- Home
- Skills
- Willoscar
- Research Units Pipeline Skills
- Opener Variator
opener-variator_skill
- Python
109
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill willoscar/research-units-pipeline-skills --skill opener-variator- SKILL.md4.6 KB
Overview
This skill rewrites the opening paragraph of flagged subsection files to remove generator-style narration and repeated opener cadences. It preserves citations and argument moves while converting overview or “we” narration into a content-bearing, varied opener that ends with a clear thesis or takeaway. The transformation is narrowly scoped to the first paragraph (1–2 sentences, up to 4) of each H3 body file.
How this skill works
The skill reads the writer self-loop report to locate files flagged for style smells and opens only those sections/S*.md files. For each file it rewrites the first paragraph using one chosen opener mode (tension, decision, failure, protocol, contrast, or lens) while keeping all citation keys and the subsection’s original thesis intact. It writes directly to the sections/S*.md files and expects the writer-selfloop check to show reduced Style Smells after rerunning.
When to use it
- After writer-selfloop is PASS but the Style Smells list flags repeated opener stems or overview narration
- When subsection openers read like a generated table-of-contents or show a repeated rhetorical cadence
- When you need minimal, targeted copy edits that preserve evidence and citations
- When the section is substantive enough to support a content-bearing opener (skip if evidence-thin)
Best practices
- Only rewrite the first paragraph (1–4 sentences); avoid touching the rest of the subsection
- Do not invent facts or move/add/remove citation keys
- Choose one opener mode per subsection and make sentence one content-bearing, not navigational
- Skip this pass for pre-C2 drafts or sections lacking evidence; route upstream instead
- Optionally consult outline/writer_context_packs.jsonl for opener_mode, tension_statement, and thesis to stay aligned
Example use cases
- A survey subsection where every H3 begins with “This subsection provides…” and needs varied framing
- A draft showing a subtle generator cadence across H3 openers while argument and citations are correct
- A paper needing consistent, concrete thesis-bearing openers so each subsection reads authored rather than templated
FAQ
No. The skill must not invent facts or add/remove/move citation keys; it only rewrites opener phrasing while preserving meaning and citations.
How large are the edits?
Edits are intentionally small: only the first paragraph (1–4 sentences) of each flagged H3 is rewritten to a content-bearing opener that ends with the same thesis.
When should I not run this skill?
Skip it for pre-C2 drafts (no prose) and for sections that are evidence-thin; those should be routed upstream rather than stylized.