@cliangdev/conductor 0.41.1 → 0.42.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (44) hide show
  1. package/assets/claude/agents/conductor-coder.md +278 -7
  2. package/assets/claude/agents/{conductor-content-art-director.md → conductor-marketing-art-director.md} +2 -2
  3. package/assets/claude/agents/{conductor-content-creative-director.md → conductor-marketing-creative-director.md} +12 -8
  4. package/assets/claude/agents/{conductor-content-ideator.md → conductor-marketing-ideator.md} +11 -5
  5. package/assets/claude/agents/{conductor-content-scriptwriter.md → conductor-marketing-scriptwriter.md} +19 -6
  6. package/assets/claude/agents/{conductor-content-strategist.md → conductor-marketing-strategist.md} +2 -2
  7. package/assets/claude/commands/conductor/implement.md +1 -1
  8. package/assets/claude/commands/conductor/marketing.md +26 -0
  9. package/assets/claude/skills/conductor-marketing-create/SKILL.md +87 -0
  10. package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/brand-kit-setup.md +2 -3
  11. package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/design-critique.md +0 -1
  12. package/assets/claude/skills/{conductor-creative/SKILL.md → conductor-marketing-create/references/direct.md} +50 -60
  13. package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/hooks.md +13 -9
  14. package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/idea-template.md +2 -2
  15. package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/layout-guide.md +0 -1
  16. package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/realism-check.md +0 -1
  17. package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/rubric.md +15 -6
  18. package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/script-template.md +4 -2
  19. package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/slideshow-formats.md +12 -9
  20. package/assets/claude/skills/{conductor-content-studio/SKILL.md → conductor-marketing-create/references/team-run.md} +33 -37
  21. package/assets/claude/skills/conductor-marketing-create/references/voice.md +120 -0
  22. package/assets/claude/skills/{conductor-publisher → conductor-marketing-publish}/SKILL.md +5 -5
  23. package/assets/claude/skills/{conductor-marketing-analyst → conductor-marketing-report}/SKILL.md +11 -4
  24. package/assets/cli-manifest.json +17 -37
  25. package/dist/lib/plugin-assets.d.ts.map +1 -1
  26. package/dist/lib/plugin-assets.js +98 -42
  27. package/dist/lib/plugin-assets.js.map +1 -1
  28. package/package.json +1 -1
  29. package/assets/claude/commands/conductor/content.md +0 -66
  30. package/assets/claude/commands/conductor/creative.md +0 -108
  31. package/assets/claude/commands/conductor/marketing-report.md +0 -38
  32. package/assets/claude/skills/conductor-coder/SKILL.md +0 -282
  33. package/assets/claude/skills/conductor-content-studio/references/layout-guide.md +0 -73
  34. package/assets/claude/skills/conductor-content-studio/references/realism-check.md +0 -95
  35. package/assets/claude/skills/conductor-content-studio/references/slideshow-formats.md +0 -135
  36. package/assets/claude/skills/conductor-creative/references/design-critique.md +0 -52
  37. package/assets/claude/skills/conductor-ux-ui-design/SKILL.md +0 -346
  38. package/assets/claude/skills/conductor-ux-ui-design/references/design-tokens.md +0 -359
  39. /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/capabilities.md +0 -0
  40. /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/compliance-checklist.md +0 -0
  41. /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/platform-specs.md +0 -0
  42. /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/publish-pack-template.md +0 -0
  43. /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/shotlist-template.md +0 -0
  44. /package/assets/claude/skills/{conductor-marketing-analyst → conductor-marketing-report}/references/html-layout.md +0 -0
@@ -1,29 +1,34 @@
1
- ---
2
- name: conductor-creative
3
- description: Creates on-brand marketing creative through the Conductor MCP tools — static ads and short videos alike, rendered locally on this machine as a draft, checked by eye and approved before anything is saved to Conductor. Covers a photo-plus-headline ad at every enabled placement (e.g. TikTok/Reels 9:16, Instagram feed 4:5, Facebook 1:1); a short video made from a still photo or an existing clip (animated text, zoom/pan, background music, a TikTok/Reels/Shorts-ready cut); posting a finished video as-is; an organic post as a native slideshow (compose_slideshow, then a CAROUSEL); posting finished images as-is (a carousel or single still designed elsewhere, or an existing media set); a story or carousel sequence; a lettered headline/hook variant for testing; and, when a workspace's Brand Kit has never been set up, setting one up first (colours, font, logo, copy rules) from a website, a brand doc, or a few direct answers. Use when asked to make, design, write or export a social ad, a promo image, a short-form video, an animated ad, or to configure/update a project's brand.
4
- user-invocable: true
5
- allowed-tools: mcp__conductor__*, AskUserQuestion, Read, Glob, Grep, WebFetch
6
- ---
7
-
8
- # Conductor Creative
9
-
10
- Turns one concept — a photo and a headline — into upload-ready artwork at every placement a
11
- workspace's Brand Kit enables. This is the front half of Conductor's marketing pillar: what this
12
- skill produces becomes a Post's media, and `conductor-publisher` takes it from there through review
13
- and out to each destination. You write the ad here; you do not post it.
14
-
15
- You are writing the ad, not just running a tool. The copy and the visual language both come from
16
- things that already exist for this workspace — a Brand Kit (data: tokens, copy rules, approved
17
- lines) and a Knowledge page (prose: positioning, voice, facts). Read them rather than inventing.
18
-
19
- Rendering happens locally, through this MCP server, with a browser on this machine — there is no
20
- separate render service to wait on. A preview typically finishes in well under a minute.
21
-
22
- **Draft first, save after approval.** Every Creative here goes draft, look, approve, commit:
23
- `preview_creative_draft` renders the artwork on this machine and writes **nothing** to Conductor — no
24
- Creative, no upload — then the person looks at it, and only on their approval does
25
- `commit_creative_draft` upload the media, save the Creative and do the full render. Never commit a
26
- draft the person has not seen and approved.
1
+ # Direct mode: making a Creative from what is in hand
2
+
3
+ Read when the person already has the headline, photo, clip or finished images (or wants a lettered
4
+ headline variant). The shared rules in `SKILL.md` (brand kit first, draft first, no Post) apply.
5
+
6
+ The copy and the visual language both come from things that already exist for this workspace: a
7
+ Brand Kit (data: tokens, copy rules, approved lines) and a Knowledge page (prose: positioning, voice,
8
+ facts). Read them rather than inventing. Rendering happens locally through this MCP server with a
9
+ browser on this machine; a preview typically finishes in well under a minute.
10
+
11
+ ## What to make
12
+
13
+ If the request does not already say, ask once with `AskUserQuestion` (single select), then gather
14
+ only what the choice needs: one or two follow-up questions, not a form.
15
+
16
+ - **Static ad** (photo + headline in the brand layout, at every enabled placement; not for organic
17
+ posts). Needs a headline (or draft one from the Knowledge page) and a photo: call
18
+ `list_creative_media({kind: "IMAGE"})` first and offer what is on hand before asking for a new one.
19
+ Nothing fits and none is offered? Offer `generate_creative_image` (step 5). A new photo is a local
20
+ file path (a URL: download it first); it is only uploaded after the person approves the preview.
21
+ - **Short video from a photo** (animated text over a still or a moving background): same inputs as a
22
+ static ad; see "Short videos from a photo or clip".
23
+ - **Video from a clip I have** (brand layout animated over an existing video) and **post an existing
24
+ video as-is** (no layout): call `list_creative_media({kind: "VIDEO"})` first; if nothing fits, ask
25
+ for a local path (a URL: download it first). A clip used as a MOTION background is previewed from
26
+ the local file too.
27
+ - **Organic post or slideshow** (a TikTok or Instagram photo post that is not an ad): skip the brand
28
+ layouts and follow the slideshow path below. Ask for the topic, the photos and a real app
29
+ screenshot if there is one.
30
+ - **Headline variant** of an existing Creative: ask which one (id or display id like `12a`) and the
31
+ new headline; flow in step 6.
27
32
 
