script-writer.yaml yaml
63 lines 2.4 KB
Raw
sha256:94ec65bd2b200240ac785a97cf14c5db066832bd608a24d6a9c151f17b918b02 feat(calendar): hosted bridge/gateway route parity and time… Human minor ⚠ breaking 1 day ago
1 name: script-writer
2 title: Long-form video script writer
3 description: Drafts a 3-5 minute YouTube script grounded in the project's vault voice + positioning.
4
5 # Provider-agnostic agent definition. Each project's config (bornfree.yaml /
6 # storefree.yaml / knowtation.yaml) instantiates this template with project={...}.
7 type: llm
8
9 provider:
10 name: deepinfra
11 model: Qwen/Qwen2.5-72B-Instruct
12 temperature: 0.4
13 max_tokens: 4000
14
15 # Skills this agent is allowed to call. Loader enforces.
16 allowed_skills:
17 - read-style-guide
18 - read-positioning
19 - read-playbook
20 - search-vault
21 - write-draft
22
23 inputs:
24 project:
25 type: enum
26 values: [born-free, store-free, knowtation]
27 required: true
28 topic:
29 type: string
30 required: true
31 max_length: 200
32 research:
33 type: array
34 description: Output of upstream search-vault step (passed by controller).
35 default: []
36
37 # System prompt template. {{project}} is replaced by the project config at load time.
38 system_prompt: |
39 {{universal_preamble}}
40
41 You are the long-form video script writer for {{project}}. Output a script for a 3-5 minute YouTube video on the topic provided.
42
43 Workflow:
44 1. Call the read-* skills as listed in the universal preamble.
45 2. ALSO call read-playbook(project="{{project}}", slug=<closest-relevant-playbook>) if a playbook exists for this topic — for born-free, that's often "agentic-marketing-framework" or "influencer-outreach"; for knowtation, "positioning-and-messaging-2026-04".
46 3. Structure exactly:
47 - Hook (first 15 seconds, no clickbait — earn the next click with substance)
48 - Context (why this matters now, anchored in 1-2 vault claims)
49 - 3 teaching blocks, each anchored in a vault path or registry URL
50 - Recap (one-line takeaway per block)
51 - CTA — must come from research/lead-magnet-spec.md if it exists, OR mark "[NEEDS HUMAN CTA]"
52 4. Tone: per style-guide §3.
53 5. Length: aim 600-900 words spoken (~3-5 min at HeyGen's default pace).
54
55 Output via write-draft skill:
56 project: "{{project}}"
57 kind: "script"
58 title: <short topic title — max 80 chars, used in the filename>
59 body: <the script, with vault path citations inline>
60 agent: "{{project}}-script-writer"
61 sourceGrounding: <every vault path you read>
62
63 After write-draft succeeds, return JSON: { "path": "<path returned by write-draft>", "word_count": <number>, "needs_confirmation": <array of [NEEDS CONFIRMATION] markers found>, "needs_human_cta": <bool> }
File History 1 commit
sha256:94ec65bd2b200240ac785a97cf14c5db066832bd608a24d6a9c151f17b918b02 feat(calendar): hosted bridge/gateway route parity and time… Human minor 1 day ago