taketomarket 0.1.0

package/workflows/discipline/email-preflight.md

@@ -0,0 +1,150 @@
+<purpose>
+Pre-send deliverability, dark-mode compatibility, and spam-trigger scan for email content.
+Evaluates against email playbook gate definitions and generates a go/no-go recommendation.
+Single-pass analysis workflow per D-07.
+</purpose>
+
+<required_reading>
+@${CLAUDE_PLUGIN_ROOT}/references/context-loading.md
+@${CLAUDE_PLUGIN_ROOT}/playbooks/email.md
+</required_reading>
+
+<constraints>
+## POSITIONING.md is READ-ONLY
+
+**Do NOT modify `.marketing/POSITIONING.md` during this workflow.**
+
+POSITIONING.md is an architectural invariant. If you detect positioning drift:
+- Flag the issue and recommend running /ttm-positioning-check
+
+Only /ttm-positioning-shift and /ttm-init may modify POSITIONING.md.
+</constraints>
+
+<process>
+
+## Text-Mode Detection
+
+**Text mode (`--text` flag):** Set `TEXT_MODE=true` if `--text` is present in `$ARGUMENTS`
+or if the runtime is not Claude Code. When TEXT_MODE is active, replace every
+`AskUserQuestion` call with a plain-text numbered list.
+
+Detection:
+```bash
+if echo "$ARGUMENTS" | grep -q -- '--text'; then TEXT_MODE=true; fi
+```
+
+If `AskUserQuestion` tool is not available in the current runtime, set `TEXT_MODE=true`.
+
+---
+
+## Step 1: Load Context
+
+```
+takeToMarket > LOADING CONTEXT FOR EMAIL PREFLIGHT
+```
+
+**Tier 1 summaries** (lines 1 to `<!-- END_SUMMARY -->`) from all 9 `.marketing/` reference files.
+**Tier 2 (full):** `.marketing/BRAND.md` (voice/banned words check).
+**Playbook gates:** @${CLAUDE_PLUGIN_ROOT}/playbooks/email.md
+
+If `.marketing/POSITIONING.md` does not exist: Error and exit.
+
+---
+
+## Step 2: Get Email Content
+
+Ask: "Paste your email content (subject line, preview text, and body). Include HTML if available."
+
+Parse into: subject line, preview text, body content, and HTML structure (if provided).
+If user provides a file path, read the file. If subject/preview are not clearly marked,
+ask for clarification.
+
+---
+
+## Step 3: Run Email Preflight Checks
+
+```
+takeToMarket > RUNNING EMAIL PREFLIGHT CHECKS
+```
+
+Evaluate against 7 email discipline gates:
+
+**1. Subject Line Spam Triggers (DISC-EMAIL-01)**
+Check for: ALL CAPS words, excessive punctuation (!!!, ???), spam trigger phrases
+(FREE, ACT NOW, LIMITED TIME, GUARANTEED, etc.), multiple formatting red flags.
+- PASS: No spam triggers; clean formatting
+- WARN: 1 borderline trigger with mitigating context
+- FAIL: 2+ spam triggers, ALL CAPS subject, or excessive punctuation
+
+**2. Preview Text Optimization (DISC-EMAIL-05)**
+Preview text present, distinct from subject, 40-130 chars, compelling.
+- PASS: Present, 40-130 chars, distinct from subject
+- WARN: Present but too short/long or repeats subject
+- FAIL: No preview text specified
+
+**3. Dark Mode Compatibility (DISC-EMAIL-02)**
+If HTML: check for dark-mode-safe colors, no hardcoded white backgrounds, image alt-text.
+- PASS: No hard-coded light backgrounds; all images have alt text
+- WARN: Background colors may render poorly in dark mode
+- FAIL: White backgrounds without alternatives; image-dependent content
+
+**4. Unsubscribe/Address (DISC-EMAIL-03)**
+CAN-SPAM compliance: unsubscribe link and physical address present.
+- PASS: Both unsubscribe link and physical address clearly present
+- WARN: Present but buried or incomplete
+- FAIL: Missing unsubscribe or physical address
+
+**5. Content-to-Image Ratio (DISC-EMAIL-04)**
+Text-to-image ratio at least 60:40, alt text on all images.
