Convert mature research notes into draft Codex skills and supporting tool scaffolds. Use when a research thread has repeated stable behavior, enough evidence to justify reuse, and needs a reviewed draft before promotion into .claude/skills.
Scanned 9/10/2026
Install to Claude Code
npx -y skills add nexuslinkproductions/yuri-os --skill research-artifact-factory --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Research Artifact Factory?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/nexuslinkproductions-research-artifact-factory-4266815f)More formats (shields.io, HTML) on the badges page.
---
name: research-artifact-factory
description: Convert mature research notes into draft Codex skills and supporting tool scaffolds. Use when a research thread has repeated stable behavior, enough evidence to justify reuse, and needs a reviewed draft before promotion into .claude/skills.
triggers:
- "convert this research into a skill"
- "promote research to skill"
- "artifact-candidate"
scope: harness
invocation: ability
---
# Research Artifact Factory
Turn durable research into reusable automation, but keep the review gate intact.
## Use This Skill When
- a research thread keeps repeating the same procedure
- the behavior is stable enough to codify
- deterministic code would reduce errors or repetition
- the result should become a skill, a tool, or both
## Required Inputs
- `02_AREAS/research-intake/ARTIFACT-CANDIDATE.md`
- supporting research notes in `RESEARCH/` or `02_AREAS/research-intake/`
- an explicit artifact name
- an explicit kind: `skill`, `tool`, or `both`
## Creation Rule
Draft first. Publish later.
Generate into `02_AREAS/skills/drafts/<slug>/` before anything touches `.claude/skills/`.
## Workflow
1. Review the candidate note.
2. Confirm the evidence is stable and repeated.
3. Separate language-only guidance from code that should become a tool.
4. Draft the skill with one reusable job.
5. Draft the tool scaffold only when the workflow needs deterministic execution.
6. Record the evidence trail in `manifest.json`.
7. Do not promote without review.
## Draft Outputs
- `SKILL.md` with the reusable behavior, guardrails, and evidence trail
- `manifest.json` with candidate metadata and source note references
- `scripts/<tool>.js` when a deterministic command surface is needed
## Guardrails
- Never turn weak or contradictory research into a published skill.
- Never bundle unrelated behaviors into one artifact.
- Never skip the review queue.
- Keep tool scaffolds minimal and deterministic.
## Generator
Use the factory script:
```bash
node _SYSTEM/research-skill-factory/generate.js
```
Add `--watch` to keep drafts up to date while research changes.
## Session Notes
### 2026-04-27
- session: 6m | peak ctx: 53% | compacts: 0
- tools: Read×27, Bash×8, Write×2, mcp×1
- corrections: none
- errors: none
### 2026-04-26
- session: 7m | peak ctx: 0% | compacts: 0
- tools: Bash×15, Read×9, Write×4, Agent×1, ToolSearch×1, ExitPlanMode×1, Edit×1
- corrections: none
- errors: none
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!