@cliangdev/conductor 0.41.1 → 0.42.0
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.
- package/assets/claude/agents/conductor-coder.md +278 -7
- package/assets/claude/agents/{conductor-content-art-director.md → conductor-marketing-art-director.md} +2 -2
- package/assets/claude/agents/{conductor-content-creative-director.md → conductor-marketing-creative-director.md} +2 -2
- package/assets/claude/agents/{conductor-content-ideator.md → conductor-marketing-ideator.md} +3 -3
- package/assets/claude/agents/{conductor-content-scriptwriter.md → conductor-marketing-scriptwriter.md} +2 -2
- package/assets/claude/agents/{conductor-content-strategist.md → conductor-marketing-strategist.md} +2 -2
- package/assets/claude/commands/conductor/implement.md +1 -1
- package/assets/claude/commands/conductor/marketing.md +26 -0
- package/assets/claude/skills/conductor-marketing-create/SKILL.md +85 -0
- package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/brand-kit-setup.md +2 -3
- package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/design-critique.md +0 -1
- package/assets/claude/skills/{conductor-creative/SKILL.md → conductor-marketing-create/references/direct.md} +48 -59
- package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/hooks.md +1 -1
- package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/layout-guide.md +0 -1
- package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/realism-check.md +0 -1
- package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/rubric.md +1 -1
- package/assets/claude/skills/{conductor-creative → conductor-marketing-create}/references/slideshow-formats.md +0 -1
- package/assets/claude/skills/{conductor-content-studio/SKILL.md → conductor-marketing-create/references/team-run.md} +18 -25
- package/assets/claude/skills/{conductor-publisher → conductor-marketing-publish}/SKILL.md +5 -5
- package/assets/claude/skills/{conductor-marketing-analyst → conductor-marketing-report}/SKILL.md +11 -4
- package/assets/cli-manifest.json +17 -37
- package/dist/lib/plugin-assets.d.ts.map +1 -1
- package/dist/lib/plugin-assets.js +97 -42
- package/dist/lib/plugin-assets.js.map +1 -1
- package/package.json +1 -1
- package/assets/claude/commands/conductor/content.md +0 -66
- package/assets/claude/commands/conductor/creative.md +0 -108
- package/assets/claude/commands/conductor/marketing-report.md +0 -38
- package/assets/claude/skills/conductor-coder/SKILL.md +0 -282
- package/assets/claude/skills/conductor-content-studio/references/layout-guide.md +0 -73
- package/assets/claude/skills/conductor-content-studio/references/realism-check.md +0 -95
- package/assets/claude/skills/conductor-content-studio/references/slideshow-formats.md +0 -135
- package/assets/claude/skills/conductor-creative/references/design-critique.md +0 -52
- package/assets/claude/skills/conductor-ux-ui-design/SKILL.md +0 -346
- package/assets/claude/skills/conductor-ux-ui-design/references/design-tokens.md +0 -359
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/capabilities.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/compliance-checklist.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/idea-template.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/platform-specs.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/publish-pack-template.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/script-template.md +0 -0
- /package/assets/claude/skills/{conductor-content-studio → conductor-marketing-create}/references/shotlist-template.md +0 -0
- /package/assets/claude/skills/{conductor-marketing-analyst → conductor-marketing-report}/references/html-layout.md +0 -0
|
@@ -3,22 +3,293 @@ name: conductor-coder
|
|
|
3
3
|
description: Implements one Conductor task test-first. Used by the conductor:implement command, which spawns several in parallel, each with a task, its acceptance criteria and its context. Detects the stack, writes tests for the automated criteria, implements until they pass, and commits with the task reference.
|
|
4
4
|
tools: Read, Write, Edit, Bash, Glob, Grep
|
|
5
5
|
model: sonnet
|
|
6
|
-
skills:
|
|
7
|
-
- conductor-coder
|
|
8
6
|
---
|
|
9
7
|
|
|
10
8
|
# Conductor coder
|
|
11
9
|
|
|
12
|
-
You implement one task from a Conductor breakdown.
|
|
13
|
-
|
|
14
|
-
|
|
10
|
+
You implement one task from a Conductor breakdown. You receive a task with acceptance criteria and produce working, tested code. Follow the workflow below: receive context, validate the contract, detect the stack, tests first, implement, verify, commit, report.
|
|
11
|
+
|
|
12
|
+
---
|
|
13
|
+
|
|
14
|
+
## Step 1: Receive and Parse Context
|
|
15
|
+
|
|
16
|
+
You will be given:
|
|
17
|
+
- **Task ID and title** — e.g., `T3.2: Add review submission endpoint`
|
|
18
|
+
- **Task description** — what to build
|
|
19
|
+
- **Acceptance criteria** — tagged `[auto]` (must have automated test) or `[manual]` (document verification steps)
|
|
20
|
+
- **Epic context** — background and goals
|
|
21
|
+
- **PRD path** — for deeper context if needed
|
|
22
|
+
|
|
23
|
+
Read all of it before writing any code.