28
33
  ## Which path
29
34
 
@@ -58,8 +63,8 @@ default, not this workspace's actual brand. Set it up before writing any ad rath
58
63
  against placeholder tokens: gather colours, font, CTA claim and any copy rules or approved lines
59
64
  (from a website via `WebFetch`, a brand doc via `Read`, or by asking directly), then call
60
65
  `update_brand_kit` for the data fields and `upload_brand_image` for the mark/wordmark/badge. Confirm
61
- with `get_brand_kit` afterward — `configured` should now read `true`. (`/conductor:creative` walks
62
- a person through exactly this when invoked with no brand set up yet, if you'd rather hand it off.)
66
+ with `get_brand_kit` afterward — `configured` should now read `true`. (`references/brand-kit-setup.md`
67
+ walks a person through exactly this.)
63
68
 
64
69
  ### 2. Pick the angle
65
70
 
@@ -74,7 +79,8 @@ of the problem or the result) before you write a line of copy.
74
79
 
75
80
  Prefer an approved line from the Knowledge page verbatim. Otherwise write one that names something
76
81
  real and specific, not a generic claim. Write 6 to 10 candidates across hook types, score them and
77
- keep the best two or three: `references/hooks.md`.
82
+ keep the best two or three: `references/hooks.md`. Every candidate must read like a person posted
83
+ it: `references/voice.md`.
78
84
 
79
85
  If the Brand Kit's `accentPhraseRequired` is true, mark exactly one phrase per headline with
80
86
  asterisks: `Every saved link, *finally usable*.` Never zero, never two — `preview_creative_draft`,
@@ -143,7 +149,7 @@ concealment: the AI label still applies.
143
149
  **Open the photo and look at it before using it** — a generated one most of all (pseudo-text, warped
144
150
  hands or faces, a fake product or app screen; the realism check covers the rest). A picture can break every copy rule with text,
145
151
  a logo or a claim burned into its pixels, and no automated check can see that — this is the first
146
- of this skill's silent failure modes (the full list is below). A human or a prior pass marks a bad
152
+ of this skill's silent failure modes (the full list is in `SKILL.md`). A human or a prior pass marks a bad
147
153
  photo `blocked`; if you notice one that should be, say so rather than using it anyway.
148
154
 
149
155
  Match the crop to the subject. A layout that lays copy over the photo (`bleed`) can put a face in
@@ -252,8 +258,19 @@ Creative is already saved — call `render_creative`, don't commit again. Verify
252
258
  Then `attach_creative_to_post` with the Post's Work Item id — it maps each placement to the right
253
259
  destinations (9:16 to TikTok/Reels, 4:5 to Instagram feed, 1:1 to Facebook, story to story targets)
254
260
  for every destination that has not set its own custom media, and never touches one that has. Verify
255
- with `get_post_status`. From here, use the `conductor-publisher` skill for scheduling, review and
256
- publishing — this skill's job ends at attaching the artwork.
261
+ with `get_post_status`. From here, use the `conductor-marketing-publish` skill for scheduling, review and
262
+ publishing — this skill's job ends at attaching the artwork. If there is no Post yet, say so plainly
263
+ and name the next step (creating one with `conductor-marketing-publish`) rather than leaving the artwork with
264
+ no stated destination.
265
+
266
+ When the Creative came from a team run, or the person points to a publish pack, look for
267
+ `.conductor/content-runs/*/publish-pack.json` whose `creativeId` matches. It carries per-destination
268
+ captions and publish options (disclosure, AI label, cover frame, alt text) and a pinned comment. When
269
+ you go on to create or edit the Post with `conductor-marketing-publish`, pre-fill each target's
270
+ `captionOverride` and options from it, show them in the confirmation table and let the person change
271
+ them. Mention the pinned comment (and, for a post that promotes an app, the Stories repost with a link
272
+ sticker) as manual steps, since Conductor cannot post comments. The Post still goes through the normal
273
+ review.
257
274
 