+- PASS: Text dominant; fully readable without images
+- WARN: Roughly equal text/image; core message survives without images
+- FAIL: Image-heavy; image-only sections or missing alt text
+
+**6. Deliverability Signals (DISC-EMAIL-07)**
+Link count (<10 recommended), no URL shorteners, SPF/DKIM/DMARC reminder.
+- PASS: <10 links, no shorteners, proper sending domain
+- WARN: 10-15 links or minor deliverability concerns
+- FAIL: Excessive links, URL shorteners, or unconfigured domain
+
+**7. Brand Voice**
+Check against BRAND.md tone guidelines and banned words list.
+- PASS: Tone matches brand voice; no banned words
+- WARN: Minor tone deviation but no banned words
+- FAIL: Banned words present or tone contradicts brand
+
+Per check: assign PASS / WARN / FAIL with evidence.
+
+---
+
+## Step 4: Generate Report
+
+Determine verdict: **GO** (0 FAIL, <=2 WARN) | **CAUTION** (>2 WARN) | **NO-GO** (any FAIL)
+
+```
+========================================
+takeToMarket > EMAIL PREFLIGHT REPORT
+========================================
+Subject: [subject line]
+Date: [current date]
+Overall: [GO / CAUTION / NO-GO]
+
+| # | Check                  | Result | Detail                |
+|---|------------------------|--------|-----------------------|
+| 1 | Spam Triggers          | [P/W/F] | [detail]             |
+| 2 | Preview Text           | [P/W/F] | [detail]             |
+| 3 | Dark Mode              | [P/W/F] | [detail]             |
+| 4 | Unsubscribe/Address    | [P/W/F] | [detail]             |
+| 5 | Content-to-Image Ratio | [P/W/F] | [detail]             |
+| 6 | Deliverability Signals | [P/W/F] | [detail]             |
+| 7 | Brand Voice            | [P/W/F] | [detail]             |
+
+ISSUES (WARN/FAIL fixes):
+- [Check]: [Specific fix recommendation]
+
+VERDICT: [GO / CAUTION / NO-GO]
+- GO: 0 FAIL + <=2 WARN -- safe to send
+- CAUTION: >2 WARN -- review recommended before sending
+- NO-GO: any FAIL -- must fix before sending
+```
+
+</process>

package/workflows/discipline/keyword-map.md

@@ -0,0 +1,125 @@
+<purpose>
+Generate a keyword cluster map with intent tags for content planning. Groups keywords
+by topic cluster, assigns search intent (informational/transactional/navigational/
+commercial), and maps to content types and funnel stages.
+Single-pass analysis workflow per D-07.
+</purpose>
+
+<required_reading>
+@${CLAUDE_PLUGIN_ROOT}/references/context-loading.md
+@${CLAUDE_PLUGIN_ROOT}/playbooks/seo.md
+</required_reading>
+
+<constraints>
+## POSITIONING.md is READ-ONLY
+
+**Do NOT modify `.marketing/POSITIONING.md` during this workflow.**
+
+POSITIONING.md is an architectural invariant. If you detect positioning drift:
+- Flag the issue and recommend running /ttm-positioning-check
+
+Only /ttm-positioning-shift and /ttm-init may modify POSITIONING.md.
+</constraints>
+
+<process>
+
+## Text-Mode Detection
+
+**Text mode (`--text` flag):** Set `TEXT_MODE=true` if `--text` is present in `$ARGUMENTS`
+or if the runtime is not Claude Code. When TEXT_MODE is active, replace every
+`AskUserQuestion` call with a plain-text numbered list.
+
+Detection:
+```bash
+if echo "$ARGUMENTS" | grep -q -- '--text'; then TEXT_MODE=true; fi
+```
+
+If `AskUserQuestion` tool is not available in the current runtime, set `TEXT_MODE=true`.
+
+---
+
+## Step 1: Load Context
+
+```
+takeToMarket > LOADING CONTEXT FOR KEYWORD MAP
+```
+
+**Tier 1 summaries** (lines 1 to `<!-- END_SUMMARY -->`) from all 9 `.marketing/` reference files.
+**Tier 2 (full):** `.marketing/POSITIONING.md`, `.marketing/CHANNELS.md`, `.marketing/COMPETITORS.md`
+
+If `.marketing/POSITIONING.md` does not exist: Error and exit.