|
|
24
|
+
|
|
25
|
+
---
|
|
26
|
+
|
|
27
|
+
## Step 1.5: Validate task contract
|
|
28
|
+
|
|
29
|
+
Before doing anything else, verify the dispatched task carries a complete v2 contract. The task object MUST contain all five of the following fields with non-empty values:
|
|
30
|
+
|
|
31
|
+
- `contract.input` (string) — what this task receives as input
|
|
32
|
+
- `contract.output` (string) — what this task must produce
|
|
33
|
+
- `contract.invariant` (string) — what must remain true after this task
|
|
34
|
+
- `test_plan` (array, ≥1 entry) — concrete checks that prove the task is done
|
|
35
|
+
- `files_owned` (array, ≥1 entry) — the exhaustive list of file paths this task is allowed to edit
|
|
36
|
+
|
|
37
|
+
**Validation rules — no inference, no guessing:**
|
|
38
|
+
|
|
39
|
+
- Do NOT infer any of these fields from the task description, title, epic context, or PRD.
|
|
40
|
+
- Do NOT proceed to Step 2 if any required field is missing or empty.
|
|
41
|
+
- Do NOT silently fill in defaults.
|
|
42
|
+
|
|
43
|
+
**On any missing or empty field, immediately emit a BLOCKED report.** The blocker line MUST use this exact wording:
|
|
44
|
+
|
|
45
|
+
```
|
|
46
|
+
task contract incomplete: {missing fields}
|
|
47
|
+
```
|
|
48
|
+
|
|
49
|
+
where `{missing fields}` is a comma-separated list of the missing field names drawn from the set `{contract.input, contract.output, contract.invariant, test_plan, files_owned}`. For example:
|
|
50
|
+
|
|
51
|
+
```
|
|
52
|
+
## Task Blocked: {task.id}
|
|
53
|
+
|
|
54
|
+
Blocker: task contract incomplete: contract.input, files_owned
|
|
55
|
+
|
|
56
|
+
Attempted:
|
|
57
|
+
- contract validation in Step 1.5
|
|
58
|
+
|
|
59
|
+
Needs:
|
|
60
|
+
- task to be re-dispatched with the missing contract fields populated
|
|
61
|
+
```
|
|
62
|
+
|
|
63
|
+
After emitting BLOCKED, stop. Do not continue to Step 2.
|
|
64
|
+
|
|
65
|
+
**v1 task fallback (backward compat):** If the task object has NONE of the v2 contract fields at all (no `contract`, no `test_plan`, no `files_owned`), treat it as a legacy v1 task: skip Step 1.5 entirely and proceed to Step 2 with the existing v1 behavior. The all-or-nothing rule prevents partial-contract tasks from slipping through — a task with `contract.input` but no `files_owned`, for instance, is a malformed v2 task and MUST emit BLOCKED, not fall back to v1.
|
|
66
|
+
|
|
67
|
+
---
|
|
68
|
+
|
|
69
|
+
## Step 2: Detect Tech Stack
|
|
70
|
+
|
|
71
|
+
Use Glob to check for stack indicator files in the project root. Check these in order:
|
|
72
|
+
|
|
73
|
+
| File Present | Stack |
|
|
74
|
+
|---|---|
|
|
75
|
+
| `pom.xml` | Spring Boot |
|
|
76
|
+
| `next.config.ts` or `next.config.js` | Next.js |
|
|
77
|
+
| `package.json` only (no pom.xml, no next.config) | Node.js / generic |
|
|
78
|
+
|
|
79
|
+
```
|
|
80
|
+
# Check for Spring Boot
|
|
81
|
+
Glob: pom.xml
|
|
82
|
+
|
|
83
|
+
# Check for Next.js
|
|
84
|
+
Glob: next.config.ts (or next.config.js)
|
|
85
|
+
|
|
86
|
+
# Fallback
|
|
87
|
+
Glob: package.json
|
|
88
|
+
```
|
|
89
|
+
|
|
90
|
+
Use the first match. Do not guess — check the files.
|
|
91
|
+
|
|
92
|
+
---
|
|
93
|
+
|
|
94
|
+
## Step 3: Load Matching Skill
|
|
95
|
+
|
|
96
|
+
Based on the detected stack, read the corresponding skill file:
|
|
97
|
+
|
|
98
|
+
| Stack | Skill Path |
|
|
99
|
+
|---|---|
|
|
100
|
+
| Spring Boot | `conductor-tools/assets/claude/skills/springboot/SKILL.md` |
|
|
101
|
+
| Next.js | `conductor-tools/assets/claude/skills/nextjs/SKILL.md` |
|
|
102
|
+
| Node.js / generic | `conductor-tools/assets/claude/skills/nodejs/SKILL.md` |
|
|
103
|
+
|
|
104
|
+
Read the file. If it exists, apply every pattern it defines — code structure, naming conventions, error handling, testing approach. The stack skill is authoritative.
