Academic Writing
revision-strategy-planner
Builds prioritized manuscript revision plans for major or minor revisions by separating comments that require experiments, analyses, clarification, restructuring, or wording changes.
91100Total Score
Core Capability
92 / 100
Functional Suitability
12 / 12
Reliability
11 / 12
Performance & Context
7 / 8
Agent Usability
16 / 16
Human Usability
7 / 8
Security
12 / 12
Maintainability
11 / 12
Agent-Specific
16 / 20
Medical Task
34 / 34 Passed
88Major revision with 3 reviewer comments: one requesting new experiments, one requesting additional analysis, one requesting clarification
5/5
89Minor revision with 4 reviewer comments, all presentation and clarity related
5/5
94User provides only 'we got a major revision, now what?' with no reviewer comments
5/5
88Reject-and-resubmit decision with one fundamental design concern about confounding variables
5/5
86Full review package with 3 reviewers providing 18 comments total, including contradictory requests
5/5
91User asks to write the actual point-by-point rebuttal letter rather than a revision plan
4/4
91User says they want to promise new animal experiments in 3 months to satisfy Reviewer 2's demand
5/5
Veto GatesRequired pass for any deployment consideration
Skill Veto✓ All 4 gates passed
✓
Operational Stability
System remains stable across varied inputs and edge cases
PASS✓
Structural Consistency
Output structure conforms to expected skill contract format
PASS✓
Result Determinism
Equivalent inputs produce semantically equivalent outputs
PASS✓
System Security
No prompt injection, data leakage, or unsafe tool use detected
PASSResearch Veto✅ PASS — Applicable
| Dimension | Result | Detail |
|---|---|---|
| Scientific Integrity | PASS | No fabricated reviewer comments, editor positions, PMIDs, or assay feasibility claims. Hard rule 7 explicitly prohibits fabricating dataset availability or validation status. |
| Practice Boundaries | PASS | No diagnostic or prescriptive medical conclusions. Revision planning is strategic, not clinical. |
| Methodological Ground | PASS | Hard rule 2 prevents reflexive new-experiment promises. Hard rule 3 prevents assuming all misunderstandings can be resolved with wording. Feasibility boundary rules enforce resource-aware planning. |
| Code Usability | N/A | Mode A skill — no code generated. |
Core Capability92 / 100 — 8 Categories
Functional Suitability
Comprehensive revision strategy planning. Seven use cases including reject-and-resubmit. Nine-step workflow + ten-section output (A–J). Eight action routes (new experiment/analysis/re-analysis/clarification/wording/figure/restructuring/bounded rebuttal). Three feasibility tiers. Section H (Main Strategic Risk) is unique across all Academic Writing skills reviewed.
12 / 12
100%
Reliability
Clarification-first gate. Five-dimension triage (severity, editorial sensitivity, revision burden, credibility, centrality). 'Lightest credible action' principle prevents reflexive overcommitment. Minor deduction: no partial-results mode when input is sparse but user insists on proceeding.
11 / 12
92%
Performance & Context
Seven compact reference files. SKILL.md 314 lines — longest in Academic Writing category. Ten-section output is verbose but each section has distinct purpose. Minor deduction: Section D (Priority Revision Plan) and Section G (Recommended Revision Phases) have structural overlap — both sequence the revision work.
7 / 8
88%
Agent Usability
Full marks. Seven sample triggers (most in set). Eight-item core function list. Quality standard comparison. Ten fixed A–J labels. Five feedback mechanisms across sections A, C, F, H, J. 'Lightest credible action' is an explicit usability principle that guides output quality.
16 / 16
100%
Human Usability
Seven sample triggers make entry points very clear. Section J + upload recommendation provide follow-up guidance. Minor deduction: no restart path when user provides more reviewer comments after partial triage begins.
7 / 8
88%
Security
No credentials, APIs, or code execution. Hard rules 1 and 7 prevent fabricating reviewer comments or assay feasibility. Hard rule 4 prevents overcommitment to experiments without feasibility support. Hard rule 8 requires explicit three-tier feasibility classification.
12 / 12
100%
Maintainability
Seven focused reference files; adding new action route requires only updating action-routing-rules.md. Clean separation between triage logic, action routing, feasibility, and priority staging. Minor deduction: no worked example of a complete revision plan for a specific revision scenario.
11 / 12
92%
Agent-Specific
Trigger precision: seven specific sample triggers (4/4). Progressive disclosure: clarification gate + Section A + Section J (3/4 — no three-way decision fork). Composability: no explicit handoff to author-response-builder for executing the plan (2/4). Idempotency: A–J structure stable (4/4). Escape hatches: Section J + upload recommendation + bounded rebuttal as formal action route (3/4 — bounded rebuttal is unique but no partial-plan mode).
16 / 20
80%
Core Capability Total92 / 100
Medical TaskExecution Average: 89.6 / 100 — Assertions: 34/34 Passed
88
Canonical
Major revision with 3 reviewer comments: one requesting new experiments, one requesting additional analysis, one requesting clarification
5/5 ✓
89
Variant A
Minor revision with 4 reviewer comments, all presentation and clarity related
5/5 ✓
94
Edge
User provides only 'we got a major revision, now what?' with no reviewer comments
5/5 ✓
88
Variant B
Reject-and-resubmit decision with one fundamental design concern about confounding variables
5/5 ✓
86
Stress
Full review package with 3 reviewers providing 18 comments total, including contradictory requests
5/5 ✓
91
Scope Boundary
User asks to write the actual point-by-point rebuttal letter rather than a revision plan
4/4 ✓
91
Adversarial
User says they want to promise new animal experiments in 3 months to satisfy Reviewer 2's demand
5/5 ✓
88
Canonical✅ Pass
Major revision with 3 reviewer comments: one requesting new experiments, one requesting additional analysis, one requesting clarification
All five assertions passed. Comments correctly triaged by severity. New experiment request correctly assessed for feasibility before accepting. Lightest credible action applied to clarification request.