+
+---
+
+## Step 2: Gather Seed Keywords
+
+Ask the user:
+1. "What are your primary seed keywords or topics? (comma-separated)"
+2. "Any specific competitors to include in keyword gap analysis? (or 'use COMPETITORS.md')"
+3. "Target geography/language? (default: English, global)"
+
+If user says "use COMPETITORS.md", extract competitor names from the loaded file.
+
+MCP Detection: Attempt WebSearch for seed keyword variations.
+- **SEARCH_MODE=web:** Use WebSearch to expand seed keywords with related terms, "people also ask" patterns, and competitor keyword gaps.
+- **SEARCH_MODE=manual:** Work with user-provided seeds + positioning + ICP context to generate clusters from domain knowledge.
+
+---
+
+## Step 3: Generate Keyword Clusters
+
+For each seed keyword, generate clusters:
+- **Head terms** (1-2 words, high volume)
+- **Body terms** (2-3 words, medium volume)
+- **Long-tail terms** (4+ words, specific intent)
+
+Assign to each keyword:
+- **Intent tag:** informational / transactional / navigational / commercial-investigation
+- **Funnel stage:** awareness / consideration / decision / retention
+- **Content type:** blog / landing page / comparison / how-to / case study / tool
+- **Priority:** H (high) / M (medium) / L (low) based on intent alignment with ICP
+
+Group into topic clusters with pillar-cluster relationships.
+
+---
+
+## Step 4: Output Keyword Map
+
+Write to `.marketing/KEYWORD-MAP.md`:
+
+```markdown
+# Keyword Map
+Generated: [date]
+Seeds: [list]
+Geography: [target]
+
+## Cluster: [Topic 1]
+Pillar: [pillar keyword]
+
+| Keyword | Intent | Funnel | Content Type | Priority |
+|---------|--------|--------|--------------|----------|
+| [term]  | [info/trans/nav/comm] | [stage] | [type] | [H/M/L] |
+
+## Cluster: [Topic 2]
+...
+
+## Gap Analysis
+Keywords competitors rank for that are not covered:
+| Keyword | Competitor | Intent | Opportunity |
+|---------|-----------|--------|-------------|
+```
+
+---
+
+## Step 5: Completion
+
+```
+========================================
+takeToMarket > KEYWORD MAP COMPLETE
+========================================
+Clusters: [N] | Keywords: [total] | File: .marketing/KEYWORD-MAP.md
+```
+
+Display a summary table of clusters with keyword counts per cluster.
+
+</process>

package/workflows/discipline/repurpose.md

@@ -0,0 +1,329 @@
+<purpose>
+Fan out a long-form source asset into derivative assets across channels. Executes full
+brief-produce-verify per derivative with hero-first Task() orchestration. Each derivative
+gets a channel-adapted brief, fresh production context, and independent verification.
+This is NOT a lightweight command -- it runs the full production lifecycle per derivative.
+</purpose>
+
+<required_reading>
+@${CLAUDE_PLUGIN_ROOT}/references/context-loading.md
+@${CLAUDE_PLUGIN_ROOT}/agents/ttm-producer.md
+@${CLAUDE_PLUGIN_ROOT}/workflows/lifecycle/brief.md
+</required_reading>
+
+<constraints>
+## POSITIONING.md is READ-ONLY
+
+**Do NOT modify `.marketing/POSITIONING.md` during this workflow.**
+
+POSITIONING.md is an architectural invariant. If you detect positioning drift:
+- In verify: use the Escalate option to launch /ttm-positioning-shift
+- In other workflows: flag the issue and recommend running /ttm-positioning-check
+
+Only /ttm-positioning-shift and /ttm-init may modify POSITIONING.md.
+
+## Campaign Context Required
+
+This workflow MUST be run within an existing campaign context. The source asset must
+exist in a campaign's ASSETS/ directory with a campaign that is in phase: produced,
+verified, reviewed, fixed, or shipped.
+</constraints>
+
+<process>
+
+## Text-Mode Detection
+
+**Text mode (`--text` flag):** Set `TEXT_MODE=true` if `--text` is present in `$ARGUMENTS`
+or if the runtime is not Claude Code. When TEXT_MODE is active, replace every
+`AskUserQuestion` call with a plain-text numbered list.
+
+Detection:
+```bash
+if echo "$ARGUMENTS" | grep -q -- '--text'; then TEXT_MODE=true; fi
+```
+
+If `AskUserQuestion` tool is not available in the current runtime, set `TEXT_MODE=true`.