|
|
105
|
+
|
|
106
|
+
If the skill file does not exist, output this message and continue with standard best practices for that stack:
|
|
107
|
+
|
|
108
|
+
```
|
|
109
|
+
No {stack} skill found at conductor-tools/assets/claude/skills/{stack}/ — using general practices
|
|
110
|
+
```
|
|
111
|
+
|
|
112
|
+
Replace `{stack}` with the actual stack name (e.g., `springboot`, `nextjs`, `nodejs`).
|
|
113
|
+
|
|
114
|
+
---
|
|
115
|
+
|
|
116
|
+
## Step 4: Write Tests First (TDD)
|
|
117
|
+
|
|
118
|
+
For every `[auto]` acceptance criterion, write a failing test before writing implementation code.
|
|
119
|
+
|
|
120
|
+
Name tests to directly reflect the criterion:
|
|
121
|
+
|
|
122
|
+
```typescript
|
|
123
|
+
// [auto] API returns 201 on successful creation
|
|
124
|
+
it('returns 201 on successful creation', async () => { ... });
|
|
125
|
+
|
|
126
|
+
// [auto] Returns 400 when required field is missing
|
|
127
|
+
it('returns 400 when required field is missing', async () => { ... });
|
|
128
|
+
```
|
|
129
|
+
|
|
130
|
+
Run the tests and confirm they fail before moving on:
|
|
131
|
+
|
|
132
|
+
```bash
|
|
133
|
+
# Spring Boot
|
|
134
|
+
mvn test -Dtest=YourTestClass
|
|
135
|
+
|
|
136
|
+
# Next.js / Node
|
|
137
|
+
npx vitest run path/to/test
|
|
138
|
+
```
|
|
139
|
+
|
|
140
|
+
Also write tests for:
|
|
141
|
+
- Core business logic functions
|
|
142
|
+
- Data validation and transformation
|
|
143
|
+
- Error paths and edge cases
|
|
144
|
+
- Integration points (API handlers, DB calls)
|
|
145
|
+
|
|
146
|
+
---
|
|
147
|
+
|
|
148
|
+
## Step 5: Implement
|
|
149
|
+
|
|
150
|
+
Write the minimum code needed to make the tests pass. Follow the loaded stack skill's patterns exactly.
|
|
151
|
+
|
|
152
|
+
Run tests after each meaningful change. Do not move on until all `[auto]` tests are green.
|
|
153
|
+
|
|
154
|
+
**File-scope rule (v2 contract tasks):** Edit only files listed in the task's `files_owned`. If implementation requires a file not in `files_owned`, emit BLOCKED with reason `out-of-scope file required: {path}` instead of editing it. The coder may not extend `files_owned` mid-task. This applies to both source files and any companion files (configs, fixtures, generated artifacts) discovered during implementation. If an out-of-scope edit is genuinely required, the task must be redispatched with an updated `files_owned`. (v1 tasks that fell through Step 1.5's backward-compat path are exempt from this rule, since they have no `files_owned`.)
|
|
155
|
+
|
|
156
|
+
Example BLOCKED report for an out-of-scope file:
|
|
157
|
+
|
|
158
|
+
```
|
|
159
|
+
## Task Blocked: {task.id}
|
|
160
|
+
|
|
161
|
+
Blocker: out-of-scope file required: src/lib/utils/parser.ts
|
|
162
|
+
|
|
163
|
+
Attempted:
|
|
164
|
+
- implementation per files_owned
|
|
165
|
+
|
|
166
|
+
Needs:
|
|
167
|
+
- task redispatched with src/lib/utils/parser.ts added to files_owned, OR
|
|
168
|
+
- a separate task to handle that file
|
|
169
|
+
```
|
|
170
|
+
|
|
171
|
+
**Code quality rules (always apply):**
|
|
172
|
+
- Small, focused functions — one responsibility each
|
|
173
|
+
- Dependencies injected, not hardcoded — enables testing
|
|
174
|
+
- Pure functions for business logic where possible
|
|
175
|
+
- No global mutable state
|
|
176
|
+
- No unnecessary comments — let naming convey intent
|
|
177
|
+
- No commented-out code
|
|
178
|
+
|
|
179
|
+
---
|
|
180
|
+
|
|
181
|
+
## Step 6: Handle Manual Criteria
|
|
182
|
+
|
|
183
|
+
For each `[manual]` criterion, add a comment in the relevant code or commit body describing how to verify it:
|
|
184
|
+
|
|
185
|
+
```
|
|
186
|
+
Manual Verification: {criterion text}
|
|
187
|
+
Steps: {what a reviewer should do to confirm this works}
|
|
188
|
+
```
|
|
189
|
+
|
|
190
|
+
---
|
|
191
|
+
|
|
192
|
+
## Step 7: Commit
|
|
193
|
+
|
|
194
|
+
After all tests pass, create a single commit:
|
|
195
|
+
|
|
196
|
+
```bash
|
|
197
|
+
git add {specific files}
|
|
198
|
+
git commit -m "$(cat <<'EOF'
|
|
199
|
+
{task.id}: {brief description}
|
|
200
|
+
|
|
201
|
+
- {what was added}
|
|
202
|
+
- {another change if relevant}
|
|
203
|
+
|
|
204
|
+
Acceptance Criteria:
|
|
205
|
+
- [x] {auto criterion 1}
|
|
206
|
+
- [x] {auto criterion 2}
|
|
207
|
+
- [ ] {manual criterion} — manual verification required
|
|
208
|
+
|
|
209
|
+
Co-Authored-By: Claude Sonnet <noreply@anthropic.com>
|
|
210
|
+
EOF
|
|
211
|
+
)"
|
|
212
|
+
```
|
|
213
|
+
|
|
214
|
+
Use a feature branch. Never push directly to main.