258
275
  ## Making a sequence instead of a single ad
259
276
 
@@ -347,7 +364,7 @@ thing).
347
364
  accepts a chosen track on photos, so each destination sends it as `carouselAs: "photos"`
348
365
  (silent) or `"video"` (the slideshow MP4 with the track). Attach defaults to video for TikTok
349
366
  and YouTube and photos for Instagram and Facebook feed, and writes the choice into the target's
350
- options; change it with `set_publish_targets` (publisher skill). Mind each platform's length limit
367
+ options; change it with `set_publish_targets` (`conductor-marketing-publish` skill). Mind each platform's length limit
351
368
  for the video.
352
369
 
353
370
  ## Short videos from a photo or clip
@@ -412,30 +429,3 @@ are the same for all three:
412
429
  ## Cleaning up
413
430
 
414
431
  `delete_creative` and `delete_media` are permanent. Never call them unless the person asked or confirmed that exact item; the server refuses a Creative still on a Post or with other variants, and Media still in use.
415
-
416
- ## Failures that are silent (carried over from this skill's predecessor)
417
-
418
- Automated checks are real but incomplete. These are the ones worth remembering to look for by eye:
419
-
420
- - **A photo with text, a logo or a claim burned into its pixels** passes every copy-rule and
421
- render check, because none of them look at pixels for meaning — only step 5's "open and look at
422
- it" catches this.
423
- - **A generated image depicting a product, result or claim that isn't real** (a made-up app screen,
424
- a before/after, an invented package) passes every check — only looking at it catches it.
425
- - **A structurally valid render is not necessarily a good one.** The render's own assertions check
426
- spill, safe zones, fonts and contrast — never whether the headline reads at a glance or the copy
427
- sits on the subject's face. Step 7 exists because of this gap.
428
- - **A Brand Kit with no copy rules, or no accent-phrase requirement, never fails a save on that
429
- account.** That is permissive by design, not a bug — but it means the model, not the platform, is
430
- the only check on brand voice for that workspace until someone configures rules.
431
- - **A draft is not a saved Creative.** `preview_creative_draft` writes nothing to Conductor, so a
432
- Creative or media id does not exist until `commit_creative_draft` runs — don't hand a draft's
433
- folder to `attach_creative_to_post` or a Post.
434
- - **A stale write is refused, not silently overwritten.** `update_creative` requires the Creative's
435
- current `version`; a mismatch comes back as a 409 naming the conflict rather than clobbering a
436
- concurrent edit — reread with `get_creative` and reapply, don't retry blind.
437
-
438
- ## Related skills
439
-
440
- - `conductor-publisher`: schedules and tracks the Post this Creative's artwork attaches to, through
441
- review and out to each destination. This skill hands off to it, never the reverse.
@@ -1,4 +1,4 @@
1
- <!-- Used by conductor-creative only. -->
1
+ <!-- Used by the Direct mode of conductor-marketing-create. -->
2
2
  # Hooks: writing the headline
3
3
 
4
4
  A hook is the line that makes a thumb stop. A tagline says what the brand is; a hook names a moment,
@@ -8,12 +8,14 @@ an object or a number the viewer recognises. Write several, score them, keep the
8
8
 
9
9
  Mix the types. Do not write ten versions of one.
10
10
 
11
- - **Situation / POV**: the moment the viewer is in. "You have 40 tabs open and no idea which one has the quote."
12
- - **Mechanism**: how it works, in one line. "Paste a link, get a summary and a reminder."
11
+ - **Situation / POV**: the moment the viewer is in. "which tab had the quote?"
12
+ - **Mechanism**: how it works, in one line. "paste a link. skip the rereading."
13
13
  - **Enemy**: the thing it replaces or kills. "The spreadsheet nobody updates."
14
+ Mechanism and Enemy hooks need an attitude to pass `voice.md`'s self-check: "the spreadsheet nobody
15
+ updates" has one; a bare feature label ("Paste a link, get a summary") does not.
14
16
  - **Exact result**: a specific number or time. "Invoices sent in 4 minutes, not 4 days."
15
17
  - **Concept**: a visual metaphor the image carries, with the line finishing the thought. A photo of one
16
- chair at a long empty table, "Standups, minus the standing."
18
+ chair at a long empty table, "this chair has been waiting since 9:15".
17
19
 
18
20
  An approved line from the Knowledge page still wins when one fits. Quote it verbatim; use the
19
21
  candidates to find the angle around it or when none fits.
@@ -26,6 +28,8 @@ promise that could open any account's post, 5 a specific, surprising line a stra
26
28
  - **Hook strength**: would it stop a thumb?
27
29
  - **Specificity**: does it name a moment, an object or a number?
28
30
  - **Competitor test**: could a competitor run this unchanged? If yes, it fails; rewrite or drop it.
31
+ - **Reads like a person**: would someone say it out loud to a friend? Run `voice.md`'s self-check;
32
+ a template frame ("X but Y"), a brand word or a line that describes the photo fails.
29
33
 