+
+When TEXT_MODE is active, replace each AskUserQuestion with a plain-text numbered list:
+```
+[HEADER]
+[QUESTION]
+  1. [OPTION_1_LABEL] -- [OPTION_1_DESCRIPTION]
+  2. [OPTION_2_LABEL] -- [OPTION_2_DESCRIPTION]
+  ...
+Type the number of your choice:
+```
+
+---
+
+## Step 1: Load Context
+
+```
+takeToMarket > LOADING REPURPOSE CONTEXT
+```
+
+Extract SOURCE_ASSET_PATH from $ARGUMENTS (strip `--text` flag if present):
+```bash
+SOURCE_ASSET_PATH=$(echo "$ARGUMENTS" | sed 's/--text//g' | xargs)
+```
+
+Read Tier 1 summaries from all 9 reference files (lines 1 to `<!-- END_SUMMARY -->`):
+- `.marketing/POSITIONING.md`
+- `.marketing/BRAND.md`
+- `.marketing/ICP.md`
+- `.marketing/CHANNELS.md`
+- `.marketing/STATE.md` (frontmatter only)
+- `.marketing/CALENDAR.md`
+- `.marketing/COMPETITORS.md`
+- `.marketing/METRICS.md`
+- `.marketing/LEARNINGS.md`
+
+Read Tier 2 (full content) for production context:
+- `.marketing/POSITIONING.md`
+- `.marketing/CHANNELS.md`
+- `.marketing/BRAND.md`
+- `.marketing/ICP.md`
+
+---
+
+## Step 2: Identify Source Asset
+
+Parse SOURCE_ASSET_PATH from $ARGUMENTS.
+
+### Validate Path (T-10-11 mitigation)
+
+Use path.resolve() equivalent to canonicalize the path, then verify:
+
+1. **File exists** at the specified path
+2. **Path is within `.marketing/CAMPAIGNS/`** directory -- reject any path that does not
+   resolve to a location under `.marketing/CAMPAIGNS/[slug]/ASSETS/`
+3. **Campaign is in appropriate phase** -- read the campaign's STATE.md and confirm phase
+   is one of: produced, verified, reviewed, fixed, or shipped
+
+Extract from the validated path and campaign data:
+- `CAMPAIGN_SLUG` -- extracted from the path (the directory name under CAMPAIGNS/)
+- `SOURCE_ASSET_ID` -- look up the source file in MANIFEST.json by matching the filename
+  against hero.file and derivatives[].file entries. Extract the matching asset_id.
+- `SOURCE_CHANNEL` -- from the matching MANIFEST.json asset entry's channel field
+- `SOURCE_CONTENT` -- read the full file content
+
+If validation fails at any step, display a specific error:
+- Path not found: "Source asset not found at [path]. Check the file path and try again."
+- Not in CAMPAIGNS: "Source asset must be within a .marketing/CAMPAIGNS/[slug]/ASSETS/ directory."
+- Wrong phase: "Campaign [slug] is in phase [phase]. Repurpose requires produced, verified, reviewed, fixed, or shipped."
+- Not in MANIFEST: "Source asset not found in MANIFEST.json. Was this asset produced by /ttm-produce?"
+
+---
+
+## Step 3: Select Target Channels
+
+Read `.marketing/CHANNELS.md` active channels list.
+
+Parse all active channels from CHANNELS.md. Remove SOURCE_CHANNEL from the list
+(do not repurpose to the same channel the source was produced for).
+
+Default selection: all remaining active channels.
+
+Ask user for channel selection (per D-12):
+
+```
+AskUserQuestion:
+  title: "Repurpose Target Channels"
+  question: "Repurpose to which channels? Default: all active channels minus source."
+  options:
+    - label: "All channels"
+      description: "[list all remaining channels]"
+    - label: "[channel 1]"
+      description: "Repurpose to [channel 1] only"
+    - label: "[channel 2]"
+      description: "Repurpose to [channel 2] only"
+    - label: "Custom selection"
+      description: "Enter comma-separated channel numbers"
+```
+
+If user selects "All channels" or accepts the default: use all remaining active channels.
+If user selects a specific channel: use only that channel.
+If user selects "Custom selection": collect comma-separated numbers and resolve to channels.
+
+### Identify Hero Channel (per D-13)
+
+From the selected target channels, identify the HERO_CHANNEL -- the channel with the
+highest reach baseline from CHANNELS.md metrics/baselines section.