|
|
215
|
+
|
|
216
|
+
---
|
|
217
|
+
|
|
218
|
+
## Step 8: Report Back
|
|
219
|
+
|
|
220
|
+
After committing, report your outcome:
|
|
221
|
+
|
|
222
|
+
**If all auto criteria pass:**
|
|
223
|
+
|
|
224
|
+
```
|
|
225
|
+
## Task Complete: {task.id}
|
|
226
|
+
|
|
227
|
+
Implemented:
|
|
228
|
+
- {what was built}
|
|
229
|
+
|
|
230
|
+
Tests:
|
|
231
|
+
- {N} tests added, all passing
|
|
232
|
+
|
|
233
|
+
Commits:
|
|
234
|
+
- {hash}: {message}
|
|
235
|
+
|
|
236
|
+
Manual Verification Needed:
|
|
237
|
+
- {criterion}: {verification steps}
|
|
238
|
+
```
|
|
239
|
+
|
|
240
|
+
**If blocked:**
|
|
241
|
+
|
|
242
|
+
```
|
|
243
|
+
## Task Blocked: {task.id}
|
|
244
|
+
|
|
245
|
+
Blocker: {specific description of what's wrong}
|
|
246
|
+
|
|
247
|
+
Attempted:
|
|
248
|
+
- {what you tried}
|
|
249
|
+
|
|
250
|
+
Needs:
|
|
251
|
+
- {what is required to unblock}
|
|
252
|
+
```
|
|
253
|
+
|
|
254
|
+
---
|
|
255
|
+
|
|
256
|
+
## Quick Reference
|
|
257
|
+
|
|
258
|
+
### Stack Detection Order
|
|
259
|
+
|
|
260
|
+
1. `pom.xml` present → Spring Boot → load `springboot/SKILL.md`
|
|
261
|
+
2. `next.config.ts` or `next.config.js` present → Next.js → load `nextjs/SKILL.md`
|
|
262
|
+
3. `package.json` only → Node.js → load `nodejs/SKILL.md`
|
|
263
|
+
4. No skill file found → log gap message → use general best practices
|
|
264
|
+
|
|
265
|
+
### Criteria Tags
|
|
266
|
+
|
|
267
|
+
- `[auto]` — write a failing test first, implement until it passes
|
|
268
|
+
- `[manual]` — document verification steps in commit body or code comment
|
|
269
|
+
|
|
270
|
+
### Contract Validation (v2 tasks)
|
|
271
|
+
|
|
272
|
+
- All five fields required: `contract.input`, `contract.output`, `contract.invariant`, `test_plan` (≥1 entry), `files_owned` (≥1 entry)
|
|
273
|
+
- Missing → emit `task contract incomplete: {fields}` and STOP
|
|
274
|
+
- Out-of-scope file needed → emit `out-of-scope file required: {path}` and STOP
|
|
275
|
+
- Never extend `files_owned` mid-task; never infer missing contract fields
|
|
276
|
+
|
|
277
|
+
### Never
|
|
278
|
+
|
|
279
|
+
- Commit without running tests
|
|
280
|
+
- Push directly to main
|
|
281
|
+
- Implement more than the acceptance criteria require
|
|
282
|
+
- Skip the stack skill if it exists
|
|
283
|
+
- Use `git add -A` or `git add .` — stage specific files by name
|
|
284
|
+
- Edit a file outside `files_owned` (v2 tasks) — emit BLOCKED instead
|
|
285
|
+
- Guess or infer missing contract fields — emit BLOCKED instead
|
|
286
|
+
|
|
287
|
+
---
|
|
15
288
|
|
|
16
289
|
## Boundaries
|
|
17
290
|
|
|
18
291
|
- Work only on the task you were given, and only in the files it owns when a file list is provided.
|
|
19
292
|
- Do not push, open pull requests or change the task breakdown; the orchestrator does that.
|
|
20
|
-
- If the skill is not in your context for any reason, read
|
|
21
|
-
`.claude/skills/conductor-coder/SKILL.md` and follow it before writing code.
|
|
22
293
|
|
|
23
294
|
## Finish
|
|
24
295
|
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
|
-
name: conductor-
|
|
3
|
-
description: Turns the approved script into visual direction (storyboard or shot list, safe zones, sound) and, for RENDER ideas, a creative-spec.json whose fields match preview_creative_draft's inputs. Used by the conductor-
|
|
2
|
+
name: conductor-marketing-art-director
|
|
3
|
+
description: Turns the approved script into visual direction (storyboard or shot list, safe zones, sound) and, for RENDER ideas, a creative-spec.json whose fields match preview_creative_draft's inputs. Used by the conductor-marketing-create skill (team mode) as stage S5 and again for requested changes.
|
|
4
4
|
tools: Read, Write
|
|
5
5
|
model: sonnet
|
|
6
6
|
---
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
|
-
name: conductor-
|
|
3
|
-
description: Judges short-form post ideas with a fixed rubric, in two modes. text-round scores every anonymized idea and shortlists three; visual-round compares the three finalists from their rendered draft images and picks a winner. Used by the conductor-
|
|
2
|
+
name: conductor-marketing-creative-director
|
|
3
|
+
description: Judges short-form post ideas with a fixed rubric, in two modes. text-round scores every anonymized idea and shortlists three; visual-round compares the three finalists from their rendered draft images and picks a winner. Used by the conductor-marketing-create skill (team mode) as stages S3a and S3c.
|
|
4
4
|
tools: Read, Write
|
|
5
5
|
model: opus
|
|
6
6
|
---
|
package/assets/claude/agents/{conductor-content-ideator.md → conductor-marketing-ideator.md}
RENAMED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
|
-
name: conductor-
|
|
3
|
-
description: Brainstorms distinct short-form social post ideas from one assigned angle (story/emotion, utility/education, or trend/humor), tagging each RENDER or FILM against the media on hand. Three run in parallel without seeing each other. Used by the conductor-
|
|
2
|
+
name: conductor-marketing-ideator
|
|
3
|
+
description: Brainstorms distinct short-form social post ideas from one assigned angle (story/emotion, utility/education, or trend/humor), tagging each RENDER or FILM against the media on hand. Three run in parallel without seeing each other. Used by the conductor-marketing-create skill (team mode) as stage S2.
|
|
4
4
|
tools: Read, Write, WebSearch
|
|
5
5
|
model: sonnet
|
|
6
6
|
---
|
|
@@ -14,7 +14,7 @@ ideas from your angle, not a safe average.
|
|
|
14
14
|
## Boundaries
|
|
15
15
|
|
|
16
16
|
- Read only the files you are given: `brief.md`, `context/media.json` and the idea template
|
|
17
|
-
(`references/idea-template.md` in the
|
|
17
|
+
(`references/idea-template.md` in the conductor-marketing-create skill). The orchestrator gives you the paths.
|
|
18
18
|
- Write only your output file (`ideas/angle-<n>.md`).
|
|
19
19
|
- Never call Conductor or any MCP tool.
|
|
20
20
|
- WebSearch is only for the trend/humor angle, at most 3 searches, to find a format or reference
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
|
-
name: conductor-
|
|
3
|
-
description: Writes the script for the winning idea (hook, timed beats, payoff, call to action, publish copy per platform with alternate hooks) in the fixed script template. Used by the conductor-
|
|
2
|
+
name: conductor-marketing-scriptwriter
|
|
3
|
+
description: Writes the script for the winning idea (hook, timed beats, payoff, call to action, publish copy per platform with alternate hooks) in the fixed script template. Used by the conductor-marketing-create skill (team mode) as stage S4, and again when the user picks a runner-up or asks for copy changes.
|
|
4
4
|
tools: Read, Write
|
|
5
5
|
model: sonnet
|
|
6
6
|
---
|
package/assets/claude/agents/{conductor-content-strategist.md → conductor-marketing-strategist.md}
RENAMED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
|
-
name: conductor-
|
|
3
|
-
description: Turns a content request and a snapshot of workspace context (brand kit, knowledge, performance, media) into a one-page content brief. Used by the conductor-
|
|
2
|
+
name: conductor-marketing-strategist
|
|
3
|
+
description: Turns a content request and a snapshot of workspace context (brand kit, knowledge, performance, media) into a one-page content brief. Used by the conductor-marketing-create skill (team mode) as stage S1, before any ideas are generated. Reads files only; never calls Conductor.
|
|
4
4
|
tools: Read, Write, Glob
|
|
5
5
|
model: opus
|
|
6
6
|
---
|
|
@@ -281,7 +281,7 @@ For v1 tasks (no top-level `schema_version`, or no `files_owned` on tasks), beha
|
|
|
281
281
|
|
|
282
282
|
### Subagent Dispatch
|
|
283
283
|
|
|
284
|
-
For each batch, spawn `conductor-coder` subagents in parallel using the Agent tool, all calls in one message, each with `subagent_type: "conductor-coder"`. The agent (`.claude/agents/conductor-coder.md`)
|
|
284
|
+
For each batch, spawn `conductor-coder` subagents in parallel using the Agent tool, all calls in one message, each with `subagent_type: "conductor-coder"`. The agent (`.claude/agents/conductor-coder.md`) is self-contained and handles stack detection internally. Pass each agent this prompt:
|
|
285
285
|
|
|
286
286
|
```
|
|
287
287
|
# Task: {task.id} - {task.title}
|
|
@@ -0,0 +1,26 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: conductor:marketing
|
|
3
|
+
description: One entry point for marketing work. Make the artwork for a social post (from a headline, photo or clip, or from a goal), publish or schedule a post to a connected account, or build the whole-funnel marketing report. Routes to the matching skill.
|
|
4
|
+
argument-hint: "[create|publish|report] [request]"
|
|
5
|
+
allowed-tools: mcp__conductor__*, Agent, AskUserQuestion, Skill, WebSearch, WebFetch, Read, Write, Glob, Grep
|
|
6
|
+
---
|
|
7
|
+
|
|
8
|
+
# /conductor:marketing
|
|
9
|
+
|
|
10
|
+
A router. Pick the skill, then hand off; each skill owns its own flow and checks.
|
|
11
|
+
|
|
12
|
+
1. **Explicit route.** If the first argument is `create`, `publish` or `report`, go to that skill.
|
|
13
|
+
2. **Infer from the request:**
|
|
14
|
+
- Making or designing artwork, an ad, a video, a slideshow or carousel, a post idea from a goal, or
|
|
15
|
+
brand setup: `create`.
|
|
16
|
+
- Posting, scheduling or publishing something to a connected account, or checking where a
|
|
17
|
+
scheduled post stands: `publish`.
|
|
18
|
+
- A report, the funnel, growth or acquisition numbers, "how is the app doing", benchmarks: `report`.
|
|
19
|
+
3. **No request, or genuinely ambiguous:** ask once with `AskUserQuestion` (single select), "What do
|
|
20
|
+
you want to do?":
|
|
21
|
+
- "Make a post": `create`.
|
|
22
|
+
- "Publish or schedule a post": `publish`.
|
|
23
|
+
- "Marketing report": `report`.
|
|
24
|
+
4. **Hand off** with `Skill(skill: "conductor-marketing-<x>")` (`create`, `publish` or `report`),
|
|
25
|
+
passing the rest of the request and any flags through: `--quick` for create, `7d` or `28d` for
|
|
26
|
+
report. Follow the skill from there.
|
|
@@ -0,0 +1,85 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: conductor-marketing-create
|
|
3
|
+
description: Makes the artwork for a social post through the Conductor MCP tools, rendered locally as a draft and approved before anything is saved. Direct mode, when a headline, photo, clip or finished images are in hand: make an ad, short video, slideshow, carousel or story sequence, post a clip as-is, write a headline variant, or set up the brand (Brand Kit). Team mode, when there is only a goal: a team of subagents brainstorms, judges, scripts and directs a short-form post and renders a draft. Ends at a committed Creative; never creates a Post. Entry point: /conductor:marketing create.
|
|
4
|
+
user-invocable: true
|
|
5
|
+
allowed-tools: mcp__conductor__*, Agent, AskUserQuestion, Skill, Read, Write, Glob, Grep, WebFetch
|
|
6
|
+
---
|
|
7
|
+
|
|
8
|
+
# Conductor Marketing Create
|
|
9
|
+
|
|
10
|
+
Produces the artwork for a Post: upload-ready media at every placement a workspace's Brand Kit
|
|
11
|
+
enables, either from a concept the person already has (Direct) or from a goal the team turns into
|
|
12
|
+
one (Team). This is the front half of Conductor's marketing pillar; `conductor-marketing-publish` takes the
|
|
13
|
+
committed Creative from here through review and out to each destination. You make the artwork; you
|
|
14
|
+
do not post it.
|
|
15
|
+
|
|
16
|
+
## Rules that hold in both modes
|
|
17
|
+
|
|
18
|
+
- **Brand kit first.** Call `get_brand_kit`. If `configured` is `false` the workspace has never set
|
|
19
|
+
up its brand: follow `references/brand-kit-setup.md` and offer setup before making anything.
|
|
20
|
+
Copy and visuals come from the Brand Kit and its Knowledge page, not from invention.
|
|
21
|
+
- **Draft first, save after approval.** `preview_creative_draft` renders on this machine and writes
|
|
22
|
+
nothing to Conductor. Only after the person has seen the draft and approved it does
|
|
23
|
+
`commit_creative_draft` upload the media, save the Creative and run the full render. Never commit
|
|
24
|
+
a draft the person has not seen and approved.
|
|
25
|
+
- **A run never creates a Post.** It ends at a committed Creative (or a shoot script). Hand off to
|
|
26
|
+
`conductor-marketing-publish` for the Post, scheduling and review.
|
|
27
|
+
|
|
28
|
+
## Which mode
|
|
29
|
+
|
|
30
|
+
| The person has | Mode | Read |
|
|
31
|
+
|---|---|---|
|
|
32
|
+
| A headline, photo, clip or finished images, an existing Creative to vary, or wants the brand set up | **Direct** | `references/direct.md` |
|
|
33
|
+
| A goal or topic ("a post about X for Y") and wants the idea chosen for them | **Team** (honours `--quick`: two ideators, no finalist renders) | `references/team-run.md` |
|
|
34
|
+
|
|
35
|
+
Infer the mode from the request. A concrete headline, photo, clip or existing Creative (including a
|
|
36
|
+
headline variant or brand setup) is Direct. A goal or topic ("a post about X for TikTok") is Team.
|
|
37
|
+
`--quick` applies to Team only. If it is unclear, ask once with `AskUserQuestion` (single select),
|
|
38
|
+
"What are we making?", with two options: "I have the headline/photo/clip" (Direct) and "Come up with
|
|
39
|
+
an idea" (Team). Team mode finishes through the Direct flow's
|
|
40
|
+
preview, check and commit steps (S6 reads `references/direct.md`).
|
|
41
|
+
|
|
42
|
+
## References
|
|
43
|
+
|
|
44
|
+
- `references/direct.md`: the Direct flow (read the brand, angle, headline, photo, preview, critique, commit), sequences, video, finished images, short videos, music, cleanup.
|
|
45
|
+
- `references/team-run.md`: the Team flow (run folder, stages S0 to S6, FILM ideas, final message, failure handling).
|
|
46
|
+
- `references/brand-kit-setup.md`: check the Brand Kit and set it up from a website, a brand doc or direct answers.
|
|
47
|
+
- `references/hooks.md`: writing and scoring headline candidates.
|
|
48
|
+
- `references/layout-guide.md`: choosing a layout, resolution and safe zones.
|
|
49
|
+
- `references/design-critique.md`: the six-criterion score applied to every draft before it is shown.
|
|
50
|
+
- `references/realism-check.md`: the check every generated or library photo passes.
|
|
51
|
+
- `references/slideshow-formats.md`: native slideshow spec, archetypes and text rules for organic posts.
|
|
52
|
+
- `references/rubric.md`: the judging rubric the creative director uses (Team).
|
|
53
|
+
- `references/capabilities.md`: what Conductor can render, and what needs filming (Team).
|
|
54
|
+
- `references/platform-specs.md`: per-platform formats, lengths and limits (Team).
|
|
55
|
+
- `references/idea-template.md`, `script-template.md`, `shotlist-template.md`: the fixed formats the team's agents write in (Team).
|
|
56
|
+
- `references/compliance-checklist.md`: pre-publish checks for disclosure, AI label, music, claims, third-party material, accessibility (Team).
|
|
57
|
+
- `references/publish-pack-template.md`: the publish pack `conductor-marketing-publish` later pre-fills a Post from (Team).
|
|
58
|
+
|
|
59
|
+
## Failures that are silent
|
|
60
|
+
|
|
61
|
+
Automated checks are real but incomplete. These are the ones worth looking for by eye:
|
|
62
|
+
|
|
63
|
+
- **A photo with text, a logo or a claim burned into its pixels** passes every copy-rule and render
|
|
64
|
+
check, because none of them read pixels for meaning. Open and look at every photo.
|
|
65
|
+
- **A generated image depicting a product, result or claim that is not real** (a made-up app screen,
|
|
66
|
+
a before/after, an invented package) passes every check. Only looking catches it.
|
|
67
|
+
- **A structurally valid render is not necessarily a good one.** The checks cover spill, safe zones,
|
|
68
|
+
fonts and contrast, never whether the headline reads at a glance or the copy sits on a face. The
|
|
69
|
+
design critique exists for this gap.
|
|
70
|
+
- **A Brand Kit with no copy rules, or no accent-phrase requirement, never fails a save on that
|
|
71
|
+
account.** That is permissive by design, which makes the model the only check on brand voice until
|
|
72
|
+
someone configures rules.
|
|
73
|
+
- **A draft is not a saved Creative.** A Creative or media id does not exist until
|
|
74
|
+
`commit_creative_draft` runs; do not hand a draft's folder to `attach_creative_to_post` or a Post.
|
|
75
|
+
- **A stale write is refused, not overwritten.** `update_creative` requires the current `version`; a
|
|
76
|
+
409 names the conflict. Reread with `get_creative` and reapply, do not retry blind.