30
34
  Exact numbers read as more credible than round ones: "11 minutes" over "10 minutes". One selling point
31
35
  per line, written the way a person would say it.
@@ -39,9 +43,9 @@ hooks for variants (`variantOf`) or the caption. Run the chosen line through the
39
43
 
40
44
  | Tagline | Hook |
41
45
  |---|---|
42
- | Smarter scheduling for busy teams | The meeting that could have been a link, gone |
43
- | Fresh food, delivered | Dinner at your door before the oven is warm |
44
- | Save money on energy | Your bill dropped $38 last month. Ours too. |
45
- | Learn faster with AI | Fluent in 30 days, or the next month is free |
46
+ | Smarter scheduling for busy teams | this meeting could've been a link |
47
+ | Fresh food, delivered | the oven's still cold. dinner's here. |
48
+ | Save money on energy | our own bill dropped $38 last month |
49
+ | Learn faster with AI | ordered lunch in Spanish on day 30 |
46
50
 
47
- Check the last row against the brand's claims rules before using a line like it: a number has to be true.
51
+ Check the number and speed rows against the brand's claims rules before using a line like them: a number or a result has to be true.
@@ -11,8 +11,8 @@ Heading: `## Idea <n>: <title>` where the title is 8 words or fewer.
11
11
 
12
12
  - **Concept:** <one line, max 25 words: what the viewer sees and why it works>
13
13
  - **Hook, spoken:** <max 12 words, or "none (text only)">
14
- - **Hook, on screen:** <max 12 words>
15
- - **Hook type:** <question | bold-claim | pattern-interrupt | relatable-pain | how-to | number | curiosity-gap | story | social-proof>
14
+ - **Hook, on screen:** <max 12 words; on the artwork aim for about 8 or fewer, `voice.md`>
15
+ - **Hook type:** <question | bold-claim | pattern-interrupt | relatable-pain | how-to | number | curiosity-gap | story | social-proof; tag it from `voice.md`'s "Shapes, and the hookType each one is tagged" table>
16
16
  - **Format:** <slideshow | still | carousel | story sequence | motion text over photo | motion text over clip | talking head | demo | skit | other>
17
17
  - **Platform:** <primary platform, then any others>
18
18
  - **Tag:** RENDER | FILM
@@ -1,4 +1,3 @@
1
- <!-- Keep identical to the copy in the other creative skill (conductor-creative / conductor-content-studio references/layout-guide.md). Skills install independently and cannot reference each other. -->
2
1
  # Layout guide
3
2
 
4
3
  Pick the layout on purpose. A Creative has one layout across all its placements (9:16, 4:5, 1:1), so
@@ -1,4 +1,3 @@
1
- <!-- Keep identical to the copy in the other creative skill (conductor-creative / conductor-content-studio references/realism-check.md). Skills install independently and cannot reference each other. -->
2
1
  # Make it real, then check it
3
2
 
4
3
  This is about image quality, not concealment. The `aiGenerated: true` provenance and the platform's
@@ -1,6 +1,6 @@
1
1
  # Judging rubric
2
2
 
3
- Used by the `conductor-content-creative-director` agent in `text-round` and `visual-round`, and by
3
+ Used by the `conductor-marketing-creative-director` agent in `text-round` and `visual-round`, and by
4
4
  the orchestrator when it needs to break a tie. Everything here exists to counter a known judge
5
5
  failure mode, so follow it as written.
6
6
 
@@ -23,11 +23,12 @@ Score each 1 to 5. Weighted total = sum(score x weight) / 5, so the maximum is 1
23
23
 
24
24
  | Criterion | Weight | 1 | 3 | 5 |
25
25
  |---|---|---|---|---|
26
- | Hook strength | 25 | Opens on a greeting, a logo or context; nothing makes a stranger stay | A clear promise or question, but generic; could open any account's post | A specific, surprising or tension-filled claim a stranger understands in 1s, spoken and on screen |
26
+ | Hook strength | 20 | Opens on a greeting, a logo or context; nothing makes a stranger stay | A clear promise or question, but generic; could open any account's post | A specific, surprising or tension-filled claim a stranger understands in 1s, spoken and on screen |
27
27
  | Audience relevance | 20 | Aimed at "everyone"; ignores the brief's tension | Names the audience's topic but not their actual friction | Opens on a friction the brief names and the audience would recognise as theirs |
28
- | Brand fit | 15 | Breaks a copy rule or approved-line, or reads off-voice | Within the rules but could be any brand | Uses an approved line or the knowledge page's voice naturally; no rule risk |
28
+ | Reads like a person | 15 | A template frame ("X but Y"), brand words, a described picture or an explained joke; nobody would say it out loud | Plain and correct, but no speaker or attitude; reads as a caption, not a post | Someone is talking, with one real detail and a reason to care; passes every question in `voice.md`'s self-check |
29
+ | Brand fit | 10 | Breaks a copy rule or approved-line, or reads off-voice | Within the rules but could be any brand | Uses an approved line or the knowledge page's voice naturally; no rule risk |
29
30
  | Feasibility with assets on hand | 15 | Needs assets that do not exist and cannot be made here (and is tagged `RENDER`) | Needs one new photo or clip the user can supply easily | Fully buildable from `media.json` or a render with no photo; or an honest `FILM` with a simple shoot |
30
- | Novelty | 15 | The mechanic is the most common one for this topic (listicle, "5 tips", before/after) | A familiar mechanic with a fresh angle | A mechanic not seen elsewhere in the pool or in the performance data, still easy to grasp |
31
+ | Novelty | 10 | The mechanic is the most common one for this topic (listicle, "5 tips", before/after) | A familiar mechanic with a fresh angle | A mechanic not seen elsewhere in the pool or in the performance data, still easy to grasp. Novelty is in the idea, not in an odd sentence: a strange word pairing scores nothing here |
31
32
  | Shareability | 10 | Nothing to send to a friend; purely informational | Useful or pleasant, but no reason to send | A person would send it to a named other person ("this is you") or save it for later |
32
33
 
