Verified Agent Skill record

Anti Defensive Writing En

A writing methodology that treats an academic paper as a press conference rather than a project summary: identify the single most publishable strength of the work and build the most favorable, complete, persuasive narrative around it, with rules for narrative, language, experiments, structure, and handling unfavorable results across the paper lifecycle.

Tasks

Revise an abstract so the first five sentences state both the problem and the contribution without implementation detail or limitations
Restructure a paper narrative around its strongest publishable strength, reordering contributions if the original story is unsupported by evidence
Apply the default decision rules to unfavorable results: delete, narrow claims, switch evaluation dimension, or explain as trade-off before stating plainly
Remove banned self-weakening phrases (e.g., "unfortunately", "merely", "only achieves") and process chronology from the manuscript
Audit experiments so each one proves the core method, the key mechanism, scenario relevance, or rules out an alternative explanation, deleting or redesigning those that do not
Run the pre-submission checklist to verify no sentence expands reviewer attack surface

Inputs

Draft manuscript sections (abstract, introduction, conclusion, experiments)
Experiment results and evaluation data
Reviewer comments or rebuttal material (per the skill's stated triggers)

Outputs

Revised manuscript text organized around the strongest publishable strength
A restructured experiment section where each experiment carries an explicit argumentative duty
An edited abstract and conclusion following the opening and reinforcement rules

Limitations and checks

The source describes writing principles only; it contains no executable tooling, commands, or automated checks
No platform support, inputs, or outputs beyond writing guidance are documented
The guide does not define verification of factual claims; it governs framing and presentation only
No banned self-weakening phrases ("unfortunately", "merely", "only achieves", "still lags far behind", "limited improvement", "severely insufficient") remain in the text
Each experiment can be tied to one of the four stated duties: prove the method works, prove the mechanism, prove scenario relevance, or rule out an alternative explanation
The abstract's first five sentences contain both the problem and the contribution, with no limitations discussed before the contribution is established
The conclusion contains no fresh self-negation or newly expanded limitations
No narrative is built on a metric the paper does not win; claims are limited to what the evidence supports