@hegemonart/get-design-done 1.57.1 → 1.57.2

package/SKILL.md

@@ -37,6 +37,8 @@ Each stage produces artifacts in `.design/` inside the current project.
 | `compare` | `get-design-done:gdd-compare` | Delta between DESIGN.md baseline and DESIGN-VERIFICATION.md → .design/COMPARE-REPORT.md |
 | `figma-write <mode>` | `get-design-done:gdd-figma-write` | Write design decisions to Figma (annotate/tokenize/mappings) |
 | `figma-extract <file-url-or-key>` | `get-design-done:gdd-figma-extract` | Off-context Figma design-system extraction → compact local digest (DESIGN.md + tokens.json + components.json), zero raw JSON in context |
+| `paper-write <mode>` | `get-design-done:gdd-paper-write` | Write design decisions back into paper.design via MCP (annotate/tokenize/roundtrip) |
+| `pencil-write <mode>` | `get-design-done:gdd-pencil-write` | Write design decisions into git-tracked `.pen` spec files (annotate/roundtrip) |
 | `graphify <subcommand>` | `get-design-done:gdd-graphify` | Manage Graphify knowledge graph (build/query/status/diff) |
 | `discuss [topic] [--all] [--spec] [--cycle <name>]` | `get-design-done:gdd-discuss` | Adaptive design interview - spawns design-discussant; appends D-XX decisions to STATE.md |
 | `list-assumptions [--area]` | `get-design-done:gdd-list-assumptions` | Surface implicit design assumptions baked into the codebase |

package/agents/README.md

@@ -33,24 +33,24 @@ The `design-` prefix prevents name collisions with agents from other Claude Code
 
 ## Frontmatter Schema
 
-Every agent file begins with a YAML frontmatter block. All fields except `model` are required. The `default-tier` and `tier-rationale` fields were added in Phase 10.1 - see `reference/model-tiers.md` for the per-agent assignment rationale.
+Every agent file begins with a YAML frontmatter block. All fields except `model` are required. See `reference/model-tiers.md` for the per-agent `default-tier` / `tier-rationale` assignment rationale.
 
 | Field | Type | Accepted values | Purpose |
 |-------|------|-----------------|---------|
 | `name` | kebab-case string | unique within plugin | Identifier passed to the `Task` tool - must match the filename without `.md` |
 | `description` | string | free-form | One sentence: what the agent does + when it is spawned |
-| `description_i18n` | object | `{ <locale>: "<description>" }` | **Phase 40.5, opt-in.** Localized descriptions keyed by locale (en/ru/uk/de/fr/zh/ja). `scripts/lib/i18n/index.cjs` `descriptionFor(frontmatter, locale)` resolves it via the fallback chain and falls back to the English `description` when a locale is absent. Backward-compatible - omit it and nothing changes. |
+| `description_i18n` | object | `{ <locale>: "<description>" }` | **Opt-in.** Localized descriptions keyed by locale (en/ru/uk/de/fr/zh/ja). `scripts/lib/i18n/index.cjs` `descriptionFor(frontmatter, locale)` resolves it via the fallback chain and falls back to the English `description` when a locale is absent. Backward-compatible - omit it and nothing changes. |
 | `tools` | comma-separated list | `Read`, `Write`, `Edit`, `Bash`, `Grep`, `Glob`, `Task`, `WebFetch`, `TodoWrite`, `mcp__*` | Claude tools the agent may use - list only what is needed |
 | `color` | enum | `yellow`, `green`, `blue`, `red` | Terminal display color for the agent's output |
 | `model` | enum (optional) | `inherit`, `sonnet`, `haiku` | Omit to use the project's configured profile default. Use `inherit` to bypass the profile and use the highest available model (quality-tier work) |
-| `default-tier` | enum | `haiku`, `sonnet`, `opus` | **Phase 10.1.** The model tier the router + budget-enforcer hook select when `.design/budget.json.tier_overrides` has no entry for this agent. Paired with `reference/model-tiers.md` - the per-agent map in that file is the source of truth; this field is the per-agent replica the hook reads. Required on all agents. |
-| `tier-rationale` | string | free-form, one line, quoted | **Phase 10.1.** One-sentence justification for the `default-tier` choice. Surfaces in `/gdd:optimize` output when the advisor suggests a tier move. Required on all agents. |
+| `default-tier` | enum | `haiku`, `sonnet`, `opus` | The model tier the router + budget-enforcer hook select when `.design/budget.json.tier_overrides` has no entry for this agent. Paired with `reference/model-tiers.md` - the per-agent map in that file is the source of truth; this field is the per-agent replica the hook reads. Required on all agents. |
+| `tier-rationale` | string | free-form, one line, quoted | One-sentence justification for the `default-tier` choice. Surfaces in `/gdd:optimize` output when the advisor suggests a tier move. Required on all agents. |
 | `parallel-safe` | enum | `always`, `never`, `conditional-on-touches`, `auto` | Whether stages may dispatch this agent in parallel with siblings. `conditional-on-touches` means safe only when `Touches:` do not overlap |
-| `typical-duration-seconds` | int | e.g. `30`, `60`, `120` | Expected wall-clock duration. Used by parallelism planner to decide whether savings clear `min_estimated_savings_seconds`. **Extensible** - Phase 10.1 adds `default-tier` override; Phase 11's `design-reflector` adds `measured-duration-seconds` from telemetry without replacing this field. |
+| `typical-duration-seconds` | int | e.g. `30`, `60`, `120` | Expected wall-clock duration. Used by parallelism planner to decide whether savings clear `min_estimated_savings_seconds`. **Extensible** - the `design-reflector` may propose `measured-duration-seconds` from telemetry without replacing this field. |
 | `reads-only` | bool | `true`/`false` | True when the agent never writes any file |
 | `writes` | list | e.g. `[".design/DESIGN-PLAN.md"]` | Files / globs the agent may write. `[]` for read-only agents |
 
-> **Frontmatter is extensible.** New fields can be added by downstream phases without removing existing ones. The `design-reflector` agent (Phase 11) may propose updates to `typical-duration-seconds` and `default-tier` based on measured telemetry - those proposals go through `/gdd:apply-reflections`, never auto-applied.
+> **Frontmatter is extensible.** New fields can be added without removing existing ones. The `design-reflector` agent may propose updates to `typical-duration-seconds` and `default-tier` based on measured telemetry - those proposals go through `/gdd:apply-reflections`, never auto-applied.
 
 Example frontmatter block:
 
@@ -67,9 +67,9 @@ color: blue
 
 ## Runtime-neutral reasoning class (alias for default-tier)
 
-**Phase 26 (v1.26.0).** Agents may carry an optional `reasoning-class: high|medium|low` field as a runtime-neutral alias for `default-tier`. The alias exists because `default-tier`'s enum (`opus|sonnet|haiku`) hard-codes Anthropic model names, while the multi-runtime installer (Phase 24) ships agents to 14 runtimes whose authors do not all use those names. `reasoning-class` describes the *reasoning density* the agent needs without naming a vendor's model lineup.
+**Introduced in v1.26.0.** Agents may carry an optional `reasoning-class: high|medium|low` field as a runtime-neutral alias for `default-tier`. The alias exists because `default-tier`'s enum (`opus|sonnet|haiku`) hard-codes Anthropic model names, while the multi-runtime installer ships agents to 14 runtimes whose authors do not all use those names. `reasoning-class` describes the *reasoning density* the agent needs without naming a vendor's model lineup.
 
-**This field is additive, not a replacement.** `default-tier: opus|sonnet|haiku` remains the authoritative, required field for v1.26 and is the source of truth that `hooks/budget-enforcer.ts`, `skills/router/SKILL.md`, and `agents/gdd-intel-updater.md` read. Both fields may coexist on the same agent during the transition window. The long-term winner - which field is canonical and which is deprecated - is data-gated per Phase 28+ measurement of adoption rates (CONTEXT D-10); no deprecation lands in v1.26.
+**This field is additive, not a replacement.** `default-tier: opus|sonnet|haiku` remains the authoritative, required field for v1.26 and is the source of truth that `hooks/budget-enforcer.ts`, `skills/router/SKILL.md`, and `agents/gdd-intel-updater.md` read. Both fields may coexist on the same agent during the transition window. The long-term winner - which field is canonical and which is deprecated - is data-gated by future measurement of adoption rates; no deprecation lands in v1.26.
 
 ### Frontmatter shape
 
@@ -77,7 +77,7 @@ color: blue
 |-------|------|-----------------|----------|---------|
 | `reasoning-class` | enum | `high`, `medium`, `low` | optional | Runtime-neutral name for the reasoning-density tier this agent needs. Equivalent to `default-tier` per the equivalence table below. |
 
-### Equivalence (locked in CONTEXT D-10)
+### Equivalence
 
 | `reasoning-class` | `default-tier` | Typical role classes |
 |-------------------|----------------|----------------------|
@@ -100,34 +100,33 @@ tier-rationale: "Authors DESIGN-PLAN.md — the contract every downstream agent
 ---
 ```
 
-When both are present, the values MUST be equivalent per the table above. Mismatched dual annotations (e.g. `default-tier: opus` paired with `reasoning-class: medium`) are a validation error - `scripts/validate-frontmatter.ts` (extended in Plan 26-08) enforces equivalence at lint time. If only one of the two is present, the validator accepts it and downstream consumers use the equivalence table to derive the missing field.
+When both are present, the values MUST be equivalent per the table above. Mismatched dual annotations (e.g. `default-tier: opus` paired with `reasoning-class: medium`) are a validation error - `scripts/validate-frontmatter.ts` enforces equivalence at lint time. If only one of the two is present, the validator accepts it and downstream consumers use the equivalence table to derive the missing field.
 
 ### How runtime-aware tooling reads either field
 
 Downstream consumers (`skills/router/SKILL.md`, `hooks/budget-enforcer.ts`, `scripts/lib/budget-enforcer.cjs`, `agents/gdd-intel-updater.md`) accept either field individually and map between them via the equivalence table:
 
 - **`default-tier` only** - consumers read `default-tier` directly. This is the v1.26 baseline state for all 26 shipped agents.
-- **`reasoning-class` only** - consumers map `high → opus`, `medium → sonnet`, `low → haiku` and feed the resulting tier into `tier-resolver.cjs` (Plan 26-02) for runtime-correct model resolution. Consumers that have not yet been updated to read `reasoning-class` natively still see a valid `default-tier` semantically (via the alias), so no consumer breaks when an agent author chooses the runtime-neutral name.
-- **Both present** - consumers prefer `default-tier` for now (v1.26 canonical), with `reasoning-class` carried through to telemetry (`gdd-intel-updater` writes both fields to `.design/intel/agent-tiers.json` per Plan 26-08) so adoption can be measured for the Phase 28 deprecation gate.
+- **`reasoning-class` only** - consumers map `high → opus`, `medium → sonnet`, `low → haiku` and feed the resulting tier into `tier-resolver.cjs` for runtime-correct model resolution. Consumers that have not yet been updated to read `reasoning-class` natively still see a valid `default-tier` semantically (via the alias), so no consumer breaks when an agent author chooses the runtime-neutral name.
+- **Both present** - consumers prefer `default-tier` for now (v1.26 canonical), with `reasoning-class` carried through to telemetry (`gdd-intel-updater` writes both fields to `.design/intel/agent-tiers.json`) so adoption can be measured for the future deprecation gate.
 
 ### Rollout policy for v1.26
 
-- The 26 existing agents continue to carry `default-tier` only - **no per-agent retrofit lands in v1.26**. New agents (added in Phase 27+) MAY carry `reasoning-class` instead of, or alongside, `default-tier`.
-- Validators, intel-updater, router, and budget-enforcer accept either field starting in v1.26 (Plans 26-04, 26-05, 26-08).
-- Adoption is measured by `gdd-intel-updater` over `agents/*.md` changes; if alias adoption stays below 50% by Phase 28, `default-tier` remains canonical and the alias is deprecated. If alias wins majority share, the reverse. **No deprecation in v1.26.**
+- The 26 existing agents continue to carry `default-tier` only - **no per-agent retrofit lands in v1.26**. New agents MAY carry `reasoning-class` instead of, or alongside, `default-tier`.
+- Validators, intel-updater, router, and budget-enforcer accept either field starting in v1.26.
+- Adoption is measured by `gdd-intel-updater` over `agents/*.md` changes; if alias adoption stays below 50% at the deprecation review, `default-tier` remains canonical and the alias is deprecated. If alias wins majority share, the reverse. **No deprecation in v1.26.**
 
 ### Cross-references
 
 - `reference/model-tiers.md` - tier-selection guide and per-agent map for `default-tier`. The same role-class rationale applies to `reasoning-class` via the equivalence table.
-- `reference/runtime-models.md` (Plan 26-01) - per-runtime tier→model adapter that consumes the resolved tier (whether sourced from `default-tier` or via `reasoning-class` alias).
-- `scripts/validate-frontmatter.ts` (Plan 26-08) - validator extension that accepts the optional field and enforces equivalence when both are present.
-- `.planning/phases/26-headless-model-resolver/CONTEXT.md` D-10, D-11 - decision lineage for additive-alias and equivalence-enforced semantics.
+- `reference/runtime-models.md` - per-runtime tier→model adapter that consumes the resolved tier (whether sourced from `default-tier` or via `reasoning-class` alias).
+- `scripts/validate-frontmatter.ts` - validator that accepts the optional field and enforces equivalence when both are present.
 
 ---
 
 ## Peer-CLI delegation (delegate_to)
 
-Phase 27 introduces an **optional** frontmatter field `delegate_to:` that lets an agent OPT IN to running on a peer CLI (Codex via ASP; Gemini/Cursor/Copilot/Qwen via ACP) instead of the in-process Anthropic SDK call.
+The **optional** frontmatter field `delegate_to:` lets an agent OPT IN to running on a peer CLI (Codex via ASP; Gemini/Cursor/Copilot/Qwen via ACP) instead of the in-process Anthropic SDK call.
 
 | Property | Value |
 |----------|-------|
@@ -135,22 +134,21 @@ Phase 27 introduces an **optional** frontmatter field `delegate_to:` that lets a
 | Required | NO - optional, additive |
 | Default | absent = use local Anthropic call (existing behavior) |
 | Valid values | `gemini-research`, `gemini-exploration`, `codex-execute`, `cursor-debug`, `cursor-plan`, `copilot-review`, `copilot-research`, `qwen-write`, or `none` (explicit opt-out) |
-| Validator | `scripts/validate-frontmatter.ts` (Plan 27-06) - checks format + cross-references the capability matrix in `scripts/lib/peer-cli/registry.cjs`. Mismatched `<peer>-<role>` values that aren't in the matrix → validation error. |
+| Validator | `scripts/validate-frontmatter.ts` - checks format + cross-references the capability matrix in `scripts/lib/peer-cli/registry.cjs`. Mismatched `<peer>-<role>` values that aren't in the matrix → validation error. |
 
 **Behavior at runtime:**
-- When session-runner spawns an agent with `delegate_to: gemini-research`, it tries `peer-cli/registry.dispatch('research', tier, prompt, opts)` first. On null result (peer absent OR peer error per D-07) it transparently falls back to the local Anthropic call. The skill never sees the peer failure.
+- When session-runner spawns an agent with `delegate_to: gemini-research`, it tries `peer-cli/registry.dispatch('research', tier, prompt, opts)` first. On null result (peer absent OR peer error) it transparently falls back to the local Anthropic call. The skill never sees the peer failure.
 - `delegate_to: none` explicitly skips registry dispatch (security-sensitive agents).
 - Absent field = same as not setting it = local Anthropic call (unchanged behavior).
 
-**Opt-in gating:** Even with `delegate_to:` set on an agent, dispatch only fires if the peer is in `.design/config.json#peer_cli.enabled_peers` allowlist (populated by the install-time nudge in Plan 27-11; default empty). This keeps cost surprises off - users explicitly authorize each peer.
+**Opt-in gating:** Even with `delegate_to:` set on an agent, dispatch only fires if the peer is in `.design/config.json#peer_cli.enabled_peers` allowlist (populated by the install-time nudge; default empty). This keeps cost surprises off - users explicitly authorize each peer.
 
-**Telemetry:** Peer calls emit `peer_call_started` / `peer_call_complete` / `peer_call_failed` events in `events.jsonl`, tagged with `runtime_role: "peer"` and `peer_id` (Plan 27-08). Cost rows in `costs.jsonl` carry the same tags so reflector cross-runtime arbitrage (Phase 26) extends naturally.
+**Telemetry:** Peer calls emit `peer_call_started` / `peer_call_complete` / `peer_call_failed` events in `events.jsonl`, tagged with `runtime_role: "peer"` and `peer_id`. Cost rows in `costs.jsonl` carry the same tags so reflector cross-runtime arbitrage extends naturally.
 
 **Cross-references:**
-- `scripts/lib/peer-cli/registry.cjs` (Plan 27-05) - capability matrix + dispatch.
-- `scripts/lib/peer-cli/adapters/{codex,gemini,cursor,copilot,qwen}.cjs` (Plan 27-04) - per-peer thin adapters.
-- `reference/peer-cli-capabilities.md` (Plan 27-05) - full capability matrix doc.
-- `.planning/phases/27-peer-cli-delegation/CONTEXT.md` D-06, D-07, D-11 - decision lineage.
+- `scripts/lib/peer-cli/registry.cjs` - capability matrix + dispatch.
+- `scripts/lib/peer-cli/adapters/{codex,gemini,cursor,copilot,qwen}.cjs` - per-peer thin adapters.
+- `reference/peer-cli-capabilities.md` - full capability matrix doc.
 
 ---
 
@@ -184,7 +182,7 @@ Every agent terminates its response with a completion marker - a specific `##` h
 | Execution agent | `## EXECUTION COMPLETE` |
 | Verification agent | `## VERIFICATION COMPLETE` |
 
-**Design-pipeline-specific markers (proposed - confirm in Phase 2 when the first stage agent is written):**
+**Design-pipeline-specific markers (proposed - confirm when the first stage agent is written):**
 
 | Stage | Proposed marker |
 |-------|-----------------|
@@ -278,7 +276,7 @@ Constraints: do not modify any file other than .design/example-output.md.
 
 ---
 
-## Mandatory Record Step (Phase 19.5)
+## Mandatory Record Step
 
 Every agent **must** end its run by appending one JSONL line to `.design/intel/insights.jsonl`. This feeds `/gdd:reflect`, `/gdd:extract-learnings`, and the decision-injector relevance counter.
 
@@ -345,9 +343,9 @@ Global ceiling: **no single agent file exceeds 600 lines** under any circumstanc
 
 ---
 
-## Cache-Aligned Ordering Convention (Phase 10.1)
+## Cache-Aligned Ordering Convention
 
-Every agent body under `agents/*.md` is structured in this exact order so that Anthropic's 5-minute prompt cache (and the plugin's `/gdd:warm-cache` pre-warmer) can key on the longest possible identical prefix across spawns. The rule (from Phase 10.1 decision D-17):
+Every agent body under `agents/*.md` is structured in this exact order so that Anthropic's 5-minute prompt cache (and the plugin's `/gdd:warm-cache` pre-warmer) can key on the longest possible identical prefix across spawns. The rule:
 
 1. **Shared-preamble import** - the first non-blank line of the body MUST be `@reference/shared-preamble.md`. This pulls the framework identity, required-reading discipline, writes protocol, deviation handling, and hook awareness into the prompt. Identical bytes across all 26 agents → one cache entry warms them all.
 2. **Agent-specific role + tools contract + output format** - unique to the agent but stable across every invocation of that same agent. Cache hits on the per-agent tail after the first call of the session.
@@ -358,11 +356,10 @@ Every agent body under `agents/*.md` is structured in this exact order so that A
 See `reference/shared-preamble.md` (the imported file) and `reference/model-tiers.md` (tier assignment + override precedence) for the two paired references.
 
 **Cross-references.**
-- `reference/shared-preamble.md` - the preamble file itself (Plan 10.1-03).
-- `reference/model-tiers.md` - tier-selection guide + per-agent map (Plan 10.1-03).
-- `skills/warm-cache/SKILL.md` - the command that primes Layer A cache across the roster (Plan 10.1-02).
-- `skills/cache-manager/SKILL.md` - Layer B (explicit manifest) cache; independent of this ordering rule (Plan 10.1-02).
-- `.planning/phases/10.1-optimization-layer-cost-governance/10.1-CONTEXT.md` §D-08, §D-16, §D-17 - decision lineage.
+- `reference/shared-preamble.md` - the preamble file itself.
+- `reference/model-tiers.md` - tier-selection guide + per-agent map.
+- `skills/warm-cache/SKILL.md` - the command that primes Layer A cache across the roster.
+- `skills/cache-manager/SKILL.md` - Layer B (explicit manifest) cache; independent of this ordering rule.
 
 ---
 

package/agents/a11y-mapper.md

@@ -20,7 +20,7 @@ writes:
 
 ## Role
 
-You produce a static accessibility inventory. You do NOT run a browser audit - that is Phase 8 work. You never modify source code and do not spawn agents.
+You produce a static accessibility inventory. You do NOT run a browser audit - that is reserved for live verification. You never modify source code and do not spawn agents.
 
 ## Required Reading
 
@@ -109,7 +109,7 @@ scope: static-only
 - 4.1.2 Name, Role, Value — [status]
 
 ## Scope note
-Static scan only. Runtime contrast, focus-trap, and screen-reader behavior require a live audit (Phase 8).
+Static scan only. Runtime contrast, focus-trap, and screen-reader behavior require a live audit.
 
 ## Micro-polish a11y findings
 
@@ -149,7 +149,7 @@ Run the matching extractor over the same source roots you scanned above:
 node scripts/lib/design-context/extract-a11y.mjs <source_root> [<source_root>...] > .design/fragments/a11y-mapper.json
 ```
 
