SkillHub

Skills warpdotdev

common-skills Skill collection

Common agent skills to share across repositories.

Some skills still carry Warp's own repo paths, commands or assumptions, as its README says.

Source on GitHubSort and filter them in the skills list

678 stars on GitHub, Nov 9, 2026 · 19 skills · 2 bundles

Its bundles

  1. Warp PR review

    Bundle of 2 skills
    Use it when:
    A CI job checks out a pull request and should get the agent's review as a file to publish.
    It does:
    Reviews the changed lines for correctness, security, error handling, performance, comments and tests, and checks them against spec_context.md when there is one.

    review-pr, check-impl-against-spec

    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    29.1K
    top 23%
    Weekly Downloads
    94
    bottom 19%

  2. Warp spec workflow

    Bundle of 5 skills
    Use it when:
    A feature is big, risky or ambiguous enough that an agent will build it better from written specs.
    It does:
    Decides whether specs are worth it, writes a product spec and, when warranted, a tech spec, builds the feature from them, and finally checks code and specs agree.

    spec-driven-implementation, write-product-spec, write-tech-spec, implement-specs, validate-changes-match-specs

    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.9K
    top 23%
    Weekly Downloads
    90
    bottom 18%

Its skills, most installed first

  1. review-pr

    Part of Warp PR review
    Use it when:
    A CI job has checked out a pull request and you want the agent's review as a file to publish, not comments posted directly.
    It does:
    Reviews the lines the PR changes for correctness, security, error handling, performance, comment quality and tests, and labels each finding by severity.
    type:workfloweffect:localneeds:nonecode-reviewcistack:github
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    29.1K
    top 23%
    Weekly Downloads
    94
    bottom 19%

  2. spec-driven-implementation

    Only works as part of Warp spec workflow
    Use it when:
    Starting a large, risky or ambiguous feature that an agent will build, such as one around 1k+ lines.
    It does:
    Decides whether specs are worth it, files a Linear issue if there's none, then has the agent write the product spec, the tech spec when warranted, and the code.
    type:workfloweffect:remoteneeds:mcpneeds:issue-trackerplanningrequirementsimplementationissue-trackingstack:linear
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.9K
    top 23%
    Weekly Downloads
    83
    bottom 17%

  3. write-product-spec

    Part of Warp spec workflow
    Use it when:
    You want a feature's behavior pinned down in writing before an agent builds it, as a product spec or PRD.
    It does:
    Asks you for missing context and for a Figma mock when there's UI, then writes the behavior as numbered rules that can each be tested.
    type:workfloweffect:localneeds:nonerequirementsplanning
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.9K
    top 23%
    Weekly Downloads
    90
    bottom 18%

  4. write-tech-spec

    Part of Warp spec workflow
    Use it when:
    A feature spans several modules, has real architectural trade-offs, or reviewers should see the plan before or with the code.
    It does:
    Reads the product spec and the relevant code, then plans the changes and maps each important rule of the spec to a test or check.
    type:workfloweffect:localneeds:noneplanningarchitecturemulti-agent
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.8K
    top 23%
    Weekly Downloads
    89
    bottom 18%

  5. resolve-merge-conflicts

    Use it when:
    A merge, rebase, cherry-pick or stash pop stops on conflicts.
    It does:
    Runs a bundled Python script that lists the unresolved files, then shows one file's conflict hunks with ours, base and theirs and a compact diff.
    type:workfloweffect:localneeds:nonegittoken-savingstack:git
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.8K
    top 23%
    Weekly Downloads
    102
    bottom 20%

  6. implement-specs

    Only works as part of Warp spec workflow
    Use it when:
    The product and tech specs are approved and it's time to build.
    It does:
    Checks the specs exist and are approved, plans steps against them, builds with tests, and edits the specs whenever behavior or design changes.
    type:workfloweffect:localneeds:noneimplementationverification
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.7K
    top 23%
    Weekly Downloads
    78
    bottom 15%

  7. update-skill

    Use it when:
    You're creating a skill, or reworking an existing one's frontmatter, structure or guidance.
    It does:
    Sets the rules: a kebab-case name, and a third-person description that starts with a verb and says what the skill does and when.
    type:guidelineseffect:noneneeds:noneskill-authoring
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.6K
    top 23%
    Weekly Downloads
    79
    bottom 15%

  8. diagnose-ci-failures

    Use it when:
    A pull request's CI is red and you want to know why before anyone touches the code.
    It does:
    Finds the branch's PR with gh, sorts its checks into passed, failed and running, pulls the failed logs and groups the errors.
    type:workfloweffect:noneneeds:clicidebuggingstack:githubstack:rust
    Age
    6 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.6K
    top 23%
    Weekly Downloads
    76
    bottom 14%

  9. pr-walkthrough

    Use it when:
    You have a pull request's branch checked out and want a map of what it touches before reading the diff.
    It does:
    Reads the PR, its diff, changed specs, review comments and the surrounding code, then draws a system overview plus data-flow, dependency and user-action graphs.
    type:workfloweffect:localneeds:clineeds:browsercode-reviewdiagramsstack:githubstack:d3
    Age
    5 months
    Stars
    678
    bottom 18%
    Total Downloads
    28.3K
    top 24%
    Weekly Downloads
    86
    bottom 17%