33
34
  ## Scoring protocol
@@ -37,13 +38,19 @@ Score each 1 to 5. Weighted total = sum(score x weight) / 5, so the maximum is 1
37
38
  2. For each idea, per criterion, write one or two sentences of reasoning FIRST, citing the idea's
38
39
  own words. Then write the score. Never write a score before its reasoning.
39
40
  3. Penalise nothing for length (ideas are length-matched) and reward nothing for confident tone.
40
- 4. Compute the weighted total. Shortlist the top 3. If ties, apply the tie-break rule below.
41
+ 4. Compute the weighted total.
41
42
  5. Any idea that breaks a copy rule or contradicts an approved line is capped at 2 on brand fit
42
43
  and cannot be a finalist unless the fix is a one-word edit you state.
44
+ 6. Any idea scoring 1 or 2 on reads-like-a-person cannot be a finalist, full stop. Do not rewrite
45
+ it and score the rewrite.
46
+ 7. Shortlist the top 3 eligible ideas (steps 5 and 6 applied first). If ties, apply the tie-break
47
+ rule below. If fewer than 3 are eligible, say so in `scores.md` and continue with what remains:
48
+ the pairwise round covers the remaining pairs, and a single eligible idea wins outright (the
49
+ visual round only checks it against the veto questions).
43
50
 
44
51
  ## Pairwise protocol (visual round)
45
52
 
46
- For each pair of finalists (A,B), (A,C), (B,C):
53
+ For each pair of finalists (for three: (A,B), (A,C), (B,C); for two, the one pair; for one, skip this):
47
54
 
48
55
  1. Judge the pair twice, once with each order, as two independent comparisons. Write reasoning
49
56
  first, then the verdict "first" or "second", then say which idea won.
@@ -65,6 +72,8 @@ Look at the actual image file, not the text description. Check:
65
72
  payoff not the opener, native look, not a banner ad), each 1 to 5. A finalist with a criterion below 3
66
73
  that the layout could have fixed (`stacked` or `split` on 9:16, a dead panel) counts against it.
67
74
  - Realism: a generated photo that looks AI-made (waxy, duplicated objects, warped hands) counts against the finalist.
75
+ - Voice: read every card and the caption against `voice.md`'s self-check. A card that describes
76
+ the picture, a template frame or a brand word counts against the finalist.
68
77
  - Whether the picture and the words do different jobs (photo shows, words claim) rather than
69
78
  repeating each other.
70
79
 
@@ -12,7 +12,7 @@ Idea: <id> Source decision: decision.md
12
12
 
13
13
  ## Hook (0-3s)
14
14
  - Spoken: <line, max 12 words, or "none">
15
- - On screen: <text, max 12 words>
15
+ - On screen: <text, max 12 words; on the artwork about 8 or fewer>
16
16
  - Visual: <what is shown>
17
17
 
18
18
  ## Beats
@@ -42,10 +42,12 @@ opening words (and in the on-screen text and speech); the first line carries the
42
42
  3. <optional>
43
43
 
44
44
  Hook types: question, bold-claim, pattern-interrupt, relatable-pain, how-to, number,
45
- curiosity-gap, story, social-proof.
45
+ curiosity-gap, story, social-proof (tag from `voice.md`'s shapes table).
46
46
 