+
+If baselines are not available or tied: use the first channel in the selected list.
+
+Store `HERO_CHANNEL` and `HERO_CHANNEL_SLUG` (URL-safe version of channel name).
+Store remaining channels as `REMAINING_CHANNELS`.
+
+---
+
+## Step 4: Generate Derivative Briefs
+
+```
+takeToMarket > GENERATING DERIVATIVE BRIEFS ([N] channels)
+```
+
+For each target channel, generate a channel-adapted derivative brief (per D-11).
+
+Read the source asset content and extract the core message/thesis.
+Cross-reference `.marketing/CHANNELS.md` for each target channel's configuration:
+- Format requirements (content type, structure)
+- Typical length/word count for the channel
+- Audience overlap with source channel
+
+Cross-reference `.marketing/BRAND.md` for tone-per-context adjustments if defined
+for the target channel.
+
+Write each brief with these sections:
+- **Header:** Derivative Brief title, source filename, campaign slug, source asset ID
+- **Adaptation Requirements:** Channel name, format (from CHANNELS.md), length, tone adjustment (from BRAND.md or "match source")
+- **Core Message:** Extract primary thesis, key claims, and CTA from source (must be preserved)
+- **Channel-Specific Adaptations:** Hook/opening, structure, length limits, CTA format, playbook rules
+- **Positioning Anchor:** Primary differentiator and proof points from POSITIONING.md (direct quote)
+- **Outcome Metric:** Inherit from campaign BRIEF.md, or channel-default from CHANNELS.md baselines
+
+Write each brief to: `.marketing/CAMPAIGNS/${CAMPAIGN_SLUG}/REPURPOSE-BRIEF-${CHANNEL_SLUG}.md`
+
+---
+
+## Step 5: Produce Hero Derivative
+
+```
+takeToMarket > REPURPOSING: HERO DERIVATIVE ([HERO_CHANNEL])
+```
+
+Read the agent prompt template from `${CLAUDE_PLUGIN_ROOT}/agents/ttm-producer.md`.
+
+Fill placeholders:
+- `[BRIEF_PATH]` --> `.marketing/CAMPAIGNS/${CAMPAIGN_SLUG}/REPURPOSE-BRIEF-${HERO_CHANNEL_SLUG}.md`
+- `[POSITIONING_PATH]` --> `.marketing/POSITIONING.md`
+- `[BRAND_PATH]` --> `.marketing/BRAND.md`
+- `[ICP_PATH]` --> `.marketing/ICP.md`
+- `[PLAYBOOK_PATH]` --> resolved playbook for HERO_CHANNEL (from `${CLAUDE_PLUGIN_ROOT}/playbooks/${CHANNEL_TYPE}.md`), or `"none"` if not found
+- `[OUTPUT_PATH]` --> `.marketing/CAMPAIGNS/${CAMPAIGN_SLUG}/ASSETS/R-01-${HERO_CHANNEL_SLUG}.md`
+- `[ASSET_TYPE]` --> derivative content type for this channel
+- `[CHANNEL]` --> HERO_CHANNEL name
+- `[HERO_PATH]` --> SOURCE_ASSET_PATH (the original source serves as the "hero" reference)
+
+Also inject a "Source Asset Reference" section into the prompt with SOURCE_CHANNEL
+and SOURCE_ASSET_PATH so the producer can reference the original content.
+
+Call Task() with the populated prompt.
+**WAIT** for Task() to complete before proceeding.
+
+After Task() returns, verify the hero derivative file exists and has content:
+```bash
+test -s ".marketing/CAMPAIGNS/${CAMPAIGN_SLUG}/ASSETS/R-01-${HERO_CHANNEL_SLUG}.md"
+```
+
+If the file is empty or missing:
+  Error: "Hero derivative production failed -- file not written by subagent."
+  Exit.
+
+---
+
+## Step 6: Produce Remaining Derivatives (Wave-Parallel)
+
+If only 1 target channel was selected (hero only): skip this step.
