cloudaeye

Plugin cloudaeye from marketplace cloudaeye in CloudAEye/cloudaeye-plugin. Commands read from the repo's manifest — today.

Skills (23)

• add-docs

  Have CloudAEye write docstrings for the undocumented code in an open pull request and post them on the pull request as one-click review suggestions. It writes on the PR; it never edits your working tree.

• add-tests

  Have CloudAEye write unit tests for the uncovered code in an open pull request and post them on the pull request as one-click review suggestions, new test files included. It writes on the PR; it never edits your working tree.

• ask

  Ask a free-form question about the uncommitted changes in this repo and get an answer grounded in the repository's code graph — callers, definitions and usage traces, not just the diff text. Returns an answer, not a review, and never edits code.

• assign-reviewer

  Suggest who should review an open pull request, say why, show their review queue, and suggest what they should check — from code ownership (.OWNERS.md) at minimum, and from whoever wrote the lines it changes, reviewed these files before or fixed bugs in them, balanced against each person's queue. Ta

• check-task

  Check whether the uncommitted changes in this repo actually do what the task asked. Takes a GitHub issue URL, Jira ticket IDs, a mixed reference list, or freeform task text, and returns a DONE / NOT DONE verdict with a per-requirement checklist and the gaps.

• check

  Run CloudAEye's checklist against an open pull request of this repository, given its number (#405) — docstring coverage, unit-test coverage, README freshness, dependency manifests, PR title and description quality, duplicate code, and a secret scan. Reports what each check found and posts the full r

• config

  Set how CloudAEye works in this repository — ultrareview's approval settings (which checks run, the hard limits it approves within, and its human review policy — auto-assigning reviewers, how many at most, and your instructions), the review's defaults and which LLM reviews the code, the control-flow

• control-flow

  Draw what the uncommitted changes in this repo did to the flow of control, as a mermaid sequence diagram with every message marked added, removed or retargeted. The arrows are resolved from the repository's code graph rather than inferred from the diff. Reports; it never edits code.

• create-issue

  File findings from a CloudAEye review as tracking issues — a Jira ticket or a GitHub issue, with an optional assignee. Takes the finding numbers the review printed. It creates real tickets; it never edits code.

• create-owners

  Write .OWNERS.md files for this repository from its git history — who writes each part of the code, with their share and the lead contributor — at the root (or a folder you name) and in every sub-folder down to three levels whose people differ from the rest. Leaves a Teams section for people to fill

• describe

  Describe the uncommitted changes in this repo — a Change Description heading and an Important Changes bullet list, in plain markdown ready to paste into a PR or a commit message. Trivial diffs short-circuit to one line.

• explain

  Explain a Jira or GitHub issue against the code it actually touches — the components involved, how they work today, and the related patterns and configuration around them. CloudAEye reads the repository's graph and vector store, which is why this answers things the issue text alone cannot.

• help

  Show the CloudAEye command catalog or a manual for one command, including arguments, options, examples, and limitations. Documentation only; never runs the command.

• implement

  Plan a fix for findings a CloudAEye review already produced. You say which ones in your own words — "fix the 2nd point, and use a context manager" — and the server returns a fix plan grounded in the repository's code graph, including the call sites and tests the change affects. It plans; it never ed

• init

  Connect the current Git repository to CloudAEye using its detected GitHub, GitLab, or Bitbucket remote.

• inspect

  Bug-hunting CloudAEye pass over the uncommitted changes in this repo, over one file or directory of them, or over an open pull request given its number — logic errors, edge cases, input validation, error handling, concurrency, dead imports. Security is left to its own pass, so this is the cheap one

• review-arch

  Review what the uncommitted changes in this repo (or a pull request) did to the architecture — which layers and components the change touched, every dependency it added or removed, layer violations, new import cycles, new calls to outside systems, breaks another repository would feel, and the questi

• review

  Full CloudAEye review — bugs and security in one pass — of the uncommitted changes in this repo, of just one file or directory of them, or of an open pull request given its number. Reports findings with file, line and severity, grouped by kind; it never edits code on its own. The widest and most exp

• risk-score

  Assess the change risk of an open pull request, or of your change before you push it — who wrote it, what it touches, and what those files and that author have been through over the last few weeks (bugs, rework, ownership, the author's pull requests, AI slop, test coverage) — and say whether it need

• security

  Security review of the uncommitted changes in this repo, of one file or directory of them, or of an open pull request given its number — OWASP-style application security plus the LLM, AI-agent and MCP surfaces, and secrets leaked on the changed lines. Reports findings with file, line and severity; i

• slop-score

  Score the uncommitted changes in this repo — or one path of them, or an open pull request — for AI-slop signals, and report how much review the change needs, with the evidence behind it. Nine weighted signals — duplicate code, hallucinated APIs, dead code, missing tests, complexity, generic naming,

• triage-prs

  Pick the pull request you should review next, across every repository your team connected to CloudAEye, and say why. Each PR waiting on you gets a priority out of 100 from its tickets' urgency, what the changed code does and the OKRs it serves, its risk, and how much else depends on it; PRs whose ne

• ultrareview

  Every CloudAEye review of a change in one run — the full review (bugs and security), the AI slop score, the change-risk verdict, the architecture review and the control-flow diagram, plus the PR checklist on a pull request and the task check when you name one. Each pass runs once, in the order that

MCP servers (1)

• cloudaeye

  connects to: ${CLOUDAEYE_URL:-https://api.cloudaeye.com/mcp}

Similar plugins