47
47
  ## Copy-rule check
48
48
  <each must-say / must-avoid item from brief.md, and ok or the fix>
49
+ ## Voice check
50
+ <each line changed against voice.md, the hook included: before, after, which question or tell it failed; or "all lines pass". Functional lines (CTA, save or follow ask, link in bio, pinned comment, plain app mention, disclosure, approved lines) are checked only against "Tells to cut". Wording the person asked for stays; note any tell it trips here.>
49
51
  ```
50
52
 
51
53
  Rules: a beat every 3 to 5 seconds (a `MOTION` idea instead has one headline, one optional body and an end card, with no timed on-screen beats); no beat longer than 5 seconds without a visual or text change;
@@ -1,4 +1,3 @@
1
- <!-- Keep identical to the copy in the other creative skill (conductor-creative / conductor-content-studio references/slideshow-formats.md). Skills install independently and cannot reference each other. -->
2
1
  # Slideshow formats
3
2
 
4
3
  For organic posts: a native-looking photo slideshow (TikTok photo post, Instagram carousel) that
@@ -70,22 +69,26 @@ A five-slide "stop doing X" post. Paths are relative to the spec file or absolut
70
69
  ## Hook archetypes
71
70
 
72
71
  Pick one per post and name it in the idea. Change one thing per post so results are comparable.
72
+ The slide-1 examples are shapes only: each must pass `voice.md`'s self-check with your own true
73
+ detail, and none may state a fact the brief does not support.
73
74
 
74
75
  | Archetype | Slide 1 | Slides |
75
76
  |---|---|---|
76
- | Saved-but-never-used guilt | "Camera-roll graveyard." | `grid` (one tile `restore`d at the end), `photo`, `screen`, ask |
77
- | Doubt or conflict | What someone said to you (needs a true story) | `messages`, `photo`, `notes`, `screen`, ask |
78
- | Stop doing X | "Stop doing X." | `photo` with the reason per slide, `screen`, ask |
79
- | Numbered list | "5 things worth saving" | `photo` per item, `screen` last, ask |
80
- | Tier list | "Ranking everything I tried" | `tier`, `photo` close-ups, `screen`, ask |
81
- | Notes-app plan | "My plan for the week" | `notes`, `week`, `screen`, ask |
82
- | POV | "POV: you finally did it" | `photo` (hands or a real scene), `screen`, `photo`, ask |
83
- | Before / after | "Week 1 vs week 4" | `stack` with `label`s, `photo`, `screen`, ask |
77
+ | Saved-but-never-used guilt | "the recipes folder is a graveyard" | `grid` (one tile `restore`d at the end), `photo`, `screen`, ask |
78
+ | Doubt or conflict | A real message someone sent you (needs a true story) | `messages`, `photo`, `notes`, `screen`, ask |
79
+ | Stop doing X | "stop checking your balance at midnight" | `photo` with the reason per slide, `screen`, ask |
80
+ | Numbered list | "5 stretches for desk-bound shoulders" | `photo` per item, `screen` last, ask |
81
+ | Tier list | "ranking airport snacks. no appeals" | `tier`, `photo` close-ups, `screen`, ask |
82
+ | Notes-app plan | "my monday plan, aka the optimistic draft" | `notes`, `week`, `screen`, ask |
83
+ | POV | "pov: it's sunday and the week is meal-prepped" | `photo` (hands or a real scene), `screen`, `photo`, ask |
84
+ | Before / after | "week 1 vs week 4, same grocery run" (needs real before and after) | `stack` with `label`s, `photo`, `screen`, ask |
84
85
 
85
86
  ## Text rules
86
87
 
87
88
  - Native box text (`box`) or white text with a black stroke (`stroke`); centred, bold, sans.
88
89
  - At most 8 words a line and 3 lines a box.
90
+ - Every card reads like a person posted it (`voice.md`): a speaker, one real detail, no template
91
+ frames, never a description of the photo.
89
92
  - Keep text between about y 16 and y 75 (percent of the 9:16 frame). That survives the 4:5 Instagram
90
93
  cut and the platform UI (top bar, bottom caption, right button column).
91
94
  - The first text of every slide needs `y`.
@@ -1,15 +1,9 @@
1
- ---
2
- name: conductor-content-studio
3
- description: Runs a small team of subagents (strategist, three ideators, creative director, scriptwriter, art director) that brainstorm competing short-form social post ideas, pick one with a bias-resistant judging method, write the script and visual direction, run pre-publish checks (disclosure, AI label, music, claims, third-party material, accessibility), and render a local draft with a publish pack the publisher later pre-fills from. The person approves at the preview; only then is the Creative uploaded to Conductor. Ideas are either RENDER (a native slideshow for organic posts, or a still, carousel, story sequence or motion text that Conductor can render) or FILM (needs real filming, so the output is a shoot-ready script). A run ends at a committed Creative or a shoot script and never creates a Post. Use when asked to come up with, choose and make a short-form post (TikTok, Reels, Shorts, Facebook) from a goal rather than from a given headline, or via /conductor:content. For a headline or photo already chosen, use conductor-creative instead.
4
- user-invocable: true
5
- allowed-tools: mcp__conductor__*, Agent, AskUserQuestion, Skill, Read, Write, Glob, Grep, WebFetch
6
- ---
7
-
8
- # Conductor Content Studio
9
-
10
- Takes a request ("a post about X for Y") to either a committed Creative or a shoot-ready script. The
11
- team chooses the idea itself; the person is asked once, at the final draft preview. You are the
12
- orchestrator and run in the main session. Subagents do the thinking and writing; you do every
1
+ # Team mode: from a goal to a committed Creative
2
+
3
+ Read when the person gives a goal ("a post about X for Y") rather than a headline, photo or clip. The
4
+ shared rules in `SKILL.md` apply. Takes the request to either a committed Creative or a shoot-ready
5
+ script. The team chooses the idea itself; the person is asked once, at the final draft preview. You
6
+ are the orchestrator and run in the main session. Subagents do the thinking and writing; you do every
13
7
  Conductor call and every file merge.
14
8
 
15
9
  ## Rules that shape the design
@@ -46,7 +40,8 @@ Each exists because of a known failure mode (sources are in `references/rubric.m
46
40
  rules into `context/banned.md` on every later run.
47
41
 
48
42
  Cost: a run is about 8 subagent calls (strategist 1, ideators 3, director 2, scriptwriter 1, art
49
- director 1), and only the strategist and the director use the top model tier. `--quick` runs two
43
+ director 1). The strategist, the director and the scriptwriter use the top model tier; the
44
+ ideators stay on the lower tier so the judge never scores its own tier's ideas. `--quick` runs two
50
45
  ideators and skips the finalist renders (about 7 calls).
51
46
 
52
47
  ## Run folder and state
@@ -107,7 +102,7 @@ template exactly; <any stage-specific rule>
107
102
  Return: at most 150 words plus the output path.
108
103
  ```
109
104
 
110
- Resolve the reference files once: Glob for `**/conductor-content-studio/references/rubric.md` under
105
+ Resolve the reference files once: Glob for `**/conductor-marketing-create/references/rubric.md` under
111
106
  `.claude/skills/` in this project, then under `~/.claude/skills/`. Use that folder for every
112
107
  `references/` path below. A subagent that returns more than 150 words or leaves its file unwritten
113
108
  has failed the stage: retry once with a one-line reminder, then tell the person.
@@ -161,20 +156,20 @@ Write `request.md` (the request verbatim, `--quick` if given, the platform if st
161
156
 