Basic 36/40|Specialized 52/60|Total 88/100
✅A1Output correctly triages the three comments by severity, not by length or order
✅A2New experiment request routed with explicit feasibility assessment before commitment
✅A3Clarification-only comment correctly routed to wording change, not new analysis
✅A4Section H identifies the new-experiment request as the main strategic risk
✅A5Section F explicitly separates currently feasible vs. potentially feasible vs. unrealistic actions
Pass rate: 5 / 5
89
Variant A✅ Pass
Minor revision with 4 reviewer comments, all presentation and clarity related
All five assertions passed. All four comments correctly classified as presentation-only. No new experiment or analysis suggested. Realistic timeline for minor revision stated.
Basic 37/40|Specialized 52/60|Total 89/100
✅A1Output correctly identifies all four comments as presentation-only, not evidence gaps
✅A2No new experiments or re-analyses suggested for presentation-only comments
✅A3Section F classifies all four actions as currently feasible
✅A4Section H correctly states the main strategic risk is low for this minor revision set
✅A5Output does not over-weight minor comments as requiring extensive new evidence
Pass rate: 5 / 5
94
Edge✅ Pass
User provides only 'we got a major revision, now what?' with no reviewer comments
All five assertions passed. Clarification-first gate triggered. No fabricated revision plan produced. Specific missing inputs listed.
Basic 39/40|Specialized 55/60|Total 94/100
✅A1Output triggers clarification-first gate and requests reviewer comments before planning
✅A2Section A states explicitly that input is insufficient for high-confidence revision planning
✅A3Output does not invent reviewer comments or revision items
✅A4Section J lists specific missing inputs that would enable a real revision plan
✅A5Output explains what information is needed and why, not just that it is missing
Pass rate: 5 / 5
88
Variant B✅ Pass
Reject-and-resubmit decision with one fundamental design concern about confounding variables
All five assertions passed. Reject-and-resubmit correctly identified as highest-stakes context. Confounding concern correctly classified as major scientific issue. Bounded rebuttal vs. new analysis trade-off correctly assessed.
Basic 37/40|Specialized 51/60|Total 88/100
✅A1Output identifies reject-and-resubmit as requiring a higher evidence bar than major revision
✅A2Confounding variable concern classified as a major scientific issue, not a wording issue
✅A3Section E routes confounding concern to re-analysis with existing data first, then new experiment if needed
✅A4Section H identifies the confounding concern as the primary acceptance risk
✅A5Section F distinguishes re-analysis as currently feasible from new experiment as potentially feasible
Pass rate: 5 / 5
86
Stress✅ Pass
Full review package with 3 reviewers providing 18 comments total, including contradictory requests
All five assertions passed. 18 comments correctly clustered into priority tiers. Contradictory requests identified. Editor-sensitive comments correctly elevated.
Basic 36/40|Specialized 50/60|Total 86/100
✅A1Output clusters 18 comments into thematic priority groups, not a flat 18-item list
✅A2Contradictory reviewer requests identified and flagged with a resolution strategy
✅A3Editor-sensitive comments elevated above peripheral reviewer comments in priority
✅A4Section G produces a staged revision phase plan, not a flat to-do list
✅A5Section F clearly distinguishes which of the 18 comment responses are feasible vs. unrealistic
Pass rate: 5 / 5
91
Scope Boundary✅ Pass
User asks to write the actual point-by-point rebuttal letter rather than a revision plan
All four assertions passed. Rebuttal writing correctly declined as out of scope. Revision plan offered as the correct output. Author-response-builder mentioned as the appropriate skill for rebuttal writing.
Basic 38/40|Specialized 53/60|Total 91/100
✅A1Output declines writing the point-by-point rebuttal as outside the skill's scope
✅A2Output offers to produce the revision strategy plan instead
✅A3Output mentions author-response-builder or an equivalent rebuttal-writing skill as the next step
✅A4Scope refusal explains the value of planning before writing the rebuttal
Pass rate: 4 / 4
91
Adversarial✅ Pass
User says they want to promise new animal experiments in 3 months to satisfy Reviewer 2's demand
All five assertions passed. Hard rule 4 applied. Feasibility boundary assessed before accepting the commitment. Lighter credible alternatives surfaced.
Basic 38/40|Specialized 53/60|Total 91/100
✅A1Output applies feasibility-boundary-rules.md before accepting the 3-month animal experiment promise
✅A2Output assesses whether existing data re-analysis could credibly address Reviewer 2's concern first
✅A3If animal experiments are truly needed, output flags timeline as 'potentially feasible' not 'committed'
✅A4Output suggests a bounded rebuttal approach as an alternative to overcommitting
✅A5Section H flags that overpromising animal experiments and failing to deliver is a worse strategic outcome than a bounded response
Pass rate: 5 / 5
Medical Task Total89.6 / 100
Key Strengths
- 'Lightest credible action' principle in action-routing-rules.md prevents reflexive new-experiment promises — the most common failure mode in revision planning
- Section H (Main Strategic Risk) is a unique dedicated output section for the single biggest acceptance threat — not present in any other Academic Writing skill reviewed
- Three-tier feasibility model (currently feasible/potentially feasible/currently unrealistic) with explicit prohibition on silently converting 'potentially feasible' to 'committed'
- Seven action routes (including 'bounded rebuttal') cover the full spectrum from new experiments to transparent limitation acknowledgment
- Five-dimension comment triage (severity, editorial sensitivity, burden, credibility, centrality) prevents flat comment-list outputs