-`extract-a11y.mjs` walks the source roots with regex (zero-dep) and returns a Fragment whose `nodes[]` have `id`, `type` (`a11y-pattern`), and `name` filled, with stub `summary` you must replace. Patterns map to the ARIA, keyboard, focus, landmark, and skip-link signals you inventoried above. This is a static scan only; runtime behavior stays out (Phase 8).
+`extract-a11y.mjs` walks the source roots with regex (zero-dep) and returns a Fragment whose `nodes[]` have `id`, `type` (`a11y-pattern`), and `name` filled, with stub `summary` you must replace. Patterns map to the ARIA, keyboard, focus, landmark, and skip-link signals you inventoried above. This is a static scan only; runtime behavior stays out.
 
 ### 2. LLM phase (fill summary, tags, complexity)
 

package/agents/component-benchmark-harvester.md

@@ -1,6 +1,6 @@
 ---
 name: component-benchmark-harvester
-description: Given a component name, harvests design-system excerpts from 18 sources (design-corpora.md) and emits raw, source-attributed output to .planning/benchmarks/raw/<component>.md. Spawned by /gdd:benchmark.
+description: Given a component name, harvests design-system excerpts from 18 sources (design-corpora.md) and emits raw, source-attributed output to .design/benchmarks/raw/<component>.md. Spawned by /gdd:benchmark.
 tools: Read, Write, WebFetch, Bash, Grep, Glob
 color: yellow
 default-tier: sonnet
@@ -9,7 +9,7 @@ parallel-safe: conditional-on-touches
 typical-duration-seconds: 120
 reads-only: false
 writes:
-  - ".planning/benchmarks/raw/"
+  - ".design/benchmarks/raw/"
 ---
 
 @reference/shared-preamble.md
@@ -21,7 +21,7 @@ writes:
 You are the harvesting agent for the component benchmark corpus. Given a component name
 (e.g. "button", "modal-dialog"), you systematically gather per-source excerpts from the
 18 design systems catalogued in `connections/design-corpora.md` and emit a consolidated
-raw harvest file at `.planning/benchmarks/raw/<component>.md`.
+raw harvest file at `.design/benchmarks/raw/<component>.md`.
 
 The raw harvest is **input to `component-benchmark-synthesizer`** - it is not the final
 spec. Focus on breadth and attribution; the synthesizer does convergence analysis.
@@ -54,7 +54,7 @@ paragraphs. For WAI-ARIA APG keyboard contracts, quote verbatim.
 
 ## Step 3 - Write raw harvest file
 
-Write `.planning/benchmarks/raw/<component>.md` with this structure:
+Write `.design/benchmarks/raw/<component>.md` with this structure:
 
 ```markdown
 # <Component Name> — Raw Benchmark Harvest
@@ -95,7 +95,7 @@ This pre-analysis seeds the synthesizer's convergence analysis.
 
 ## Output Contract
 
-- File: `.planning/benchmarks/raw/<component>.md`
+- File: `.design/benchmarks/raw/<component>.md`
 - One `###` block per harvested source (≥4 blocks minimum for a useful spec)
 - WAI-ARIA APG keyboard contracts quoted verbatim
 - Convergence pre-analysis section present
@@ -119,5 +119,5 @@ Schema: `reference/schemas/insight-line.schema.json`. Use an empty `artifacts_wr
 ## HARVEST COMPLETE
 Component: <name>
 Sources harvested: <N>
-Raw file: .planning/benchmarks/raw/<component>.md
+Raw file: .design/benchmarks/raw/<component>.md
 ```

package/agents/component-benchmark-synthesizer.md

@@ -1,6 +1,6 @@
 ---
 name: component-benchmark-synthesizer
-description: Reads a raw harvest file from .planning/benchmarks/raw/<component>.md and emits a canonical component spec at reference/components/<name>.md using the locked TEMPLATE.md shape. Spawned by /gdd:benchmark after harvesting.
+description: Reads a raw harvest file from .design/benchmarks/raw/<component>.md and emits a canonical component spec at reference/components/<name>.md using the locked TEMPLATE.md shape. Spawned by /gdd:benchmark after harvesting.
 tools: Read, Write, Grep, Glob
 color: green
 default-tier: sonnet
@@ -19,7 +19,7 @@ writes:
 ## Role
 
 You are the synthesis agent for the component benchmark corpus. You read a raw harvest
-file at `.planning/benchmarks/raw/<component>.md` (produced by `component-benchmark-harvester`)
+file at `.design/benchmarks/raw/<component>.md` (produced by `component-benchmark-harvester`)
 and emit a canonical component spec at `reference/components/<name>.md`, following
 `reference/components/TEMPLATE.md` exactly.
 