162
157
  ### S1 — Strategist -> `brief.md`
163
158
 
164
- `Agent(subagent_type: "conductor-content-strategist")` with the seven context files (including `context/banned.md`, whose every entry goes into the brief's must-avoid), `request.md`
159
+ `Agent(subagent_type: "conductor-marketing-strategist")` with the seven context files (including `context/banned.md`, whose every entry goes into the brief's must-avoid), `request.md`
165
160
  and `references/capabilities.md` (so the brief states rendering constraints correctly: motion text
166
161
  over a photo is renderable and needs no filming).
167
162
 
168
163
  ### S2 — Ideators, in parallel -> `ideas/angle-<n>.md`
169
164
 
170
165
  Three `Agent` calls in ONE message so they run in parallel, `subagent_type:
171
- "conductor-content-ideator"`, with these angles and no knowledge of each other:
166
+ "conductor-marketing-ideator"`, with these angles and no knowledge of each other:
172
167
 
173
168
  1. story/emotion 2. utility/education 3. trend/humor (the only one allowed web search)
174
169
 
175
- Each gets `brief.md`, `context/media.json`, `references/idea-template.md` and
176
- `references/capabilities.md` (the `RENDER`/`FILM` test), and writes 4 ideas.
177
- With `--quick`: only angles 1 and 2, 3 ideas each.
170
+ Each gets `brief.md`, `context/media.json`, `references/idea-template.md`,
171
+ `references/capabilities.md` (the `RENDER`/`FILM` test) and `references/voice.md`, and writes 4 ideas.
172
+ With `--quick`: only angles 1 and 2, 3 ideas each. Any ideator rerun passes the same inputs, `references/voice.md` included.
178
173
 
179
174
  ### Merge (you) -> `ideas/pool.md`
180
175
 
@@ -189,12 +184,14 @@ Build one pool from all the angle files:
189
184
 
190
185
  ### S3a — Creative director, text round -> `judging/scores.md`
191
186
 
192
- `Agent(subagent_type: "conductor-content-creative-director")`, mode `text-round`. Inputs:
193
- `brief.md`, `ideas/pool.md`, `references/rubric.md`. It scores every idea independently, reasoning
194
- before each score, and names the top 3. Store the finalists in `run.json`.
187
+ `Agent(subagent_type: "conductor-marketing-creative-director")`, mode `text-round`. Inputs:
188
+ `brief.md`, `ideas/pool.md`, `references/rubric.md`, `references/voice.md`. It scores every idea independently, reasoning
189
+ before each score, and names the top 3 eligible ideas (fewer when the rubric's gates leave fewer; it says so). Store the finalists in `run.json`.
195
190
 
196
191
  ### S3b — Finalist drafts (you; skipped in `--quick`)
197
192
 
193
+ Draft each finalist the judge named (up to three).
194
+
198
195
  A slideshow finalist has no cheap draft: judge it from its idea text (archetype and slide-by-slide
199
196
  text) unless its slides can be composed from photos already on hand. In that case write its spec to
200
197
  `<run>/drafts/finalist-<id>/slideshow-spec.json`, call `compose_slideshow` with `draftDir` set to
@@ -219,7 +216,7 @@ path as `photoPath` (generate only for finalists, not the whole pool).
219
216
 
220
217
  Same agent, mode `visual-round`. Inputs: `brief.md`, `judging/scores.md`, `ideas/pool.md`, the
221
218
  finalist image paths (`drafts/finalist-<id>/sheet.jpg`; the director views them with Read), the
222
- rubric, `references/design-critique.md` and `references/compliance-checklist.md`. It compares finalists pairwise in both orders, runs
219
+ rubric, `references/voice.md`, `references/design-critique.md` and `references/compliance-checklist.md`. It compares finalists pairwise in both orders (a single finalist wins outright and is only checked), runs
223
220
  the devil's-advocate pass against the leader (including the veto questions: an unsubstantiated
224
221
  claim, trademark or parody risk, a representation or sensitivity problem, disclosure needed), and
225
222
  writes `decision.md`: the winner, why it won, the runners-up, one line per cut idea and the veto
@@ -234,8 +231,8 @@ which idea to take; show the decision at the gate.
234
231
  ### S4 — Scriptwriter -> `script.md`
235
232
 
236
233
  First copy the winning idea's block from `ideas/pool.md` to `ideas/winner.md`. Then
237
- `Agent(subagent_type: "conductor-content-scriptwriter")` with `brief.md`, `decision.md`, `ideas/winner.md`, `references/script-template.md`, `references/platform-specs.md` and
238
- `references/publish-pack-template.md`. Besides the script it writes the "Numbers and sources" list (every number on a card or in a caption, with its source record) and the Publish copy section: a
234
+ `Agent(subagent_type: "conductor-marketing-scriptwriter")` with `brief.md`, `decision.md`, `ideas/winner.md`, `references/script-template.md`, `references/platform-specs.md`,
235
+ `references/publish-pack-template.md` and `references/voice.md` (required on every scriptwriter run, including the reruns below). Besides the script it writes the "Numbers and sources" list (every number on a card or in a caption, with its source record) and the Publish copy section: a
239
236
  caption per platform whose call to action names the app so it can be searched, hashtags, a pinned
240
237
  comment and 2 to 3 alternate hooks with a hookType each.
241
238
  For a `MOTION` winner tell it that the render shows one headline and one optional body, animated
@@ -254,7 +251,7 @@ through `generate_creative_image` and the realism check first, as below. Every p
254
251
  through the resolution gate below. The rest of this stage is
255
252
  the STILL/MOTION path.
256
253
 
