clearly scoped to command-template improvements applicable to the Claude-native fork.

Findings

No issues found above confidence threshold.

Convention Compliance

CLAUDE.md conventions (terse rewrites, ≤40–50 line limits) apply to command templates, not documentation files. No conventions violated. The fork's stated sync point (v0.8.1) matches the document's stated current sync point.

Closes vs Refs

No Closes keywords found.

Verdict

Clean — proceed to double-check

[pylot-app]

Double-Check Review: PR #8 — docs: upstream spec-kit improvements to consider (2026-07-20)

Reviewer: Automated double-check
Branch: docs/upstream-improvements-2026-07-20 → main

---

Intent

This PR adds a bot-generated reference document cataloguing cherry-pickable improvements from upstream github/spec-kit v0.8.1 → v0.13.0. It is explicitly marked "Do not merge" and exists purely for team discussion and tracking — the PR delivers exactly what it describes.

Implementation

• Single new file: docs/upstream-improvements-2026-07-20.md (+113 / -0)
• Structured as: upstream status → 6 recommended cherry-picks with priority ratings → already-applied section → out-of-scope section → priority order
• No changes to any command templates, configs, or executable code

Curated CI Findings

No CI review comments found — reviewed diff directly. First review returned findings: [], confirmed accurate against the diff.

New Issues (not caught by first review)

None.

(Minor cosmetic observations noted but not defects: items 4 and 5 lack an explicit "Recommendation:" line; PR body says "5 improvements" while doc lists 6 (item 6 is LOW/low-urgency and absent from priority order, so "5 actionable" is accurate); top-picks order in PR body differs from doc section order — all cosmetic only in a reference-only document.)

Tests After Fixes

• Suite: Not applicable — documentation-only PR (single new markdown file, no code changes)
• Regressions: N/A

Verdict

Ready for CTO review — documentation-only reference PR, no findings, clean diff, correctly marked "Do not merge". No fixes required.

[pylot-app]

CTO Review: fellowship-dev/spec-kit PR #8 — docs: upstream spec-kit improvements to consider (2026-07-20)

Date: 2026-07-20
Repo: fellowship-dev/spec-kit
PR: fellowship-dev/spec-kit#8 — docs: upstream spec-kit improvements to consider (2026-07-20)
Branch: docs/upstream-improvements-2026-07-20 → main
Labels: reviewed, double-checked

---

Verdict

✅ LGTM (hold) — Reference doc only; clear, accurate, well-scoped; explicitly not for merge; no blockers

---

CTO Checklist

Documentation

Check	Status
.env.example — new env vars	✅ No new env vars introduced
README.md — new features	✅ N/A — internal reference doc, not a shipped feature
Setup docs / docs/setup.md	✅ N/A — no setup changes
CHANGELOG.md	✅ N/A — docs-only PRs do not require an entry
New file docs/upstream-improvements-2026-07-20.md	✅ Self-documenting; date-stamped; clearly scoped in PR body

External Dependencies

• None — no new packages, APIs, or services introduced

Downstream Impact

• fellowship-dev/booster-pack, fellowship-dev/inbox-angel-worker, fellowship-dev/mtg-lotr — no impact; document is internal to spec-kit only
• All changes are opt-in — read-only reference document; no template, command, or config changes reach consumer repos

Merge Strategy

• ✅ Hold — PR body explicitly states "Do not merge — this is a reference document only." Content is high-quality and actionable; this PR should remain open as a tracking artifact. Each cherry-pick should be a separate implementation PR when actioned.

Process Verification

Check	Status
Related code searched	✅ PR body calls out what is "Already Applied" (checklist dimensions, PR #4)
Docs updated	✅ This PR IS the documentation; no companion code changes to document
Merge strategy	N/A — reference doc only
FlowChad flows affected	✅ none — no .flowchad/ or flows/ files touched
Production impact assessed	✅ low risk — markdown file added to docs/; zero production surface

---

Action Items Before Merge

None — this PR is intentionally held as a reference artifact. When actioning cherry-picks, open separate PRs:

1. templates/commands/converge.md — add speckit.converge new command + lean preset variant (≤40 lines); rated HIGH VALUE
2. templates/commands/taskstoissues.md — cherry-pick speckit.taskstoissues dedup fix; rated HIGH VALUE (prevents duplicate issues on re-run)
3. All 9 command templates — apply hook enforcement directive; rated MEDIUM VALUE (fixes compliance gap where agents emit hook block but don't execute it)

[pylot-app]

Auto-pylot Status Check

This PR is intentionally held open as a reference document ("Do not merge" stated in PR body). CTO verdict was LGTM, action=held (2026-07-20).

Waiting on: Human decision to either merge (adopt the upstream improvements as a reference artifact) or close (if the discussion is complete and this is no longer needed as an open PR).

Auto-pylot will check again next cycle.