Skip to content
Back to search

roborev

MarketplaceWorks with Claude CodeGitHub

Marketplace for roborev agent skills.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026. Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/kenn-io/roborev with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/kenn-io/roborev'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/kenn-io-roborev/badge.svg)](/listings/kenn-io-roborev)

HTML

<a href="/listings/kenn-io-roborev"><img src="/api/listings/kenn-io-roborev/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

cd your-repo
roborev init          # Install post-commit hook
git commit -m "..."   # Reviews happen automatically
roborev tui           # View reviews in interactive UI
roborev ui            # Open reviews in the native browser UI

If roborev is managed by a version manager, `roborev init` and `roborev agent hook install` try to install hooks with the stable shim/symlink. You can also choose the exact binary path with `roborev init --binary ~/.local/share/mise/shims/roborev`, or `roborev agent hook install --binary ~/.local/share/mise/shims/roborev`.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 20 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "plugins": [
  3. {
  4. "version": "0.75.0",
  5. "name": "roborev",
  6. "description": "Agent skills for roborev code review, fix, refine, and response workflows.",
  7. "source": "./",
  8. "author": {
  9. "name": "roborev",
  10. "url": "https://github.com/roborev-dev"
  11. }
  12. }
  13. ],
  14. "name": "roborev",
  15. "description": "Marketplace for roborev agent skills.",
  16. "owner": {
  17. "name": "roborev",
  18. "url": "https://github.com/roborev-dev"
  19. }
  20. }
README.md · README · 644 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. ---
  2. last_edited: 2026-11-08
  3. ---
  4. <p align="center">
  5. <picture>
  6. <source media="(prefers-color-scheme: dark)" srcset="https://roborev.io/docs/assets/static/logo-with-text-dark-bg.svg">
  7. <img alt="roborev" src="https://roborev.io/docs/assets/static/logo-with-text-light.svg">
  8. </picture>
  9. </p>
  10. <p align="center">
  11. <a href="https://go.dev/"><img src="https://img.shields.io/badge/Go-1.27.4+-00ADD8?logo=go&logoColor=white" alt="Go 1.27.4+"></a>
  12. <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-yellow.svg" alt="License: MIT"></a>
  13. <a href="https://roborev.io"><img src="https://img.shields.io/badge/Docs-roborev.io-blue" alt="Docs"></a>
  14. </p>
  15. <p align="center">
  16. <a href="https://roborev.io/docs/">Documentation</a> ·
  17. <a href="https://roborev.io/docs/quickstart/">Quick Start</a> ·
  18. <a href="https://roborev.io/docs/installation/">Installation</a>
  19. </p>
  20. **Continuous code review for AI coding agents.**
  21. roborev runs in the background, reviews every commit as agents write code,
  22. and surfaces issues in seconds -- before they compound. Pull code reviews
  23. into your agentic loop while context is fresh.
  24. <p align="center">
  25. <img src="https://roborev.io/docs/assets/generated/web-ui.png" alt="Roborev browser application">
  26. </p>
  27. ## How It Works
  28. 1. Run `roborev init` to install a post-commit hook
  29. 2. Every commit triggers a background review -- agents write, roborev reads
  30. 3. View findings in the TUI, feed them to your agent, or let `roborev fix` handle it
  31. ### Automation, two layers
  32. ![How roborev works](https://roborev.io/docs/assets/static/how-it-works.svg)
  33. - **Post-commit reviews** - a git hook reviews every commit in the background (any agent).
  34. - **Agent hook** - watches supported coding-agent sessions and brings open
  35. roborev findings back into the active workflow.
  36. ```bash
  37. roborev init # layer 1: per-commit reviews
  38. ```
  39. Not in the latest release yet
  40. ```bash
  41. roborev agent setup # layer 2: wire Claude Code and Codex
  42. ```
  43. For other supported agents, run `roborev agent hook install --agent all`.
  44. Before you ship, run the `/roborev-refine` skill: it re-reviews and fixes your
  45. whole branch until every review passes, catching bugs before the PR.
  46. New here? Run `roborev quickstart` and point your agent at it.
  47. Not in the latest release yet
  48. Add `--mcp` to `roborev agent setup` to register MCP. If Codex asks you to
  49. review hooks, run `/hooks` in Codex, then `roborev agent setup --check`.
  50. ## Quick Start
  51. ```bash
  52. cd your-repo
  53. roborev init # Install post-commit hook
  54. git commit -m "..." # Reviews happen automatically
