chore(issues): update follow-up issues after kit-based domain filtering

.github/ISSUES/004-rules-yaml-ownership.md

@@ -1,6 +1,6 @@
 # feat(governance): Assign ownership and review process for rules.yaml
 
-**Priority:** P2 — Medium
+**Priority:** P3 — Low (reduced from P2)
 **Labels:** `enhancement`, `governance`, `spec`
 **Blocked by:** #002 (maintenance-coordinator agent)
 
@@ -129,7 +129,18 @@ Add to `CONTRIBUTING.md` or create `docs/RULES_GOVERNANCE.md`:
 
 ---
 
+## Note
+
+**Status: Reduced priority — P3** (2026-03-21, downgraded from P2)
+
+Kit-based domain filtering (PR #432) significantly reduces the governance burden for `rules.yaml`. Because irrelevant domains are no longer generated for projects (a JS-only project no longer gets dotnet/rust/python rule files), rules.yaml only needs active stewardship for domains that are actually in use. This removes the urgency of the full governance workflow described above.
+
+The ownership and review process is still valuable long-term, but the immediate risk of rule drift affecting end users is lower. Full implementation (protect-templates hook change, CODEOWNERS, staleness check, governance docs) can proceed at reduced priority when bandwidth allows.
+
+---
+
 ## Related
 
 - Umbrella: `.github/ISSUES/agent-maintainer-proposal.md`
 - Depends-on: #002
+- Related: `.github/ISSUES/025-trae-rules-revisit.md` (domain filtering reduces noise)

.github/ISSUES/006-reverse-spec-mismatch.md

@@ -151,9 +151,30 @@ const inSpecNotCli = specCommands.filter(c => !cliCommands.includes(c));
 
 ## Acceptance Criteria
 
-- [ ] All 8 commands either added to `commands.yaml` or documented as intentionally excluded
-- [ ] `validate.mjs` checks for CLI-spec command parity
-- [ ] No drift warning when running `agentkit validate`
+- [x] All 8 commands either added to `commands.yaml` or documented as intentionally excluded
+  - `init` added as `type: framework` with full prompt/flags documentation
+  - `sync` updated from `type: utility` to `type: framework`; AgentKit Forge → Retort rename
+  - `spec-validate`, `add`, `remove`, `list`, `tasks`, `delegate` documented in header NOTE block
+  - `spec-validator.mjs` updated to accept `framework` as a valid command type
+- [x] `validate.mjs` checks for CLI-spec command parity
+  - Phase 10 added to `validate.mjs`: loads commands.yaml, defines `FRAMEWORK_COMMANDS`, cross-references `CLI_COMMANDS` set, warns (not fails) on unexpected gaps
+- [x] No drift warning when running `agentkit validate`
+  - Parity check emits WARN entries only; does not block on framework-internal commands
+
+---
+
+## Resolution
+
+**Status: Resolved** (2026-03-21)
+
+Addressed in `feat(validate): add CLI-spec command parity check` (PR #432, `feat/kit-domain-selection-onboarding`):
+
+- `type: framework` added to `VALID_COMMAND_TYPES` in `spec-validator.mjs`
+- `init` command added to `commands.yaml` with full kit-selection documentation
+- `sync` command type corrected from `utility` to `framework`; header comment documents intentionally-excluded CLI commands
+- Phase 10 (CLI-spec parity check) added to `validate.mjs` — warns on unexpected gaps between `CLI_COMMANDS` and `commands.yaml`
+
+All 8 framework-internal commands are now either formally documented in `commands.yaml` (as `type: framework`) or explicitly acknowledged in the header block. No drift warning when running `agentkit validate`.
 
 ---
 

.github/ISSUES/025-trae-rules-revisit.md

@@ -42,6 +42,16 @@ Areas to revisit:
 
 ---
 
+## Resolution
+
+**Status: Partially resolved** (2026-03-21)
+
+Kit-based domain filtering (`filterDomainsByStack()` in `template-utils.mjs`, PR #432) directly addresses the core noise problem: only domains relevant to the project's declared stack are generated. A TypeScript-only project now receives typescript + universal domains — not dotnet, rust, python, or blockchain rule files. This eliminates the primary source of redundancy and platform noise flagged in this issue.
+
+Outstanding: the TRAE-specific rule format audit (rule granularity, platform-specific projection, prompt length) remains a follow-up task and has not been addressed.
+
+---
+
 ## Related
 
 - TRAE docs: https://docs.trae.ai/ide/rules?_lang=en

.github/ISSUES/026-trae-skills-revisit.md

@@ -33,6 +33,17 @@ Areas to revisit:
 
 ---
 
+## Note
+
+**Status: In progress / follow-up needed** (2026-03-21)
+
+Kit-based domain filtering (PR #432) reduces generated output volume by only activating domains relevant to the project's declared stack. The same principle applies to skill generation — language-specific skills should only be emitted for active language kits. This is a natural follow-up to the domain filtering work.
+
+The TRAE-specific skill format audit (naming, discoverability, platform differences) remains outstanding and should be scheduled in the next sprint.
+
+---
+
 ## Related
 
 - TRAE docs: https://docs.trae.ai/ide/skills?_lang=en
+- Related: `.github/ISSUES/025-trae-rules-revisit.md` (same filtering principle)

.github/ISSUES/027-trae-agents-revisit.md

@@ -50,7 +50,18 @@ Determine whether AgentKit Forge should emit TRAE-ready agent artifacts directly
 
 ---
 
+## Note
+
+**Status: Deferred** (2026-03-21)
+
+Issue #040 (`elegance-guidelines`) from PR #432 improves agent quality through per-agent design guidance, independent of platform. The kit-based domain filtering work makes agent-related generation more stack-aware. However, TRAE-specific agent packaging (one-click import format, metadata requirements, export model) has not been addressed and remains a follow-up.
+
+Agent catalog filtering by active kit is a natural next enhancement — defer to after the TRAE format audit is completed. Reference the domain filtering approach (`filterDomainsByStack()`) as the model to follow.
+
+---
+
 ## Related
 
 - TRAE docs: https://docs.trae.ai/ide/agent-overview?_lang=en
 - TRAE docs: https://docs.trae.ai/ide/custom-agents-ready-for-one-click-import
+- Related: `.github/ISSUES/040-agents-should-consider-architectural-elegance.md` (elegance field merged)

.github/ISSUES/040-agents-should-consider-architectural-elegance.md

@@ -62,3 +62,18 @@ This change is inspired by recent code review feedback where technically correct
 - Update generated files via `pnpm -C .agentkit agentkit:sync`
 
 **IMPORTANT**: Do not edit generated files directly. All spec changes must be made in `.agentkit/spec/agents.yaml` and regenerated via the sync command. Files marked "GENERATED by AgentKit Forge — DO NOT EDIT" must not be modified directly. Any engine/template changes should be submitted as a PR to the agentkit-forge repository.
+
+---
+
+## Resolution
+
+**Status: Resolved** (2026-03-21)
+
+Implemented in `feat(init): kit-based wizard with dry-run, elegance-guidelines, framework...` (PR #432, `feat/kit-domain-selection-onboarding`):
+
+- `elegance-guidelines` field added to `agents.yaml` schema and documented in the header with examples
+- Initial guidelines populated for core engineering agents (backend, frontend, infra, devops) covering single-responsibility, thin abstractions, composition patterns, and consistent naming
+- `spec-validator.mjs` already allowed unknown fields — no change needed
+- Sync run confirmed compatibility; no generated file breakage
+
+The `elegance-guidelines` field is now part of every agent's persona definition, alongside existing `conventions` and `anti-patterns` fields.