Skills warpdotdev common-skills
research
- Use it when:
- An answer means reading many files, long logs or a big diff: how X works, where Y is used, why Z fails.
- It does:
- Briefs one local, read-only subagent, or several in parallel for independent sub-questions, on the exact question, where to look and what to return.
- You get:
- A direct answer, the key file paths and symbols as evidence, and any caveats, without the noise in the main context.
- Skip it when:
- You'd read two or three known files, run one grep, or need the raw files to edit next.
- Models:
- Names gpt-6-luna-medium for simple searches and gpt-6-luna-xhigh for involved ones.
TL;DR
Context is the main agent's scarcest resource, and reading twenty files to find the three that matter fills it with noise. This skill hands the digging to a subagent and keeps only the conclusion. The test is whether the noise avoided outweighs a subagent's latency and tokens: good fits are a failing test's root cause, how a class is used across a repo, or a summary of a 4,000-line PR. Subagents always run locally, even under a cloud agent, and use a search model rather than the parent's. One is the default; several run in parallel only when the question splits into independent parts. The brief states the question, where to look, that the work is read-only, and that it should return a distilled answer with file:line evidence rather than a transcript. A bloated report gets a follow-up to the same subagent. Afterwards the agent works from the result and reads files itself only when it needs to edit them.
Summarizes the skill at 8792394.
npx skills add https://github.com/warpdotdev/common-skills --skill research skills.sh pageSource on GitHubAll skills from warpdotdev/common-skills
- Total Downloads
- 22,000 on skills.sh, Nov 9, 2026
- Repo stars
- 678 warpdotdev/common-skills
- First seen
- Jun 18, 2026 on skills.sh
Weekly values
| Week to | Installs |
|---|---|
| Sep 21 | 644 |
| Sep 28 | 586 |
| Oct 5 | 924 |
| Oct 12 | 498 |
| Oct 19 | 365 |
| Oct 26 | 362 |
| Nov 2 | 113 |
| Nov 9 | 80 |
| Category |
Build agents, orchestrate workers and give agents memory. Everything for building software: planning, coding, debugging, design, frameworks, cloud, agents and skills. |
|---|---|
| Type |
A multi-step process the agent follows: interview, debug, run a TDD loop, review. |
| Effect |
Adds no side effects of its own. It shapes how the agent thinks, writes or reviews. Guidelines and routers are always effect:none, even when the task they support edits code. |
| Needs |
Relies on the agent being able to spawn subagents. |
| Publisher |
Everyone else. Unverified vendor claims default to community. |
| Topics |
SKILL.md
No copy of its SKILL.md yet.