ValentinNikolaev/agent-skills / agent-plugins/skills/editorial-writing
editorial-writing
VerifiedDevelop standalone posts or articles from an idea or source through evidence, structure, drafting, and bounded review. Not discussion replies, private correspondence, profile copy, or wording-only edits.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.5 KB |
| agents/openai.yaml | config | 274 B |
| references/article-development.md | markdown | 3.9 KB |
references/article-development.md
View on GitHub# Article development
Use for a new article or a substantive rewrite. Scale to the actual change; a typo does not require reconstructing the whole process.
## Decide whether the concept earns article form
Before prose, identify a plausible reader and reading occasion, the question or intellectual purpose, what changes after reading, and evidence that can earn that gain. Explain either how the reasoning extends beyond a worked example or why a deliberately bounded finding matters. Do not manufacture a universal lesson from one lab result or insist that every essay solve a practical problem.
Choose an evidence-backed decision:
- `pass`: the proposed gain and supporting path are credible;
- `reframe`: the reader, question, promise, or evidence plan needs correction before prose;
- `reclassify`: the material better supports a note, explanation, reference, or research record;
- `stop`: no worthwhile supported article is currently available;
- `blocked`: an essential source or user decision is unavailable.
Preserve a user-defined audience or purpose. Propose a material scope change explicitly rather than making a weaker promise after writing and calling the original objective met. Reopen this decision when author feedback rejects the article's value, even if technical checks passed.
## Build a concrete skeleton
After the value decision passes, map sections to reader questions, intended gains, key claims, evidence, examples, visual decisions, and transitions. Check prerequisites and whether the ending earns the promised gain. Mark evidence gaps; never prescribe an experiment's result. These are working planning fields, not compulsory manuscript headings.
For an existing article, label a reconstructed skeleton as reconstructed. Follow project storage and manuscript ownership conventions. Keep working notes ephemeral unless the user or project requests a retained plan.
Decide whether a diagram, chart, table, screenshot, or worked example materially improves understanding. Use source-grounded quantities and labels, meaningful alternatives for inaccessible visuals, and appropriate rendering checks. A decision that no visual helps is valid; decoration is not evidence.
## Keep one revision history
Pin the first complete prose draft as baseline and retain its existing revision count. For a requested retained plan, record each batch's concrete changes, reasons, completion evidence, and remaining issues. Count a substantive rewrite followed by assessment as one batch, including batches before first delivery. Reviewer passes on unchanged text are validation, not extra editing iterations. File relocation and a new filename do not restart the budget.
Diagnose value, evidence, and scope before polishing. Prioritize the correction that resolves the actual objection. When evidence is missing, stop prose work or propose a supported alternative within the user's scope. At the budget limit, report `not ready` if any mandatory requirement fails; continuing requires additional scope authorization.
## Assess the article a reader will receive
When independent assessment is required, first give the reviewers the exact draft, intended audience, and prerequisites. Ask what question it answers, why the reader would continue, what changes after reading, and what passages earn that result. Record this reading before revealing the full brief, evidence, or proposed repairs. Withhold historical scores and disclose any prior context reviewers already possess.
Then supply the identical brief, evidence, and material author feedback. Resolve objections against the text before quality scoring. Do not invent a use case on the article's behalf. Neither technical correctness nor reviewer agreement can waive a missing reader gain or a known unresolved author objection.
Verify applicable factual claims, links, examples, code, visuals, and destination constraints on the final version. Identify checks not performed. Separate readiness for author review from personal approval and publication authority.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |