Workbench research
Skill Fearvox/multica-ultimate-workbench/skills/workbench-research
Public multi-agent workbench skill pack for SDD, review gates, context discipline, VM lanes, Sanity context, and autonomous agent coordination.
npx -y skills add Fearvox/multica-ultimate-workbench --skill workbench-researchAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 23 stars23 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Evidence-first research, architecture discovery, vendor documentation review, benchmark reading, and unknown-system analysis.
SKILL.md
1.5 KB, 294 tokens by cl100k_base, as published. Nobody here has run it
Workbench Research
Use this skill for research, architecture discovery, vendor docs, benchmark reading, and unknown-system analysis.
Research Standard
- Read the real system first when local state exists.
- Prefer primary sources, live commands, local files, and official docs.
- Separate confirmed facts, source claims, and inference.
- Treat current software behavior, API shape, prices, versions, and docs as drift-prone.
- Cite file paths, command names, issue/comment IDs, or source URLs when they support a claim.
- Use targeted reads first: indexes,
rg,jq, table summaries, and exact IDs before full documents or full histories. - Summarize large sources before handing them to another agent.
Research Output
Return:
QUESTION: what was investigated.CONFIRMED: facts with evidence.UNCERTAIN: assumptions or unresolved gaps.OPTIONS: viable paths with tradeoffs.RECOMMENDATION: the smallest next action.SOURCES: local paths, commands, issue IDs, or URLs used.
Research Boundaries
- Do not present speculation as fact.
- Do not copy secrets or private tokens into reports.
- Do not rewrite working architecture while researching it.
- If source material conflicts, explain the conflict and prefer live runtime evidence.
- Stop if the next read would be broad and low-yield; report the smaller evidence query needed instead.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.