  55. roborev tui # View reviews in interactive UI
  56. roborev ui # Open reviews in the native browser UI
  57. ```
  58. If roborev is managed by a version manager, `roborev init` and
  59. `roborev agent hook install` try to install hooks with the stable shim/symlink.
  60. You can also choose the exact binary path with
  61. `roborev init --binary ~/.local/share/mise/shims/roborev`, or
  62. `roborev agent hook install --binary ~/.local/share/mise/shims/roborev`.
  63. ![roborev review](https://roborev.io/docs/assets/generated/tui-review.svg)
  64. ## What you can do
  65. - **Background Reviews** - Every commit is reviewed automatically via
  66. git hooks. No remote review workflow required.
  67. - **Auto-Fix** - `roborev fix` feeds review findings to an agent that
  68. applies fixes and commits. `roborev refine` iterates until reviews pass.
  69. - **Agent Hook** - Optional hooks for Claude Code, Codex, Copilot CLI, Cursor,
  70. Factory Droid, Gemini CLI, Hermes, and Qwen bring open findings back into the
  71. active agent session.
  72. - **Custom Reviews** - Write your own rubrics and run them as review panels,
  73. for code or for documents such as legal drafts. `/roborev-setup-review`
  74. interviews you and writes the setup. Reviewers can read PDFs that a commit
  75. adds or changes, and strict review types get only read tools plus the MCP
  76. tools you grant.
  77. - **Code Analysis** - Built-in analysis types (duplication, complexity,
  78. refactoring, test fixtures, dead code, security) that agents can fix
  79. automatically.
  80. - **Multi-Agent** - Works with Codex, Claude Code, Gemini, Copilot,
  81. OpenCode, Cursor, Kiro, Kilo, Droid, Pi, and Muse Code.
  82. - **Runs Locally** - No hosted service or additional infrastructure.
  83. Reviews are orchestrated on your machine using the coding agents
  84. you already have configured.
  85. - **Interactive TUI** - Real-time review queue with vim-style navigation.
  86. Large terminals show the queue and selected review in a split-screen layout.
  87. - **Native Browser UI** - Browse, filter, inspect, comment on, close, cancel,
  88. and rerun reviews from the application embedded in the local daemon, then
  89. explore cost, latency, reliability, and outcomes in Analytics.
  90. - **Review Verification** - `roborev job compact` verifies findings against
  91. current code, filters false positives, and consolidates related issues
  92. into a single review.
  93. - **Review History Search** - Search completed reviews across repositories by
  94. keyword or meaning. Lexical search stays local and needs no embedding
  95. provider; semantic and hybrid modes are optional.
  96. - **Extensible Hooks** - Run shell commands on review events. Built-in
  97. [beads](https://github.com/steveyegge/beads) and [kata](https://github.com/kenn-io/kata)
  98. integrations create trackable issues from review failures automatically.
  99. ## The Agentic Fix Loop
  100. When reviews find issues, copy-and-paste the reviews into your
  101. interactive agent sessions, or invoke `/roborev-fix` in Claude Code
  102. or `$roborev-fix` in Codex. You can also address open reviews on the
  103. command line non-interactively with `roborev fix`.
  104. `roborev fix` shows the review findings to an agent, which applies
  105. changes and commits. The new commit gets reviewed automatically,
  106. closing the loop.
  107. `roborev agent hook install` auto-detects installed Claude Code, Codex, Copilot
  108. CLI, Cursor, Factory Droid, Gemini CLI, Hermes, Qwen, Grok Build, and ZCode
  109. harnesses
  110. and adds optional hooks after configured turn, commit, or failed-review
  111. thresholds are met.
  112. Reminders name exact review IDs and invoke the bundled `roborev-fix` skill.
  113. <!-- release-commands: expected-failure -->
  114. Never run `roborev fix --open`.
  115. Hook installation updates bundled skills for every supported profile. Hermes delivers queued
  116. post-tool reminders at `Stop`; Cursor records the same events but emits no control
  117. response.
  118. Installed hooks post events to the regular roborev daemon. That daemon evaluates
  119. the reminders and persists session counters and delivered review IDs in
  120. `${ROBOREV_DATA_DIR:-~/.roborev}/agent-hook/state.json`. Hook callbacks fail open
  121. when the daemon is unavailable, so they do not block the coding agent.
  122. For fully automated iteration (advanced feature), use `refine`:
  123. ```bash
  124. roborev refine # Fix, re-review, repeat until passing
  125. roborev refine --plan # Plan each fix before editing
  126. roborev refine --plan-only # Inspect plans for existing failed reviews
  127. roborev refine --skip-trivial-rereview # Do not wait on docs/test-only fix re-reviews
  128. ```
  129. `refine` runs in an isolated worktree and loops: fix findings, wait for
  130. re-review, fix again, until all reviews pass or `--max-iterations` is hit.
  131. ## Rebutting a Finding
  132. Use a stored reviewer session to re-evaluate one finding with your explanation:
  133. ```bash
  134. roborev job comment 42 --rebuttal --finding "Missing null check in handler." \
  135. -m "The caller rejects null input before invoking the handler."
  136. ```
  137. The assessment is saved as a comment. Add `--close` to close the whole review
  138. once it is recorded. See
  139. [Responding to Reviews](https://roborev.io/guides/responding-to-reviews/) for
  140. supported agents and recovery.
  141. ## Code Analysis
  142. Run targeted analysis across your codebase and optionally auto-fix:
  143. ```bash
  144. roborev analyze duplication ./... # Find duplication
  145. roborev analyze refactor --fix *.go # Suggest and apply refactors
  146. roborev analyze complexity --wait main.go # Analyze and show results
  147. roborev analyze test-fixtures *_test.go # Find test helper opportunities
  148. roborev analyze security ./... # Find security risks in existing code
  149. ```
  150. Available types: `test-fixtures`, `duplication`, `refactor`, `complexity`,
  151. `api-design`, `dead-code`, `architecture`, `security`.
  152. Analysis jobs appear in the review queue. Use `roborev fix` to apply
  153. open findings later, target a specific job with `roborev fix <id>`, or
  154. pass `--fix` to apply immediately.
  155. Analysis types can pin their own agent settings in config:
  156. ```toml
  157. [analyze.refactor]
  158. agent = "claude-code"
  159. model = "sonnet"
  160. reasoning = "fast"
  161. ```
  162. ## Installation
  163. **macOS ([Homebrew](https://formulae.brew.sh/formula/roborev)):**
  164. ```bash
  165. brew install roborev
  166. ```
  167. Homebrew manages updates with `brew upgrade roborev`.
  168. **Linux (install script):**
  169. ```bash
  170. curl -fsSL https://roborev.io/install.sh | bash
  171. ```
  172. Linux users can also install a [DEB or RPM package](https://roborev.io/docs/installation/#linux-packages-deb-and-rpm).
  173. **Windows (PowerShell):**
  174. ```powershell
  175. powershell -ExecutionPolicy ByPass -c "irm https://roborev.io/install.ps1 | iex"
  176. ```
  177. **With Go:**
  178. ```bash
  179. go install go.kenn.io/roborev/cmd/roborev@latest
  180. ```
  181. The Go module source archive does not include generated browser assets. This
  182. installation path provides the CLI and terminal UI; install a release package
  183. or build with `make install` for the embedded browser application.
  184. ## Developer Setup
  185. This repo uses [`prek`](https://prek.j178.dev/) for local Git hooks. Commits
  186. run only the Renovate config validator and jactionlint, and only when
  187. `renovate.json` or a workflow file changes. The whole-repo checks run on push:
  188. a Git-test isolation guard, `make lint-ci` for non-mutating Go lint,
  189. `make markdown-ci` for Zensical Markdown formatting, and the web check and test
  190. scripts when web files change. If you installed the hooks before, rerun
  191. `prek install` to add the pre-push hook.
  192. ```bash
  193. brew install prek uv # or use your preferred install method
  194. mise use --global npm:renovate@latest
  195. prek install # install the pre-commit and pre-push hooks
  196. prek run --all-files --hook-stage pre-push # run the push checks manually
  197. ```
  198. Use `make lint` when you explicitly want golangci-lint to apply fixes. Use
  199. `make markdown` to wrap prose in published Zensical pages at 80 columns while
  200. leaving Markdown tables unchanged. Use `make check-renovate-config` to validate
  201. `renovate.json` directly.
  202. ## Commands
  203. | Command | Description |
  204. |---------|-------------|
  205. | `roborev init` | Initialize roborev in current repo |
  206. | `roborev tui` | Interactive terminal UI |
  207. | `roborev daemon status` | Show daemon, browser UI, and queue status |
  208. | `roborev review <sha>` | Queue a commit for review |
  209. | `roborev review --branch` | Review all commits on current branch |
  210. | `roborev review --dirty` | Review uncommitted changes |
  211. | `roborev fix` | Fix open reviews (or specify job IDs) |
  212. | `roborev refine` | Auto-fix loop: fix, re-review, repeat |
  213. | `roborev analyze <type>` | Run code analysis with optional auto-fix |
