evidence-review

Name: evidence-review
Author: yha9806

24stars3forks

How to install evidence-review?

yha9806/academic-writing-toolkit/evidence-review

$npx -y skills add yha9806/academic-writing-toolkit --skill evidence-review

Installs into the current project.

Summary

Build evidence-controlled literature reviews and gap maps with source-status labels, claim registers, citation-role plans, traceability tables, and overclaim audits. Use when drafting or auditing review papers, thesis literature reviews, scoping reviews, or evidence syntheses whe

Files · 1

View on GitHub

SKILL.md

1	# /evidence-review - Evidence-Controlled Review Skill
2
3	## Purpose
4
5	Create literature-review and review-paper workflows that are traceable from evidence to claim. This skill extends the toolkit's read-note-map-integrate pipeline with stricter evidence-status control, gap mapping, citation-role planning, paragraph-level claim traceability, and overclaim auditing.
6
7	## Trigger Words
8
9	This skill activates on: `evidence review`, `gap review`, `gap map`, `claim traceability`, `overclaim audit`, `review paper`, `scoping review`, `systematic narrative review`, `/evidence-review`.
10
11	## Core Rules
12
13	1. Keep the central review scope explicit.
14	2. Separate direct evidence, background evidence, methodological support, workflow/governance support, candidate-only records, unpublished work, and in-progress work.
15	3. Do not fabricate references, DOI, PMID, sample size, venue, abstracts, results, or claims.
16	4. Do not treat candidate-only, metadata-only, abstract-only, generated-stub, or manual-review records as full-text evidence.
17	5. Do not treat adjacent-domain evidence as direct validation of the target domain.
18	6. Preserve uncertainty and boundary language.
19	7. Use the project's citation style or placeholder convention. If the project uses review placeholders, preserve `[CITE: CitationKey]`.
20	8. Draft section by section. Do not assemble a final chapter or paper until section-level controls are clean.
21
22	## Workflow
23
24	### 1. Scope And Boundary Setup
25
26	Create a short scope statement with:
27
28	- central review question
29	- direct evidence domain
30	- adjacent background domains
31	- methodological support domains
32	- excluded topics
33	- key overclaim risks
34
35	### 2. Evidence Inventory
36
37	Inspect existing project files before drafting. Common locations include:
38
39	- `literature/reading_notes/`
40	- `summaries/`
41	- `evidence/`
42	- `outputs/references_master.csv`
43	- `outputs/search_log.md`
44
45	Classify each source using `references/evidence_status_schema.md`.
46
47	### 3. Gap Map
48
49	Create or update:
50
51	- `evidence/evidence_matrix.csv`
52	- `evidence/claim_register.csv`
53	- `evidence/citation_plan.csv`
54	- `evidence/remaining_gap_notes.csv`
55	- `evidence/overclaim_risk_register.csv`
56
57	Map each source to what it can legitimately support. If direct evidence is sparse, state that as a gap instead of filling the gap with background literature.
58
59	### 4. Section Drafting
60
61	For each section:
62
63	1. Create an input-check report.
64	2. Draft first-pass prose.
65	3. Create paragraph-level claim traceability.
66	4. Create citation usage audit.
67	5. Create overclaim audit.
68	6. Create revision notes.
69	7. Revise into second-pass prose only after controls are clean.
70
71	### 5. Assembly Readiness
72
73	Before merging sections:
74
75	- check every section has compatible draft maturity
76	- check traceability and audit files exist
77	- check CSV files import cleanly
78	- audit cross-section flow
79	- audit repetition and boundary-language duplication
80	- audit integrated citation consistency
81	- audit integrated overclaim risks
82
83	### 6. Export Or Sharing
84
85	Use `/export` only after the review has passed assembly and overclaim checks. Export is packaging, not revision.
86
87	## Reading Notes Bridge
88
89	When the review starts from source reading, use the template in `references/reading_notes_template.md`. Every source note should record:
90
91	- citation key or provisional source ID
92	- source provenance
93	- evidence status
94	- relevance
95	- key arguments
96	- detailed notes
97	- review connections
98	- claim candidates
99	- limitations and boundary cautions
100	- follow-up questions
101
102	## Output Templates
103
104	Use `references/review_workflow_templates.md` for recommended CSV columns and report structures.
105
106	Minimum section controls:
107
108	- `reports/<section>_input_check.md`
109	- `docs/<section>_first_pass_draft.md` or `chapters/<section>_first_pass_draft.md`
110	- `evidence/<section>_claim_traceability.csv`
111	- `evidence/<section>_citation_usage_audit.csv`
112	- `reports/<section>_overclaim_audit.md`
113	- `reports/<section>_revision_notes.md`
114
115	## Optional Package Check
116
117	If Python is available, run:
118
119	```bash
120	python {skill_dir}/scripts/check_review_package.py <project_root>
121	```
122
123	This checks common directories, common evidence-control files, and CSV readability. It does not judge scientific validity.
124
125	## Stop Conditions
126
127	Stop and report a blocker if:
128
129	- the user prohibits searching but the requested claim requires new evidence
130	- only candidate or abstract evidence exists for a central claim
131	- a clinical, causal, deployment, or outcome claim lacks direct evidence
132	- unpublished or in-progress work cannot be separated from published literature
133	- c

