limitation-weaver

Name: limitation-weaver
Author: WILLOSCAR

489stars39forks

How to install limitation-weaver?

willoscar/research-units-pipeline-skills/limitation-weaver

$npx -y skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver

Installs into the current project.

Summary

Rewrite limitation passages so the paper keeps limitations without falling into count-based slot phrases (e.g., \"Two limitations…\") across many H3s. Trigger: limitation weaver, rewrite limitations, remove two limitations, 去Two limitations, 局限改写, caveat rewrite.

Files · 1

View on GitHub

SKILL.md

1	# Limitation Weaver (keep caveats, lose the slot phrase)
2
3	Purpose: keep survey-grade intellectual honesty without triggering a strong generator-voice tell:
4	- repeated count-based openers (\"Two limitations…\", \"Three takeaways…\")
5
6	This is not about removing limitations.
7	It is about expressing them in a paper-like way that varies naturally across sections.
8
9	## Inputs
10
11	Required:
12	- `output/WRITER_SELFLOOP_TODO.md` (Style Smells section)
13	- the referenced `sections/S<sub_id>.md` files
14
15	Optional (helps keep limitations grounded):
16	- `outline/writer_context_packs.jsonl` (use `failures_limitations` / `limitation_hooks` / `verify_fields` when present)
17
18	## Workflow (explicit inputs)
19
20	- Start from `output/WRITER_SELFLOOP_TODO.md` (Style Smells) to locate the exact `sections/S*.md` files to rewrite.
21	- Use `outline/writer_context_packs.jsonl` to keep limitations grounded in the subsection's evidence boundary (no guessing).
22
23	## Outputs
24
25	- Updated `sections/S<sub_id>.md` files (still body-only; no headings)
26
27	## Role prompt: Caveat Editor (paper voice)
28
29	```text
30	You are editing the limitation content of a survey subsection.
31
32	Goal:
33	- preserve the subsection-specific limitation(s)
34	- remove count-based opener slots and repetitive cadence
35	- keep limitations tied to the protocol/evidence boundary (what changes interpretation)
36
37	Constraints:
38	- do not invent facts
39	- do not add/remove/move citation keys
40	- do not weaken the section by deleting real limitations
41	```
42
43	## Anti-pattern (rewrite immediately)
44
45	- `Two limitations stand out. First, ... Second, ...`
46	- `Three key takeaways are ...`
47
48	Why it hurts: it creates a reusable template slot that repeats across H3s and reads auto-generated.
49
50	## Rewrite moves (choose one; vary across H3s)
51
52	1) Fold caveat into a contrast paragraph (preferred)
53	- Put one caveat sentence as the last sentence of the A-vs-B paragraph.
54	- Shape: “However, …; this matters because …”
55
56	2) Single caveat paragraph without counting
57	- Start with a natural opener (rotate across H3s; avoid repeating the same stem):
58	- “These results hinge on …”
59	- “Interpretation depends on …”
60	- “Evidence is thin when …”
61	- “A caveat is that …” (use sparingly)
62	- Then add one sentence that explains why it changes interpretation.
63
64	3) Verification-target framing (when evidence is abstract-only / underspecified)
65	- Convert the limitation into a checkable condition:
66	- “To make this comparison robust, evaluations need to report …”
67	- Keep it concrete (budget/tool access/logging/threat model), and do not repeat this pattern across many H3s.
68
69	## Mini examples (paraphrase; do not copy)
70
71	Bad:
72	- `Two limitations temper strong conclusions. First, budgets differ. Second, ablations are missing.`
73
74	Better (folded into contrast):
75	- `...; however, reported budgets and retry policies vary widely, which makes head-to-head comparisons fragile unless those constraints are normalized.`
76
77	Better (single caveat paragraph):
78	- `These results hinge on under-specified verification and retry policies; this matters because success rates can shift substantially along the success–cost frontier.`
79
80	## Done checklist
81
82	- [ ] No rewritten subsection uses count-based limitation openers as a default structure.
83	- [ ] Limitations still exist and remain subsection-specific.
84	- [ ] Citation keys are unchanged.
85	- [ ] `writer-selfloop` remains PASS and Style Smells shrink.