+
+```
+takeToMarket > REPURPOSING: DERIVATIVES ([N] channels, wave-parallel)
+```
+
+For each channel in REMAINING_CHANNELS:
+
+1. Read agent prompt template from `${CLAUDE_PLUGIN_ROOT}/agents/ttm-producer.md`
+
+2. Assign sequential filename: `R-[NN]-${CHANNEL_SLUG}.md` where NN starts at 02
+
+3. Fill all placeholders (same pattern as Step 5, with channel-specific values and
+   `[HERO_PATH]` pointing to the hero derivative `R-01-${HERO_CHANNEL_SLUG}.md`)
+4. Inject Source Asset Reference (same as Step 5)
+5. Call Task() with the populated prompt
+
+All derivative Task() calls run in parallel (per D-13 wave-parallel pattern).
+
+After ALL Task() calls complete, verify each output file exists and has content (> 50 chars):
+```bash
+test -s ".marketing/CAMPAIGNS/${CAMPAIGN_SLUG}/ASSETS/R-${NN}-${CHANNEL_SLUG}.md"
+```
+
+For any derivative that failed (file empty or missing):
+- Log warning: `takeToMarket > WARNING: Derivative R-${NN}-${CHANNEL_SLUG}.md failed`
+- Continue with remaining assets. Do NOT abort the repurpose run.
+
+Track successful and failed derivatives for the manifest and completion banner.
+
+---
+
+## Step 7: Verify Derivatives
+
+```
+takeToMarket > VERIFYING DERIVATIVES
+```
+
+For each successfully produced derivative, run simplified inline gate checks:
+
+Run 3 simplified gates per derivative (PASS / WARN / FAIL each):
+
+1. **Positioning Drift** -- maintains primary differentiator, claims trace to proof points, no must-not-say terms
+2. **Format Correctness** -- meets channel length constraints, has required structural elements (hook, CTA), follows channel conventions
+3. **Voice Drift** -- matches BRAND.md voice archetype, no banned words, tone consistent with source
+
+Record results as `VERIFY_RESULTS` array with per-derivative gate outcomes.
+
+If any FAIL: flag for user attention but do NOT auto-fix. Recommend `/ttm-fix` or `/ttm-verify`.
+
+---
+
+## Step 8: Update MANIFEST.json
+
+```
+takeToMarket > UPDATING MANIFEST
+```
+
+Read existing MANIFEST.json to determine the next available asset_id.
+Find the maximum asset_id across hero.asset_id and all derivatives[].asset_id.
+Set NEXT_ID = max_id + 1.
+
+Build `DERIVATIVES_JSON` array -- one entry per successfully produced derivative, each with
+`asset_id` (sequential from NEXT_ID), `name`, `type: "derivative"`, `channel`, and `file`.
+
+Run the campaign.cjs CLI to update MANIFEST.json:
+```bash
+node "${CLAUDE_PLUGIN_ROOT}/bin/ttm-tools.cjs" campaign repurpose-manifest "${CAMPAIGN_SLUG}" ${SOURCE_ASSET_ID} '${DERIVATIVES_JSON}'
+```
+
+Verify the command succeeded (exit code 0 and output contains "derivatives_added").
+
+---
+
+## Step 9: Completion Banner
+
+Display a banner with:
+- Source asset name and channel
+- Campaign slug
+- Derivative count (successful / attempted)
+- Per-derivative table: #, Channel, File, Positioning result, Format result, Voice result
+- Warnings for any failed productions or FAIL gate results
+- Manifest and brief file paths
+- Next steps: `/ttm-verify`, `/ttm-review`, `/ttm-fix`
+
+</process>
+
+<success_criteria>
+- [ ] Source asset validated (exists, in campaign, campaign in correct phase)
+- [ ] Channel selection defaults to all active minus source, with user override
+- [ ] Hero derivative produced first (blocking) in highest-reach channel
+- [ ] Remaining derivatives produced in wave-parallel via Task()
+- [ ] Each derivative gets a channel-adapted brief cross-referencing CHANNELS.md
+- [ ] Simplified verification (positioning drift, format, voice) runs per derivative
+- [ ] All derivatives tracked in MANIFEST.json with source_asset_id
+- [ ] Completion banner with per-derivative verification results
+</success_criteria>
+
+<output>
+- `.marketing/CAMPAIGNS/${SLUG}/REPURPOSE-BRIEF-*.md` (per-channel derivative briefs)
+- `.marketing/CAMPAIGNS/${SLUG}/ASSETS/R-*.md` (produced derivative assets)
+- `.marketing/CAMPAIGNS/${SLUG}/MANIFEST.json` (updated with derivative entries)
+</output>