  10. council

    Use it when:
    You want second opinions on a judgment call: an architecture tradeoff, a risky fix, a rollout decision, or whether an alternative is worth pursuing.
    It does:
    Gives each subagent a model and an angle, such as Opus on correctness, GPT on testability and an open-source model as contrarian, and keeps them independent.
    type:workfloweffect:noneneeds:subagentsmulti-agentarchitectureagent-workflows
    Age
    5 months
    Stars
    678
    bottom 18%
    Total Downloads
    28K
    top 24%
    Weekly Downloads
    113
    bottom 21%

  11. check-impl-against-spec

    Only works as part of Warp PR review
    Use it when:
    An automated review with review-pr finds spec_context.md, the PR's spec context: product behavior, tech details or both.
    It does:
    Lists the spec's commitments, compares them with the diff and checked-out code, and flags only material drift: missing behavior, contradictions, significant unplanned scope, skipped validation or migration steps.
    type:workfloweffect:localneeds:nonecode-reviewverificationci
    Age
    5 months
    Stars
    678
    bottom 18%
    Total Downloads
    27.1K
    top 24%
    Weekly Downloads
    81
    bottom 16%

  12. respond-to-pr-comments-in-blocklist

    Use it when:
    You want to answer a PR's review comments and post the replies back to GitHub, not just read them or fix code.
    It does:
    Skips bot overviews and threads you already answered, then asks per comment: apply the recommended fix, explain it, acknowledge without changes, or something else.
    type:workfloweffect:remoteneeds:clicode-reviewgitstack:github
    Age
    5 months
    Stars
    678
    bottom 18%
    Total Downloads
    25.3K
    top 25%
    Weekly Downloads
    77
    bottom 14%

  13. validate-changes-match-specs

    Part of Warp spec workflow
    Use it when:
    A spec-driven change is up for review or nearly done, and you want to know where the code and its specs disagree.
    It does:
    Finds the specs the branch adds or changes, extracts their commitments, reads the code and tests, and lists each material mismatch with its sources and a recommended resolution.
    type:workfloweffect:localneeds:noneverificationrequirementscode-reviewsecurity
    Age
    5 months
    Stars
    678
    bottom 18%
    Total Downloads
    25.3K
    top 25%
    Weekly Downloads
    77
    bottom 14%

  14. cross-critique

    Use it when:
    Independent subagents produced divergent proposals on an architecture choice, a review disagreement or competing root-cause theories.
    It does:
    Sends each author the other proposals, labelled neutrally, and asks for pros and cons of each, whether they'd revise their own, and a ranking with confidence.
    type:workfloweffect:noneneeds:subagentsmulti-agentarchitectureagent-workflows
    Age
    4 months
    Stars
    678
    bottom 18%
    Total Downloads
    22.1K
    top 27%
    Weekly Downloads
    93
    bottom 19%

  15. 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.
    type:workfloweffect:noneneeds:subagentsmulti-agenttoken-savingagent-workflowsresearch
    Age
    4 months
    Stars
    678
    bottom 18%
    Total Downloads
    22K
    top 28%
    Weekly Downloads
    80
    bottom 15%

  16. saga

    Use it when:
    A medium-to-large feature should be built mostly autonomously, or you're resuming a saga from its directory.
    It does:
    Asks multiple-choice questions until no requirement is open, splits the work into milestones and tasks with validation criteria, then delegates tasks to parallel workers.
    type:workfloweffect:localneeds:subagentsmulti-agentplanningimplementationverificationagent-workflows
    Age
    4 months
    Stars
    678
    bottom 18%
    Total Downloads
    20.8K
    top 28%
    Weekly Downloads
    77
    bottom 14%

  17. readout

    Use it when:
    You say "write this up" mid-investigation, or want a readable explainer such as how a server processes webhook events.
    It does:
    Asks two to four scoping questions with concrete options, then hands a short brief to one local child agent that mines the conversation or researches the code.
    type:workfloweffect:localneeds:subagentsdocs-writingresearch
    Age
    4 months
    Stars
    678
    bottom 18%
    Total Downloads
    17.7K
    top 30%
    Weekly Downloads
    78
    bottom 15%

  18. skill-doctor

    Use it when:
    You want to know which of your installed skills actually help, judged from your real conversation history.
    It does:
    Samples up to 12 conversations from the last 45 days by default, scores each for efficiency, code quality, procedure compliance and verbosity, and grades the setup.
    type:workfloweffect:localneeds:noneskill-authoringself-improvementagent-workflows
    Age
    2 months
    Stars
    678
    bottom 18%
    Total Downloads
    8.9K
    bottom 45%
    Weekly Downloads
    209
    bottom 40%

  19. write-pr-description

    Use it when:
    The agent opens a PR, fills a repo's PR template, or refreshes a description that no longer matches the branch.
    It does:
    Reads commit bodies before the diff, checks their claims against the code, follows the repo's template, then cuts every paragraph that helps no review decision.
    type:workfloweffect:noneneeds:nonegitcode-reviewwritingverification
    Age
    2 months
    Stars
    678
    bottom 18%
    Total Downloads
    6.2K
    bottom 28%
    Weekly Downloads
    88
    bottom 18%