# Limitation Weaver

Rewrite limitation passages so the paper keeps limitations without falling into...

- Repository: https://github.com/WILLOSCAR/research-units-pipeline-skills (517 stars)
- Category: Engineering
- License: see repository
- Last updated: 2026-10-05
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver -g -a claude-code -y`
Codex: `npx skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver -g -a codex -y`
Cursor: `npx skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver -g -a cursor -y`
Gemini CLI: `npx skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver -g -a gemini-cli -y`

## Description

Rewrite limitation passages so the paper keeps limitations without falling into count-based slot phrases (e.g., \\\"Two limitations…\\\") across many H3s.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Limitation Weaver (keep caveats, lose the slot phrase)

## Triggers & routing

- **Trigger**: limitation weaver, rewrite limitations, remove two limitations, 去Two limitations, 局限改写, caveat rewrite.


Purpose: keep survey-grade intellectual honesty **without** triggering a strong generator-voice tell:
- repeated count-based openers (\"Two limitations…\", \"Three takeaways…\")

This is not about removing limitations.
It is about expressing them in a paper-like way that varies naturally across sections.

## Inputs

Required:
- `output/WRITER_SELFLOOP_TODO.md` (Style Smells section)
- the referenced `sections/S<sub_id>.md` files

Optional (helps keep limitations grounded):
- `outline/writer_context_packs.jsonl` (use `failures_limitations` / `limitation_hooks` / `verify_fields` when present)

## Workflow (explicit inputs)

- Start from `output/WRITER_SELFLOOP_TODO.md` (Style Smells) to locate the exact `sections/S*.md` files to rewrite.
- Use `outline/writer_context_packs.jsonl` to keep limitations grounded in the subsection's evidence boundary (no guessing).

## Outputs

- Updated `sections/S<sub_id>.md` files (still body-only; no headings)

## Role prompt: Caveat Editor (paper voice)

```text
You are editing the limitation content of a survey subsection.

Goal:
- preserve the subsection-specific limitation(s)
- remove count-based opener slots and repetitive cadence
- keep limitations tied to the protocol/evidence boundary (what changes interpretation)

Constraints:
- do not invent facts
- do not add/remove/move citation keys
- do not weaken the section by deleting real limitations
```

## Anti-pattern (rewrite immediately)

- `Two limitations stand out. First, ... Second, ...`
- `Three key takeaways are ...`

Why it hurts: it creates a reusable template slot that repeats across H3s and reads auto-generated.

## Rewrite moves (choose one; vary across H3s)

The rest: https://github.com/WILLOSCAR/research-units-pipeline-skills/blob/HEAD/.codex/skills/limitation-weaver/SKILL.md