257
- `Agent(subagent_type: "conductor-content-art-director")` with `script.md`, `context/media.json`,
254
+ `Agent(subagent_type: "conductor-marketing-art-director")` with `script.md`, `context/media.json`,
258
255
  `context/brand-kit.json`, `context/banned.md`, `references/shotlist-template.md`,
259
256
  `references/platform-specs.md`, `references/realism-check.md` and
260
257
  `references/publish-pack-template.md`. Besides the spec it writes the Publish assets section of
@@ -304,7 +301,7 @@ does not apply is answered "not applicable".
304
301
  sets `brandContentToggle`; gifted or affiliate sets `brandContentToggle` and adds the
305
302
  disclosure to the caption's first line.
306
303
  4. A claims or trademark check that fails and cannot be fixed by you in one word goes back to the
307
- scriptwriter once, with the failing line; re-run S4's merge and S5 only if the copy changed. If
304
+ scriptwriter once, with the failing line and `references/voice.md`; re-run S4's merge and S5 only if the copy changed. If
308
305
  it still fails, show it at the gate as a known issue and never hide it.
309
306
  5. Fact lint: merge the scriptwriter's "Numbers and sources" into the pack and check each number
310
307
  against its source record (checklist item 9). Fix a mismatch before the gate.
@@ -338,7 +335,7 @@ any placement is fixed now, not shown.
338
335
  `audio.source: "track"` previews as its slideshow MP4 too (`slideshowVideoPath`), so give the
339
336
  person that path to watch along with the sheet, and put the track's licence in the pack's music
340
337
  line (`references/compliance-checklist.md` section 3).
341
- 2. Eye-check the image. Load the `conductor-creative` skill with `Skill(skill: "conductor-creative")` and
338
+ 2. Eye-check the image. Read `references/direct.md` (same skill) and
342
339
  apply its review guidance (step 7 and the silent-failures list), scoring the draft with
343
340
  `references/design-critique.md` (the slideshow criteria for a slideshow, the STILL criteria
344
341
  otherwise; any criterion below 3 is fixed before anyone sees it). For a STILL: does the headline read at a
@@ -369,10 +366,10 @@ any placement is fixed now, not shown.
369
366
  its state into `publish-pack.json` as `creativeState` (and both into `run.json`).
370
367
  - **Use a runner-up**: ask a second `AskUserQuestion` listing the runners-up by title (with
371
368
  their tag; label one that could not be rendered as such and say why, rather than offering it as
372
- if it would work). Then set the chosen idea as `winner` and re-run S4 onward for it (S5, S5b and S6,
373
- or the FILM path if it is tagged `FILM`).
369
+ if it would work). Then set the chosen idea as `winner` and re-run S4 onward for it, passing
370
+ `references/voice.md` again (S5, S5b and S6, or the FILM path if it is tagged `FILM`).
374
371
  - **Request changes**: take the free text. If it is about the copy, send it with the current
375
- script to the scriptwriter (S4), then the art director (S5); if only about the look, send it to
372
+ script and `references/voice.md` to the scriptwriter (S4), then the art director (S5); if only about the look, send it to
376
373
  the art director. Redo S5b, then re-preview. Do not commit the old draft.
377
374
  - **Stop**: upload nothing. Tell the person where the drafts are
378
375
  (`drafts.final` and the finalist folders) and that the run can be resumed.
@@ -380,7 +377,7 @@ any placement is fixed now, not shown.
380
377
  ### FILM winner
381
378
 
382
379
  1. After S4, have the art director write the shoot version: `Agent(subagent_type:
383
- "conductor-content-art-director")` with `script.md`, `context/media.json`, `context/brand-kit.json`
380
+ "conductor-marketing-art-director")` with `script.md`, `context/media.json`, `context/brand-kit.json`
384
381
  and `references/shotlist-template.md`, output `shoot/shoot-script.md`: script, shot list, capture
385
382
  settings (9:16, 30 fps, sound on), B-roll, safe zones, and the on-screen text and captions ready to
386
383
  paste. It also writes the Publish assets section there. It writes no `creative-spec.json` here.
@@ -403,11 +400,10 @@ Keep it short:
403
400
  alternate hooks); say the publisher will pre-fill the Post from `publish-pack.json`, and that the
404
401
  pinned comment, the link in bio and the Stories repost with a link sticker are manual steps;
405
402
  - next steps:
406
- - make the Post with `/conductor:creative` or the `conductor-publisher` skill (the Creative is
403
+ - make the Post with the `conductor-marketing-publish` skill (the Creative is
407
404
  not on a Post yet, because a run never creates one);
408
405
  - to test a runner-up's hook, cut it as a lettered variant of the Creative with `variantOf`;
409
- - for `FILM`: film it, upload the clip, then run `/conductor:creative` and pick "Video from a
410
- clip I have".
406
+ - for `FILM`: film it, upload the clip, then run `/conductor:marketing create` and pick a clip I have.
411
407
 
412
408
  ## Failure handling
413
409
 
@@ -421,6 +417,6 @@ Keep it short:
421
417
 
422
418
  ## Related
423
419
 
424
- - `conductor-creative`: the single-Creative skill whose checklist S6 uses.
420
+ - `references/direct.md`: the single-Creative flow whose checklist S6 uses.
425
421
  - `references/slideshow-formats.md`: the slideshow spec, archetypes and text rules.
426
- - `conductor-publisher`: takes the committed Creative onto a Post, pre-filled from `publish-pack.json`.
422
+ - `conductor-marketing-publish`: takes the committed Creative onto a Post, pre-filled from `publish-pack.json`.