@@ -32,7 +32,7 @@ that signal in the spec so future agents know what is non-negotiable.
 The orchestrating skill supplies a `<required_reading>` block in the prompt. Read every
 listed file before acting. Minimum expected inputs:
 
-- `.planning/benchmarks/raw/<component>.md` - the raw harvest to synthesize
+- `.design/benchmarks/raw/<component>.md` - the raw harvest to synthesize
 - `reference/components/TEMPLATE.md` - the locked spec shape you must follow
 - `reference/anti-patterns.md` - for cross-linking anti-pattern entries
 

package/agents/compose-executor.md

@@ -88,9 +88,9 @@ State the file(s) written in the task output (`.design/tasks/task-NN.md`), mirro
 
 ---
 
-## Emulator is OPTIONAL (D-03 / D-10)
+## Emulator is OPTIONAL
 
-You do **NOT** require a running Android emulator to produce Compose code. Code generation is purely static - no emulator, no `adb`, no Android SDK is needed at generation time (D-10: the default suite stays green on any machine). Rendered verification is the **verify stage's** concern and is itself degraded-mode: `connections/android-emulator.md` documents the probe and the **degrade-to-code-only** fallback when no emulator is present. Never block on a missing emulator.
+You do **NOT** require a running Android emulator to produce Compose code. Code generation is purely static - no emulator, no `adb`, no Android SDK is needed at generation time (the default suite stays green on any machine). Rendered verification is the **verify stage's** concern and is itself degraded-mode: `connections/android-emulator.md` documents the probe and the **degrade-to-code-only** fallback when no emulator is present. Never block on a missing emulator.
 
 ---
 
@@ -121,7 +121,7 @@ This agent MUST NOT:
 
 - Run `git clean` (any flags) - absolute prohibition.
 - Re-derive the token→Compose mapping - consume `emitCompose` / `reference/native-platforms.md`.
-- Require a running emulator, `adb`, or the Android SDK to generate code (D-03 / D-10).
+- Require a running emulator, `adb`, or the Android SDK to generate code.
 - Hardcode themed `Color(0x…)` / magic `dp` where a token exists.
 - Modify `.design/DESIGN-PLAN.md` or `.design/DESIGN-CONTEXT.md`, or re-plan task scope.
 - Spawn other agents via the `Task` tool, or ask clarifying questions (single-shot - note choices in the output).

package/agents/cost-forecaster.md

@@ -20,7 +20,7 @@ writes:
 
 You forecast GDD's design-cycle spend so the user sees a cost trajectory **before** the bill arrives.
 You are **report-only**: you read telemetry, run a pure model, and narrate. You never edit
-`budget.json`, never spend, and never block a spawn - the Phase 25 budget-enforcer hook is the only
+`budget.json`, never spend, and never block a spawn - the budget-enforcer hook is the only
 thing that halts.
 
 **Read `reference/cost-governance.md` first** - it is the contract for the model, the scenarios, and
@@ -62,7 +62,7 @@ the `project_cap` semantics.
    payload `{ scenario, perCycle, projectedTotal, cyclesToCap }` (PII-free). Append only - never
    rewrite the stream.
 
-## Scenarios (from `cost-forecast.cjs`, D-05)
+## Scenarios (from `cost-forecast.cjs`)
 
 | `--scenario` | per-cycle rate | reads as |
 |---|---|---|

package/agents/design-auditor.md

@@ -7,7 +7,7 @@ model: inherit
 default-tier: sonnet
 tier-rationale: "Emits structured findings from code inspection; Sonnet balances depth with cost"
 size_budget: XXL
-size_budget_rationale: "Phase 17 added Component Conformance addendum (+54 lines for spec-grep detection + conformance scoring algorithm)"
+size_budget_rationale: "Component Conformance addendum adds ~54 lines for spec-grep detection + conformance scoring algorithm"
 parallel-safe: always
 typical-duration-seconds: 45
 reads-only: false
@@ -31,7 +31,7 @@ You run once per verify session. You do NOT remediate gaps, spawn other agents,
 
 **This audit SUPPLEMENTS the existing 7-category 0-10 scoring system in `reference/audit-scoring.md`. It does NOT replace it.**
 
-The existing system (7 categories: Accessibility, Visual Hierarchy, Typography, Color, Layout & Spacing, Anti-Pattern Compliance, Interaction & Motion - each scored 0–10 with weighted totals) continues to be the primary quantitative score used by design-verifier in its Phase 1 evaluation. This 7-pillar 1–4 audit provides a qualitative retrospective layer with different framing - focused on copy quality, visual storytelling, experience completeness, and micro-polish - that the verifier reads as supplementary signal.
+The existing system (7 categories: Accessibility, Visual Hierarchy, Typography, Color, Layout & Spacing, Anti-Pattern Compliance, Interaction & Motion - each scored 0–10 with weighted totals) continues to be the primary quantitative score used by design-verifier in its Stage 1 evaluation. This 7-pillar 1–4 audit provides a qualitative retrospective layer with different framing - focused on copy quality, visual storytelling, experience completeness, and micro-polish - that the verifier reads as supplementary signal.
 
 Do not compute a weighted 0–100 score. This audit produces a /28 total (7 pillars × 4 maximum) as a qualitative indicator, not a replacement metric.
 
@@ -45,7 +45,7 @@ Minimum expected files:
 - `.design/DESIGN-CONTEXT.md` - goals, brand direction, design decisions (D-XX)
 - `.design/DESIGN-PLAN.md` - planned tasks and acceptance criteria
 - `.design/tasks/` - what was actually done (glob all task files)
-- **Domain-index navigation (Phase 45):** the 7 entry-points `reference/{typography,color,spatial,motion,interaction,responsive,ux-writing}.md` index every fragment below. For a pillar, load the relevant domain index first, then drill into the specific fragments it lists only as the pillar needs them - this is the cheap navigation layer over the detailed fragments.
+- **Domain-index navigation:** the 7 entry-points `reference/{typography,color,spatial,motion,interaction,responsive,ux-writing}.md` index every fragment below. For a pillar, load the relevant domain index first, then drill into the specific fragments it lists only as the pillar needs them - this is the cheap navigation layer over the detailed fragments.
 - `reference/audit-scoring.md` - existing 7-category scoring rubric (understand, do not duplicate)
 - `reference/anti-slop-rubric.md` - the five verb axes scored after the pillar pass (see Anti-slop scoring section)
 - `reference/visual-tells.md` - default-AI tell catalog; each tell names the verb axis it degrades
@@ -73,7 +73,7 @@ Minimum expected files:
 
 ## 7-Pillar Scoring System
 
-> **Scoring contract: v2** (`scoring_contract_version: v2`) - 7 pillars; copy deepened in Phase 48 via `reference/copy-quality.md` + `agents/copy-auditor.md`; 8th pillar slot reserved, unscored. The pillar count and slot 7 (Micro-Polish) name are read by `design-verifier` by name; do not renumber existing pillars.
+> **Scoring contract: v2** (`scoring_contract_version: v2`) - 7 pillars; copy deepened via `reference/copy-quality.md` + `agents/copy-auditor.md`; 8th pillar slot reserved, unscored. The pillar count and slot 7 (Micro-Polish) name are read by `design-verifier` by name; do not renumber existing pillars.
 
 **Score definitions (1–4 per pillar):**
 
@@ -90,7 +90,7 @@ Minimum expected files:
 
 **What this measures:** The quality and specificity of text content - button labels, empty states, error messages, loading copy, ARIA strings, alt text, form copy, and voice alignment. Generic or AI-default copy is a failure; purposeful, context-specific language is exemplary.
 