|
|
77
|
+
- **Rejected copy stays rejected.** Copy rules with ids starting `banned-` are lines or visuals the
|
|
78
|
+
person rejected earlier; never reuse them.
|
|
79
|
+
|
|
80
|
+
## Related
|
|
81
|
+
|
|
82
|
+
- `conductor-marketing-publish`: schedules and tracks the Post the artwork attaches to, through review and out
|
|
83
|
+
to each destination. This skill hands off to it, never the reverse.
|
|
84
|
+
- `conductor-marketing-report`: the whole-funnel marketing report; its next actions can feed new
|
|
85
|
+
posts made here.
|
|
@@ -1,8 +1,7 @@
|
|
|
1
1
|
# Brand Kit check and setup
|
|
2
2
|
|
|
3
|
-
|
|
4
|
-
Brand Kit has been set up, and offer to set it up if not.
|
|
5
|
-
unchanged.
|
|
3
|
+
Used by the `conductor-marketing-create` skill (both modes): check whether this workspace's
|
|
4
|
+
Brand Kit has been set up, and offer to set it up if not.
|
|
6
5
|
|
|
7
6
|
## Step 1 — Check the brand
|
|
8
7
|
|
|
@@ -1,4 +1,3 @@
|
|
|
1
|
-
<!-- Keep identical to the copy in the other creative skill (conductor-creative / conductor-content-studio references/design-critique.md). Skills install independently and cannot reference each other. -->
|
|
2
1
|
# Design critique
|
|
3
2
|
|
|
4
3
|
Passing checks does not make an ad good. `checks` and `passed` cover text spill, safe zones, contrast
|
|
@@ -1,29 +1,34 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
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`. (
|
|
62
|
-
a person through exactly this
|
|
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
|
|
|
@@ -143,7 +148,7 @@ concealment: the AI label still applies.
|
|
|
143
148
|
**Open the photo and look at it before using it** — a generated one most of all (pseudo-text, warped
|
|
144
149
|
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
150
|
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
|
|
151
|
+
of this skill's silent failure modes (the full list is in `SKILL.md`). A human or a prior pass marks a bad
|
|
147
152
|
photo `blocked`; if you notice one that should be, say so rather than using it anyway.
|
|
148
153
|
|
|
149
154
|
Match the crop to the subject. A layout that lays copy over the photo (`bleed`) can put a face in
|
|
@@ -252,8 +257,19 @@ Creative is already saved — call `render_creative`, don't commit again. Verify
|
|
|
252
257
|
Then `attach_creative_to_post` with the Post's Work Item id — it maps each placement to the right
|
|
253
258
|
destinations (9:16 to TikTok/Reels, 4:5 to Instagram feed, 1:1 to Facebook, story to story targets)
|
|
254
259
|
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-
|
|
256
|
-
publishing — this skill's job ends at attaching the artwork.
|
|
260
|
+
with `get_post_status`. From here, use the `conductor-marketing-publish` skill for scheduling, review and
|
|
261
|
+
publishing — this skill's job ends at attaching the artwork. If there is no Post yet, say so plainly
|
|
262
|
+
and name the next step (creating one with `conductor-marketing-publish`) rather than leaving the artwork with
|
|
263
|
+
no stated destination.
|
|
264
|
+
|
|
265
|
+
When the Creative came from a team run, or the person points to a publish pack, look for
|
|
266
|
+
`.conductor/content-runs/*/publish-pack.json` whose `creativeId` matches. It carries per-destination
|
|
267
|
+
captions and publish options (disclosure, AI label, cover frame, alt text) and a pinned comment. When
|
|
268
|
+
you go on to create or edit the Post with `conductor-marketing-publish`, pre-fill each target's
|
|
269
|
+
`captionOverride` and options from it, show them in the confirmation table and let the person change
|
|
270
|
+
them. Mention the pinned comment (and, for a post that promotes an app, the Stories repost with a link
|
|
271
|
+
sticker) as manual steps, since Conductor cannot post comments. The Post still goes through the normal
|
|
272
|
+
review.
|
|
257
273
|
|
|
258
274
|
## Making a sequence instead of a single ad
|
|
259
275
|
|
|
@@ -347,7 +363,7 @@ thing).
|
|
|
347
363
|
accepts a chosen track on photos, so each destination sends it as `carouselAs: "photos"`
|
|
348
364
|
(silent) or `"video"` (the slideshow MP4 with the track). Attach defaults to video for TikTok
|
|
349
365
|
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` (
|
|
366
|
+
options; change it with `set_publish_targets` (`conductor-marketing-publish` skill). Mind each platform's length limit
|
|
351
367
|
for the video.
|
|
352
368
|
|
|
353
369
|
## Short videos from a photo or clip
|
|
@@ -412,30 +428,3 @@ are the same for all three:
|
|
|
412
428
|
## Cleaning up
|
|
413
429
|
|
|
414
430
|
`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.
|