research-papers
Find scientific literature relevant to the algorithmic core of a problem before design. Use when the problem has a scientific or algorithmic core (scheduling, matching, probabilistic reasoning, parsing theory, concurrency protocols, ML) or the user names a paper. Runs start-research, which dispatches the workbench:research-paper-scout agent; cites DOI or arXiv id and separates peer-reviewed from preprint.
- License 许可
- MIT
- 最近更新
- 2026-10-26
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- research-papers
- description
- Find scientific literature relevant to the algorithmic core of a problem before design. Use when the problem has a scientific or algorithmic core (scheduling, matching, probabilistic reasoning, parsing theory, concurrency protocols, ML) or the user names a paper. Runs start-research, which dispatches the workbench:research-paper-scout agent; cites DOI or arXiv id and separates peer-reviewed from preprint.
03 SKILL.md 正文
research-papers
Finds the literature behind the algorithmic core of the problem, with peer-reviewed work told apart from preprints.
When to run it
Run when:
- the problem has a scientific or algorithmic core (scheduling, matching, probabilistic reasoning, parsing theory, concurrency protocols, ML);
- the user names a paper.
Skip when:
- CRUD, infra plumbing, UI, documentation.
Procedure
- Get the brief path:
bash ${CLAUDE_PLUGIN_ROOT}/skills/start-research/scripts/doc_path.sh research "<topic>-papers". - Invoke
start-researchwith lenses [paper-scout] (it dispatches theworkbench:research-paper-scoutagent) and the topic, writing the brief to that path. Model:sonnet. - Check each finding carries a DOI or arXiv id, year, venue, peer-reviewed or
preprint, and an
applicabilityline. Drop findings without them. - Distil into the Plan inputs below; every adopted idea names its paper.
Output: Plan inputs
Return the four ## Plan inputs lists (reuse, do-not-build, fix-as-ac, ideas) exactly as
defined in phase 1 of ${CLAUDE_PLUGIN_ROOT}/skills/start-development/SKILL.md; do not restate
them.
Rules
- Read-only: research only, no edits to the repos in scope.
- Every claim cited with a source URL, file:line, or DOI / arXiv id.
- Model sonnet unless stated; set the model explicitly on every agent call.
04 同仓库更多 Skills
claudy
d claudy
gamma
Literal line one. Literal line two.
beta
Quoted: description with a colon.
alpha
Folded description that spans two lines and ends here.
worktree-merge
Rebases one subagent's worktree onto the base branch, optionally verifies the rebased tree, fast-forwards it into the worktree that has the base checked out (an epic integration tree or the main checkout), verifies every commit landed, and only then removes the worktree and its branch. Keeps the worktree intact on any conflict, dirty tree, or failed check. Use after a subagent returns work created with worktree-create.
worktree-create
Creates one isolated git worktree for one subagent, at <workbench>/.claude/worktrees/<name>, branched from main with the newest origin/main and local main merged in. Refuses a name already in use so two agents can never share a tree. Use once per implementation or review subagent before dispatching it; pair with worktree-merge to land the work and clean up.