Evidence-Bound Press-Conference Revision skill
A Codex skill that cuts defensive wording in an academic paper without raising claims above the evidence.
- 282 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The skill only edits text and works with manuscript files with the author's permission.
lensback940701/evidence-bound-press-conference-revision-skill
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/evidence-bound-press-conference-revisionDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill only edits text and works with manuscript files with the author's permission.
This is third-party code. Review the repository files before installing.
What it does
Evidence-Bound Press-Conference Revision, the `press-conference-revision-evidence-bound` skill, reviews a manuscript whose contribution is buried under repeated caveats, self-defence and work-log narration. The aim is to remove rhetorical defensiveness while keeping scholarly caution: scope conditions, source status, rival explanations and method limits. It works in two modes. Diagnosis only reads the text and produces a candidate register, KEEP decisions and a section map. Revision starts after the author gives explicit authority and yields a tracked version, a clean text and a claim regression report. Before either mode the skill drafts a contribution-and-scope contract.
Who it is for. Authors and editors of academic manuscripts, especially qualitative and historical research, where a small wording change shifts the status of a claim.
Good fit when
- A manuscript went through several review rounds and turned defensive.
- The paper's contribution appears only after several paragraphs of caveats.
- A co-author wants stronger prose, but the evidence ceiling must not move.
- AI-assisted revision needs an auditable protocol instead of unrestricted rewriting.
Not a fit when
- You need a first draft of a paper.
- You are looking for literature, citations or fact checking.
- Weak evidence needs to be hidden behind polished wording.
Example request
Run a read-only diagnostic of the manuscript: draft the contribution contract and show the defensive-writing register with KEEP decisions.Limitations
The README and skill target Codex. The `scan_defensive_cues.py` script pre-screens only DOCX files for cues. The skill does not replace author judgement: doubtful passages stay as queries. The repository carries an example of guardrails for one specific urban geography manuscript. General use does not depend on it, so your own paper gets a comparable file. The last commit dates from August 2026.
How to disable. Delete the `~/.codex/skills/press-conference-revision-evidence-bound` directory and restart Codex. Roll back manuscript edits using the tracked-changes version.
FAQ
Does the skill strengthen the claims of the paper?
No. It never raises certainty, causality, novelty or generality above what the evidence supports: "may shape" does not become "determines".
Does the diagnosis change the text?
No. Diagnosis only reads the manuscript; the skill must not edit without the author's explicit authority.
What comes out of a revision?
A tracked manuscript, a clean version, a change summary by pattern, a claim regression report, a list of deliberate non-edits and questions for the author.
What is the script for?
`scripts/scan_defensive_cues.py` pre-screens passages in DOCX files. Each passage is decided by reading it in context, not by word frequency.
Which agents does the skill run on?
The README and installation target Codex: the `~/.codex/skills` directory and invocation through `$press-conference-revision-evidence-bound`. The repository also holds `agents/openai.yaml`.
Discussion
No questions or reviews yet.
Related
SkillsA skill that spots defensive academic prose: pre-emptive replies to reviewers, stacked caveats and excuses for weak results.
A skill that stress-tests a research paper before submission: mocks a review, gives a verdict, helps revise and re-verify
A skill for editing academic writing and grant proposals: strips AI-writing tells, keeps the scholarly voice and ties claims to evidence
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
Research Weave turns a pile of papers into a reading queue, an evidence map and a sharper research question
A skill that strips defensive caveats, apologies and excess hedging from text and makes claims direct while keeping needed limits
README in short
The skill separates defensive rhetoric from needed caution and requires a contribution contract, an audit register and a regression check.