[marius-kilocode]

What Problem This Solves

The migration skills told agents which done-proof to write, but not how to decide that a cell is done. In the Gateway port (#15237) a given classification would have closed cells with real gaps: native for request defaults that v2 does not send, dropped for cloud session import and the event service client (both have plan units), and ported for modules whose v2 code lacks part of the behavior.

Why This Change Was Made

The rules now sit where agents read them before and during a port:

• .kilo/skills/v2-migration/SKILL.md, section "Progress map": an ordered choice of proof (marker, kilo_port, v1 file name, ported, native, facade, dropped), and a verification list: compare behavior by behavior, find v1 callers, dropped only for behavior that is not carried (never for behavior a plan unit still ports), no closing with a known gap, mixed hunks, test cells, claims in tasks or plans are verified, git add before regenerating, regenerate twice and --check, a reason for each open cell.
• .kilo/skills/v2-feature-placement/SKILL.md, step 7: the same choice in short form, with the commands, and a pointer to the full rules.
• .kilo/skills/v2-migration/SKILL.md, section "Evidence": every PR that changes the map gets a ## Progress map section with the box before and after, and one short line per cell or group on why it is done, why it is dropped (what it did on v1, why v2 does not need it, the evidence), and why it stays open (what is missing, which unit owns it), plus corrections to a given classification. v2-feature-placement step 8 points to it.
• script/upstream/progress.ts: one counting rule in the generated migration-tracking/progress/README.md that says what a proof claims and when dropped applies.

User Impact

None at runtime. Future ports close progress map cells by the same rules.

Evidence

• bun script/upstream/progress.ts twice with identical output, then --check: up to date. The map numbers do not change.
• Docs and generator text only. No em dashes.

This PR and #15237 both regenerate migration-tracking/progress/README.md. The second one to merge regenerates it.

[github-actions]

➖ v2 migration progress: 0.9% → 0.9% (0 done, 0 total, 0 dropped; 86/9829 cells)

Placement: 0 new marked lines in shared files, 0 new kilo_port annotations, 0 new core-internal imports.

Ratchets: 8 core-internal imports (base 8); shared patches by rung 16 seam, 1 patch (base 16 seam, 1 patch).

No change to the progress map.

Full map · how cells close

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Incremental review of dbf4eca over the previously reviewed d754045. The only new change is commit 3, which regenerates migration-tracking/progress/README.md and history.svg for the merge base.

The regenerated history now matches script/upstream/progress.ts: at the PR merge base (2d6722a9), history() skips that commit because its progress.json equals the current output (86/9829), then takes 48335f4 as the last 2026-10-19 point (39/9841) and a20041a as 2026-10-16 (12/9854). The README table and the SVG (169.85 = 0.1%, 169.40 = 0.4%, 168.65 = 0.9%) are internally consistent and match the generator's chart() math. The changed image references stay in Markdown syntax, and no runtime code changed, so there is no leak or behavior risk.

Files Reviewed (4 files)

• .kilo/skills/v2-migration/SKILL.md
• .kilo/skills/v2-feature-placement/SKILL.md
• migration-tracking/progress/README.md
• script/upstream/progress.ts

Note: review was read-only, so the generator was not executed; the history values were verified against the committed progress.json files and the generator source.

Previous Review Summaries (2 snapshots, latest commit d754045)

Current summary above is authoritative. Previous snapshots are kept for context only.

Previous review (commit d754045)

Status: No Issues Found | Recommendation: Merge

Incremental review of d754045 (docs-only, +33/-2 in two skill files). The new ## Progress map PR-section rule and the updated v2-feature-placement step 8 pointer match the current skills structure and the generator: box done/total with dropped excluded from the denominator, per-kind module/patch/test ratios, and landed <n> plugin all match script/upstream/progress.ts (RUNG, Box, rungText), and note/by are indeed the required dispositions.json fields. The cross-reference from v2-feature-placement to the v2-migration section "Evidence" resolves to the new subsection. The example feature names (server, cloud session import) and unit U-E3-04 exist in the map config and plan. No markdown images were touched, and no runtime code changed, so there is no leak or behavior risk.

Files Reviewed (4 files)

• .kilo/skills/v2-migration/SKILL.md
• .kilo/skills/v2-feature-placement/SKILL.md
• migration-tracking/progress/README.md
• script/upstream/progress.ts

Note: review was read-only, so the generator was not executed; the README and generator sources are unchanged since the previous review and still agree.

Previous review (commit af7ebd2)

Status: No Issues Found | Recommendation: Merge

Documentation-only change. Every new rule was checked against script/upstream/progress.ts: --list <feature> filtering, the pins.v1Target pin, the ported/native/facade/dropped states, rung assignment for kilo_port (2 in packages/kilo-cli, 1 elsewhere), the git ls-files/git grep dependence on tracked files (so the git add step is correct), and the stated inability to prove unattributed:<area> cells (the proof regex only accepts [a-z0-9-]+ feature names). The three regenerated README.md lines match the generator strings exactly, and the changed README.md image is already Markdown syntax. No runtime code changed, so no leak or behavior risk.

Files Reviewed (4 files)

• .kilo/skills/v2-migration/SKILL.md
• .kilo/skills/v2-feature-placement/SKILL.md
• migration-tracking/progress/README.md
• script/upstream/progress.ts

Note: review was read-only, so bun script/upstream/progress.ts --check was not executed; the PR body reports it as up to date and the README diff matches the generator source.

---

_{Reviewed by deepseek-v4.1-flash-1017 · Input: 54.4K · Output: 11.2K · Cached: 688.3K}

_{Review guidance: REVIEW.md from base branch v2}