People who liked this also viewed

6 picks

Type

evidence-review

24stars3forks

View on GitHub

How to install evidence-review?

yha9806/academic-writing-toolkit/evidence-review

$npx -y skills add yha9806/academic-writing-toolkit --skill evidence-review

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# /evidence-review - Evidence-Controlled Review Skill
2
3	## Purpose
4
5	Create literature-review and review-paper workflows that are traceable from evidence to claim. This skill extends the toolkit's read-note-map-integrate pipeline with stricter evidence-status control, gap mapping, citation-role planning, paragraph-level claim traceability, and overclaim auditing.
6
7	## Trigger Words
8
9	This skill activates on: `evidence review`, `gap review`, `gap map`, `claim traceability`, `overclaim audit`, `review paper`, `scoping review`, `systematic narrative review`, `/evidence-review`.
10
11	## Core Rules
12
13	1. Keep the central review scope explicit.
14	2. Separate direct evidence, background evidence, methodological support, workflow/governance support, candidate-only records, unpublished work, and in-progress work.
15	3. Do not fabricate references, DOI, PMID, sample size, venue, abstracts, results, or claims.
16	4. Do not treat candidate-only, metadata-only, abstract-only, generated-stub, or manual-review records as full-text evidence.
17	5. Do not treat adjacent-domain evidence as direct validation of the target domain.
18	6. Preserve uncertainty and boundary language.
19	7. Use the project's citation style or placeholder convention. If the project uses review placeholders, preserve `[CITE: CitationKey]`.
20	8. Draft section by section. Do not assemble a final chapter or paper until section-level controls are clean.
21
22	## Workflow
23
24	### 1. Scope And Boundary Setup
25
26	Create a short scope statement with:
27
28	- central review question
29	- direct evidence domain
30	- adjacent background domains
31	- methodological support domains
32	- excluded topics
33	- key overclaim risks
34
35	### 2. Evidence Inventory
36
37	Inspect existing project files before drafting. Common locations include:
38
39	- `literature/reading_notes/`
40	- `summaries/`
41	- `evidence/`
42	- `outputs/references_master.csv`
43	- `outputs/search_log.md`
44
45	Classify each source using `references/evidence_status_schema.md`.
46
47	### 3. Gap Map
48
49	Create or update:
50
51	- `evidence/evidence_matrix.csv`
52	- `evidence/claim_register.csv`
53	- `evidence/citation_plan.csv`
54	- `evidence/remaining_gap_notes.csv`
55	- `evidence/overclaim_risk_register.csv`
56
57	Map each source to what it can legitimately support. If direct evidence is sparse, state that as a gap instead of filling the gap with background literature.
58
59	### 4. Section Drafting
60
61	For each section:
62
63	1. Create an input-check report.
64	2. Draft first-pass prose.
65	3. Create paragraph-level claim traceability.
66	4. Create citation usage audit.
67	5. Create overclaim audit.
68	6. Create revision notes.
69	7. Revise into second-pass prose only after controls are clean.
70
71	### 5. Assembly Readiness
72
73	Before merging sections:
74
75	- check every section has compatible draft maturity
76	- check traceability and audit files exist
77	- check CSV files import cleanly
78	- audit cross-section flow
79	- audit repetition and boundary-language duplication
80	- audit integrated citation consistency
81	- audit integrated overclaim risks
82
83	### 6. Export Or Sharing
84
85	Use `/export` only after the review has passed assembly and overclaim checks. Export is packaging, not revision.
86
87	## Reading Notes Bridge
88
89	When the review starts from source reading, use the template in `references/reading_notes_template.md`. Every source note should record:
90
91	- citation key or provisional source ID
92	- source provenance
93	- evidence status
94	- relevance
95	- key arguments
96	- detailed notes
97	- review connections
98	- claim candidates
99	- limitations and boundary cautions
100	- follow-up questions
101
102	## Output Templates
103
104	Use `references/review_workflow_templates.md` for recommended CSV columns and report structures.
105
106	Minimum section controls:
107
108	- `reports/<section>_input_check.md`
109	- `docs/<section>_first_pass_draft.md` or `chapters/<section>_first_pass_draft.md`
110	- `evidence/<section>_claim_traceability.csv`
111	- `evidence/<section>_citation_usage_audit.csv`
112	- `reports/<section>_overclaim_audit.md`
113	- `reports/<section>_revision_notes.md`
114
115	## Optional Package Check
116
117	If Python is available, run:
118
119	```bash
120	python {skill_dir}/scripts/check_review_package.py <project_root>
121	```
122
123	This checks common directories, common evidence-control files, and CSV readability. It does not judge scientific validity.
124
125	## Stop Conditions
126
127	Stop and report a blocker if:
128
129	- the user prohibits searching but the requested claim requires new evidence
130	- only candidate or abstract evidence exists for a central claim
131	- a clinical, causal, deployment, or outcome claim lacks direct evidence
132	- unpublished or in-progress work cannot be separated from published literature
133	- c

People who liked this also viewed

6 picks

Type