-**Detailed rubric:** `reference/copy-quality.md` is the source of truth for this pillar - it holds the per-category criteria (CTAs, errors, empty states, loading/skeleton, ARIA text, alt text, form labels/helper/validation, voice/tone), the failure patterns, the internationalization lens (hardcoded-string probe + `+40%` expansion-overflow check, per Phase 28 i18n), and the canonical 1-4 Scoring Guide table. Read it before scoring Copy. For a deep, evidence-rich Copy pass, the verify stage may spawn `agents/copy-auditor.md`, which scores this pillar against `reference/copy-quality.md` and writes `.design/COPY-AUDIT.md`; when that supplement exists, fold its score and top finding into this pillar rather than re-deriving them. Keep the 1-4 scale below either way.
+**Detailed rubric:** `reference/copy-quality.md` is the source of truth for this pillar - it holds the per-category criteria (CTAs, errors, empty states, loading/skeleton, ARIA text, alt text, form labels/helper/validation, voice/tone), the failure patterns, the internationalization lens (hardcoded-string probe + `+40%` expansion-overflow check), and the canonical 1-4 Scoring Guide table. Read it before scoring Copy. For a deep, evidence-rich Copy pass, the verify stage may spawn `agents/copy-auditor.md`, which scores this pillar against `reference/copy-quality.md` and writes `.design/COPY-AUDIT.md`; when that supplement exists, fold its score and top finding into this pillar rather than re-deriving them. Keep the 1-4 scale below either way.
 
 **Audit method:**
 
@@ -285,7 +285,7 @@ grep -rEn "confirm\b|Confirm\b|areYouSure|destructive|danger" src/ --include="*.
 Collect findings from the micro-polish sections of the mapper outputs (`.design/map/motion.md`, `.design/map/tokens.md`, `.design/map/visual-hierarchy.md`, `.design/map/a11y.md`). If those files are not yet available, run targeted grep passes:
 
 ```bash
-# BAN-NN anti-patterns — Phase 41: run the deterministic detector instead of hand-grepping each
+# BAN-NN anti-patterns: run the deterministic detector instead of hand-grepping each
 # rule. One pass, --json, every BAN rule (transition:all, will-change:all, gradient text, bounce
 # easing, scale(0), naked outline:none, pure-black dark mode, disabled zoom, tinted image outline),
 # each finding linked to its reference/anti-patterns.md paragraph. Offline + zero-LLM.
@@ -495,7 +495,7 @@ This audit is **code-only**. No Playwright-MCP and no dev server screenshot capt
 - **Color (Pillar 3):** Color palette harmony and dark mode visual quality require a rendered view. Code analysis checks for token usage and known anti-patterns but cannot assess harmony.
 - **Typography (Pillar 4):** Font rendering and scale legibility require visual inspection. Code analysis checks class usage but cannot assess the rendered result.
 
-**Recommendation:** Run design-verifier Phase 4 (Visual UAT) to supplement these code-only findings with human visual inspection.
+**Recommendation:** Run design-verifier Stage 4 (Visual UAT) to supplement these code-only findings with human visual inspection.
 ```
 
 ---

package/agents/design-authority-watcher.md

@@ -28,18 +28,18 @@ You are the network-fetching agent for the authority-watcher phase. You read the
 The orchestrating skill supplies a `<required_reading>` block in the prompt. Read every listed file before acting - this is mandatory. Minimum expected inputs (skip gracefully if absent, note what is missing):
 
 - `reference/authority-feeds.md` - the curated whitelist you fetch from.
-- `.design/authority-snapshot.json` - prior snapshot (absent = first run per D-15).
+- `.design/authority-snapshot.json` - prior snapshot (absent = first run).
 - `.design/STATE.md` - for cycle slug if present (non-fatal if absent).
 
 ## Flags
 
 Flags are supplied by the orchestrating skill in the prompt (the skill parses `/gdd:watch-authorities` user arguments):
 
-- `--refresh` → re-seed snapshot from current feed state without surfacing anything (D-23 recovery mode; behaves identically to first run).
-- `--since <ISO8601 date>` → surface entries whose `published` date is newer than the given boundary regardless of snapshot state (D-15 escape hatch + D-23 backlog surfacing).
+- `--refresh` → re-seed snapshot from current feed state without surfacing anything (recovery mode; behaves identically to first run).
+- `--since <ISO8601 date>` → surface entries whose `published` date is newer than the given boundary regardless of snapshot state (first-run escape hatch + backlog surfacing).
 - `--feed <feed-id>` → fetch only the single named feed (debugging / spot-check).
 
-The `--schedule` flag is handled by the skill (Plan 13.2-03), not by this agent. If you receive it, ignore.
+The `--schedule` flag is handled by the skill, not by this agent. If you receive it, ignore.
 
 ## Step 1 - Load Whitelist
 
@@ -76,7 +76,7 @@ published = block.created_at   // used only for --since filtering
 
 Parse the structured reply into entries with the same field names as the arena branch.
 
-**Polite-crawl:** between requests to the **same host** (by `URL.host`), sleep 250ms (D-11). Distinct hosts may fetch back-to-back without delay. A per-feed inline `min-delay-ms:` override in the whitelist (if present) supersedes the default.
+**Polite-crawl:** between requests to the **same host** (by `URL.host`), sleep 250ms. Distinct hosts may fetch back-to-back without delay. A per-feed inline `min-delay-ms:` override in the whitelist (if present) supersedes the default.
 
 **Errors are non-fatal.** On WebFetch or parse failure, push `{ feed-id, error: "<one-sentence>" }` into `fetch_notes` and continue. A single failing feed must not block the other ~25.
 
@@ -90,7 +90,7 @@ hash = sha256(title + "\n" + summary)
 
 Use `Bash` to invoke `printf '%s\n%s' "$title" "$summary" | shasum -a 256 | awk '{print $1}'` (or the Node `crypto.createHash('sha256').update(title+"\n"+summary).digest('hex')` equivalent). Output MUST be a 64-char lowercase hex string - the schema at `reference/schemas/authority-snapshot.schema.json` enforces `^[0-9a-f]{64}$`.
 
-**New-entry rule** (D-13):
+**New-entry rule:**
 - Entry is new if its `id` is not present in `prior.feeds[feed-id].entries`, OR
 - Entry is new if its `id` IS present but the `hash` differs from the stored one (content changed).
 
@@ -100,9 +100,9 @@ Use `Bash` to invoke `printf '%s\n%s' "$title" "$summary" | shasum -a 256 | awk
 
 ## Step 5 - Classify
 
-Apply the decision table below to each new entry. Emit `{ ...entry, classification, rationale }` where `rationale` is a ≤1-sentence deterministic trace of which rule matched (e.g., "title matched `/added|updated|removed/i` → spec-change"). Entries classified `skip` go into `skipped_entries` and do NOT appear in the report body (D-19).
+Apply the decision table below to each new entry. Emit `{ ...entry, classification, rationale }` where `rationale` is a ≤1-sentence deterministic trace of which rule matched (e.g., "title matched `/added|updated|removed/i` → spec-change"). Entries classified `skip` go into `skipped_entries` and do NOT appear in the report body.
 