People who liked this also viewed

6 picks

Type

limitation-weaver

489stars39forks

View on GitHub

How to install limitation-weaver?

willoscar/research-units-pipeline-skills/limitation-weaver

$npx -y skills add WILLOSCAR/research-units-pipeline-skills --skill limitation-weaver

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Limitation Weaver (keep caveats, lose the slot phrase)
2
3	Purpose: keep survey-grade intellectual honesty without triggering a strong generator-voice tell:
4	- repeated count-based openers (\"Two limitations…\", \"Three takeaways…\")
5
6	This is not about removing limitations.
7	It is about expressing them in a paper-like way that varies naturally across sections.
8
9	## Inputs
10
11	Required:
12	- `output/WRITER_SELFLOOP_TODO.md` (Style Smells section)
13	- the referenced `sections/S<sub_id>.md` files
14
15	Optional (helps keep limitations grounded):
16	- `outline/writer_context_packs.jsonl` (use `failures_limitations` / `limitation_hooks` / `verify_fields` when present)
17
18	## Workflow (explicit inputs)
19
20	- Start from `output/WRITER_SELFLOOP_TODO.md` (Style Smells) to locate the exact `sections/S*.md` files to rewrite.
21	- Use `outline/writer_context_packs.jsonl` to keep limitations grounded in the subsection's evidence boundary (no guessing).
22
23	## Outputs
24
25	- Updated `sections/S<sub_id>.md` files (still body-only; no headings)
26
27	## Role prompt: Caveat Editor (paper voice)
28
29	```text
30	You are editing the limitation content of a survey subsection.
31
32	Goal:
33	- preserve the subsection-specific limitation(s)
34	- remove count-based opener slots and repetitive cadence
35	- keep limitations tied to the protocol/evidence boundary (what changes interpretation)
36
37	Constraints:
38	- do not invent facts
39	- do not add/remove/move citation keys
40	- do not weaken the section by deleting real limitations
41	```
42
43	## Anti-pattern (rewrite immediately)
44
45	- `Two limitations stand out. First, ... Second, ...`
46	- `Three key takeaways are ...`
47
48	Why it hurts: it creates a reusable template slot that repeats across H3s and reads auto-generated.
49
50	## Rewrite moves (choose one; vary across H3s)
51
52	1) Fold caveat into a contrast paragraph (preferred)
53	- Put one caveat sentence as the last sentence of the A-vs-B paragraph.
54	- Shape: “However, …; this matters because …”
55
56	2) Single caveat paragraph without counting
57	- Start with a natural opener (rotate across H3s; avoid repeating the same stem):
58	- “These results hinge on …”
59	- “Interpretation depends on …”
60	- “Evidence is thin when …”
61	- “A caveat is that …” (use sparingly)
62	- Then add one sentence that explains why it changes interpretation.
63
64	3) Verification-target framing (when evidence is abstract-only / underspecified)
65	- Convert the limitation into a checkable condition:
66	- “To make this comparison robust, evaluations need to report …”
67	- Keep it concrete (budget/tool access/logging/threat model), and do not repeat this pattern across many H3s.
68
69	## Mini examples (paraphrase; do not copy)
70
71	Bad:
72	- `Two limitations temper strong conclusions. First, budgets differ. Second, ablations are missing.`
73
74	Better (folded into contrast):
75	- `...; however, reported budgets and retry policies vary widely, which makes head-to-head comparisons fragile unless those constraints are normalized.`
76
77	Better (single caveat paragraph):
78	- `These results hinge on under-specified verification and retry policies; this matters because success rates can shift substantially along the success–cost frontier.`
79
80	## Done checklist
81
82	- [ ] No rewritten subsection uses count-based limitation openers as a default structure.
83	- [ ] Limitations still exist and remain subsection-specific.
84	- [ ] Citation keys are unchanged.
85	- [ ] `writer-selfloop` remains PASS and Style Smells shrink.

People who liked this also viewed

6 picks

Type