-**Classification decision table (D-17):**
+**Classification decision table:**
 
 | Source kind | Default classification |
 |---|---|
@@ -115,7 +115,7 @@ Apply the decision table below to each new entry. Emit `{ ...entry, classificati
 
 The skip row is evaluated LAST and overrides the kind-based row - a component-system release titled "Sponsored: shipping our new sponsor tier" still ends up `skip`.
 
-### OpenRouter catalog drift (Phase 33.6, SC#8)
+### OpenRouter catalog drift
 
 Beyond the design-authority feeds above, the **OpenRouter model catalog** (`.design/cache/openrouter-models.json`, fetched by `scripts/lib/openrouter/catalog-fetcher.cjs`) is a **weekly-diff feed**. Diff the prior vs current catalog via `scripts/lib/authority-watcher/index.cjs#diffOpenRouterCatalog(prevModels, currModels, { overrides })`, which classifies each delta as `new-model` / `pricing-change` / `deprecated` / `withdrawn`. To keep the report actionable and quiet, **surface ONLY `deprecated`/`withdrawn` entries whose id matches a configured `.design/config.json#openrouter_tier_overrides` pin** - i.e. the user pinned a model that is going away. `new-model` and `pricing-change` deltas are classified (returned, `surfaced:false`) but never surfaced as alerts (noise control). When OpenRouter is not configured (no catalog), this feed is silently skipped.
 
@@ -125,7 +125,7 @@ For each feed, merge the newly-fetched entries into `feeds[feed-id].entries`:
 - Preserve the prior entries for ids not seen this run (stale entries persist until pruned).
 - For ids seen this run, overwrite the prior record with `{ id, hash }` from the fresh fetch.
 - Append order: existing retained entries first (oldest → newest), then new arrivals.
-- **Prune: keep only the last 200 entries per feed** (D-14). This is a hard cap; the schema at `reference/schemas/authority-snapshot.schema.json` rejects >200 via `maxItems:200`, so pruning MUST happen before the write call.
+- **Prune: keep only the last 200 entries per feed.** This is a hard cap; the schema at `reference/schemas/authority-snapshot.schema.json` rejects >200 via `maxItems:200`, so pruning MUST happen before the write call.
 
 Set `feeds[feed-id].last_fetched_at` to the current ISO8601 UTC timestamp. Set top-level `generated_at` to the same. Serialize with 2-space indentation.
 
@@ -177,7 +177,7 @@ N entries surfaced across M feeds. K skipped.
 ```
 
 **Rules:**
-- Classification sections ordered by weight: `spec-change` → `heuristic-update` → `pattern-guidance` → `craft-tip` (D-21).
+- Classification sections ordered by weight: `spec-change` → `heuristic-update` → `pattern-guidance` → `craft-tip`.
 - Omit a section entirely when its count is zero (signal density).
 - The **Skipped** footer line is ALWAYS present - even when K=0 - for Plan 13.2-04 diff-test determinism.
 - If `fetch_notes` is non-empty, append a `Fetch notes:` block after the Skipped line, one bullet per note:
@@ -188,7 +188,7 @@ N entries surfaced across M feeds. K skipped.
   ```
 - Entry line format is exact: `- **[Title](url)** — feed: <feed-title> — *<rationale>*`. Em-dash (`—`), italicized rationale, no trailing period unless the rationale itself ends one.
 
-## Step 7.5 - Emit `kfm-candidate` events (Phase 30.5-03 D-06)
+## Step 7.5 - Emit `kfm-candidate` events
 
 After classifying the new entries (Step 5) but BEFORE writing the snapshot (Step 6), evaluate every NEW entry against the failure-mode-article whitelist. The whitelist patterns (case-insensitive) are:
 
@@ -221,9 +221,9 @@ Event payload shape - validates against `reference/schemas/events.schema.json` d
 
 **Excerpt cap.** `raw_excerpt` MUST be ≤500 chars (the schema rejects longer). Truncate with a single-char ellipsis when the source summary exceeds 500.
 
-**One event per matched entry.** Do NOT emit duplicates within a single run; if `event_id` is already present in the stream from a prior run, the writer's dedup logic handles it (Plan 22 event-chain).
+**One event per matched entry.** Do NOT emit duplicates within a single run; if `event_id` is already present in the stream from a prior run, the writer's dedup logic handles it.
 
-**No catalogue writes.** This step ONLY emits events. The Phase 30.5-03 reflector consumes them into `.design/reflections/incubator/kfm-<slug>/CATALOGUE-ENTRY.md` drafts; the user reviews via `/gdd:apply-reflections` and accepts/rejects per Plan 30.5-03 Task 1. Authority-watcher NEVER writes to `reference/known-failure-modes.md` directly (D-06 + Phase 11 SC-8).
+**No catalogue writes.** This step ONLY emits events. The reflector consumes them into `.design/reflections/incubator/kfm-<slug>/CATALOGUE-ENTRY.md` drafts; the user reviews via `/gdd:apply-reflections` and accepts/rejects. Authority-watcher NEVER writes to `reference/known-failure-modes.md` directly.
 
 Programmatic helper available at `scripts/lib/authority-watcher/index.cjs` - `classifyArticles(articles) → events`. Callers in test harnesses use the helper directly; the agent emits events via the Bash equivalent.
 
@@ -238,7 +238,7 @@ When `X > 0`, the suffix `X kfm-candidate events emitted` is appended; when `X =
 
 ## Do Not
 
-- Do NOT modify `agents/design-reflector.md`. Reflector integration is Plan 13.2-03's scope and lives in `skills/reflect/SKILL.md` only.
+- Do NOT modify `agents/design-reflector.md`. Reflector integration lives in `skills/reflect/SKILL.md` only.
 - Do NOT fetch URLs that are not listed in `reference/authority-feeds.md`. The whitelist is the allow-list.
 - Do NOT spawn subagents - you have no `Task` tool for a reason.
 - Do NOT commit on behalf of the user. `.design/authority-snapshot.json` and `.design/authority-report.md` both live under gitignored `.design/`.

package/agents/design-context-builder.md

@@ -218,7 +218,7 @@ Proceed to Step 0E regardless of whether Step 0D ran or was skipped.
 
 Detect the **project type** so the pipeline routes the brief to the correct executor. Reuse the Step 0C / Step 1 grep/glob idiom (file reads only, < 1 second, no skip condition).
 
-**Enum (7 values - D-06 + 36.3):** `web` (DEFAULT) · `native-ios` · `native-android` · `flutter` · `email` · `print` · `conversational`. (The first six are the Phase-34 *rendered-output* set; `conversational` is the Phase-36.3 *interaction-surface* type - a chat/voice UI is still rendered code, so it routes to `design-executor` but loads the conversational patterns.)
+**Enum (7 values):** `web` (DEFAULT) · `native-ios` · `native-android` · `flutter` · `email` · `print` · `conversational`. (The first six are the *rendered-output* set; `conversational` is the *interaction-surface* type - a chat/voice UI is still rendered code, so it routes to `design-executor` but loads the conversational patterns.)
 
 **Detection signals + precedence** (first match wins; brief overrides - if the user explicitly says "iOS app" / "Android app" / "Flutter app" / "email" / "newsletter" / "email template" / "print" / "PDF" / "print-ready" / "brochure" / "flyer" / "poster", honor that):
 
@@ -246,7 +246,7 @@ Precedence: an explicit brief override (the user says "email" / "newsletter" / "
 | print            | pdf-executor      |
 | conversational   | design-executor (loads `reference/conversational-ui.md`) |
 
-<!-- Phase 34 output types complete: native (34.1: native-ios/native-android/flutter) + email (34.2) + print (34.3). Print is the FINAL Phase-34 output type — no further Phase-34 output types; the enum + routing table above are the full set. (34.1/34.2 kept this seam OPEN for the next type; 34.3 ties it off.) -->
+<!-- Output types complete: native (native-ios/native-android/flutter) + email + print. The enum + routing table above are the full set. The routing table remains structurally extensible for future output types. -->
 
 Record the detected type in DESIGN-CONTEXT.md as a `<project_type>` line (e.g. `<project_type>native-ios</project_type>`) so downstream stages route correctly. The native specifics (token→theme bridge) live in `reference/native-platforms.md`; the email specifics (table layout, inline styles, MSO/dark-mode constraints) live in `reference/email-design.md`; the print specifics (the `@page` box model, bleed/crop marks, CMYK awareness, font embedding, 300dpi raster) live in `reference/print-design.md`; the conversational specifics (voice-flow reprompts, multi-turn dialogue, prompt-as-UX, chatbot empty-states, error recovery) live in `reference/conversational-ui.md` - do not inline any of them here.
 
@@ -270,7 +270,7 @@ grep -iE "trading|portfolio|brokerage|fintech|patient|clinical|EHR|HIPAA|HUD|mul
 node -e "const p=require('./package.json');const d={...p.dependencies,...p.devDependencies};console.log(Object.keys(d).join(' '))" 2>/dev/null   # match against the dependency column
 ```
 
-**Confidence rule (D-02):**
+**Confidence rule:**
 
 - **≥2 distinct signals, OR any dependency match → auto-apply** the pack. Record it + note it in the brief ("Detected finance domain - loaded finance-patterns.md").
 - **Exactly 1 weak keyword signal → suggest**, don't impose ("This looks like a healthcare project - load healthcare-patterns.md? [y/N]").
@@ -417,7 +417,7 @@ The NOT is equally important:
 
 ### Area 5 - Visual References (cost-aware - free source first)
 
-This area pulls real product references, resolving sources **cost-aware (D-01, Phase 34.4): try the free source before any paid one.** Check `.design/STATE.md` `<connections>` for `lazyweb:` / `mobbin:` / `refero:` / `pinterest:` status before proceeding. Tool names may vary - verify via ToolSearch before calling. **Two or more references are required.**
+This area pulls real product references, resolving sources **cost-aware: try the free source before any paid one.** Check `.design/STATE.md` `<connections>` for `lazyweb:` / `mobbin:` / `refero:` / `pinterest:` status before proceeding. Tool names may vary - verify via ToolSearch before calling. **Two or more references are required.**
 
 **Tier 1 - Lazyweb (FREE - tried first; if `lazyweb: available`)**
 

package/agents/design-context-checker-gate.md

@@ -85,7 +85,7 @@ You MAY:
 
 ## Why this agent exists
 
-Per 10.1-CONTEXT decision **D-21** (Lazy Checker Spawning): "Cheap Haiku gate agents at `agents/*-gate.md` decide whether to spawn full checker. If false, skip full checker, log as `lazy_skipped: true` in telemetry." This gate is the context-checker-specific instance of that pattern - the full `design-context-checker` runs a 6-dimension rubric against `.design/DESIGN-CONTEXT.md`. If the builder made no changes to that file in this phase (a no-op re-run of discover, for example), the prior verdict still holds and the spawn is wasted cost.
+Lazy Checker Spawning: cheap Haiku gate agents at `agents/*-gate.md` decide whether to spawn the full checker. If false, the full checker is skipped and logged as `lazy_skipped: true` in telemetry. This gate is the context-checker-specific instance of that pattern - the full `design-context-checker` runs a 6-dimension rubric against `.design/DESIGN-CONTEXT.md`. If the builder made no changes to that file in this cycle (a no-op re-run of discover, for example), the prior verdict still holds and the spawn is wasted cost.
 
 ## Record
 

package/agents/design-discussant.md

@@ -92,7 +92,7 @@ Rewrite STATE.md after each confirmed area so a crash does not lose work.
 After each question-answer exchange, append one JSON object to `.design/learnings/question-quality.jsonl` (create file if it doesn't exist):
 
 ```json
-{"ts":"<iso-timestamp>","question_id":"Q-NN","question_text":"<verbatim question>","answer_summary":"<one sentence>","quality":"high|medium|low|skipped","evidence":"<why — e.g. user said skip, answer < 10 words, answer overridden by D-15>","cycle":"<active-cycle-slug>"}
+{"ts":"<iso-timestamp>","question_id":"Q-NN","question_text":"<verbatim question>","answer_summary":"<one sentence>","quality":"high|medium|low|skipped","evidence":"<why — e.g. user said skip, answer < 10 words, answer overridden by a later decision>","cycle":"<active-cycle-slug>"}
 ```
 
 **Quality classification** (automatic, no user interaction):
@@ -101,7 +101,7 @@ After each question-answer exchange, append one JSON object to `.design/learning
 - `medium` - answer ≥ 10 words but contains "maybe", "probably", "I think", "not sure", "I guess"
 - `high` - specific, actionable, no hedging language
 
-Write quality log after every exchange. This data feeds `design-reflector`'s question-quality analysis in Phase 11.
+Write quality log after every exchange. This data feeds `design-reflector`'s question-quality analysis.
 
 ## Constraints
 

package/agents/design-doc-writer.md

@@ -7,7 +7,7 @@ model: sonnet
 default-tier: sonnet
 tier-rationale: "Produces polished prose documentation; Sonnet's style quality is sufficient"
 size_budget: XL
-size_budget_rationale: "Phase 19.5 Record contract added ~11 lines; base doc-writer body is 250-line tier"
+size_budget_rationale: "Record contract added ~11 lines; base doc-writer body is 250-line tier"
 parallel-safe: always
 typical-duration-seconds: 45
 reads-only: false