  214. | `roborev agent hook install` | Install hooks for detected coding agents |
  215. | `roborev agent hook install --agent all` | Install all nine supported integrations |
  216. | `roborev agent hook snooze` | Silence Agent Hook reminders in the current worktree and branch |
  217. | `roborev agent hook snooze off` | Resume Agent Hook reminders in the current worktree and branch |
  218. | `roborev job compact` | Verify and consolidate open review findings |
  219. | `roborev job show [sha]` | Display review for commit |
  220. | `roborev job search <query>...` | Search completed review history across repositories |
  221. | `roborev export reviews` | Export completed reviews as JSON |
  222. | `roborev export ci-metrics` | Export finalized CI panel metrics as JSON |
  223. | `roborev export ci-costs` | Export job-level CI costs as JSON |
  224. | `roborev run "<task>"` | Execute a task with an AI agent |
  225. | `roborev job close <id>` | Close a review |
  226. | `roborev agent skills install` | Install bundled skills for supported agent profiles |
  227. See [full command reference](https://roborev.io/docs/commands/) for all options.
  228. Review history search is global by default and supports repository, branch,
  229. time, verdict, and open/closed filters. Lexical search works with no provider
  230. configuration or network call. See the
  231. [search guide](https://roborev.io/docs/search/) for semantic and hybrid setup,
  232. freshness, privacy, and recovery behavior.
  233. ### Exporting review history
  234. Use `roborev export reviews` to emit completed reviews as one JSON document for
  235. local reporting or archival workflows:
  236. ```bash
  237. roborev export reviews
  238. roborev export reviews --profile metadata --since 2026-06-01 --until 2026-06-30
  239. roborev export reviews --closed-only --repo github.com/org/repo --limit 1000
  240. roborev export reviews --cursor "$NEXT_CURSOR" --until 2026-07-01
  241. ```
  242. The default `content` profile includes each review twice: `document` is the
  243. stored JSON review document with the summary, verdict, and findings, and
  244. `content` is the Markdown rendering of it. Review text may contain sensitive
  245. repository details, so handle exported files carefully.
  246. Use `--profile metadata` when you only need identifiers, timestamps, verdicts,
  247. cost metadata, and related review metadata.
  248. Exports include a stable `database_id` for the local review database and, when
  249. at least one review is emitted, an opaque `next_cursor`. Pass
  250. `--cursor <next_cursor>` to resume after the previous page; `--cursor` cannot
  251. be combined with `--since`. If a cursor belongs to a previous database
  252. generation, `roborev export reviews` exits with code `3`; discard the cursor
  253. and retry with a window backfill. Other cursor rejections also require
  254. discarding the cursor before backfilling.
  255. Each exported review reports `closed` and `updated_at`. Closing or reopening a
  256. review does not move it past a `completed_at` cursor, so combine the cursor pull
  257. with `roborev export reviews --updated-since <time>` to pick up later close and
  258. reopen changes. See the
  259. [export reference](https://roborev.io/docs/commands/#exporting-reviews).
  260. Use `roborev export ci-metrics` to emit finalized CI panel runs — terminal
  261. outcome (`review_posted`, `review_suppressed`, `no_review_posted`,
  262. `giveup_posted`, `abandoned`, or `unknown` for panels finalized before outcomes
  263. were recorded), first-attempt and posting timestamps, attempt count, and each
  264. panel's member/synthesis jobs — for external review turnaround tracking.
  265. `giveup_posted` is a historical outcome; exhausted retries now finish with
  266. `no_review_posted` and an error status without a PR comment. The export follows
  267. the same cursor contract as `roborev export reviews`, ordered by `posted_at`,
  268. and exits with code `3` when a cursor's `database_id` no longer matches so
  269. callers can discard the cursor and backfill.
  270. Pass `--legacy` to export the frozen pre-panel CI era instead (rows with
  271. outcome `legacy_review`, one per reviewed PR head, from before panel runs
  272. existed) as a one-time backfill; legacy and panel cursors are namespaced
  273. and cannot be resumed against each other's export.
  274. Use `roborev export ci-costs` to emit cost-eligible CI jobs, including terminal
  275. retry attempts that are no longer retained by a panel. Each row records its
  276. completion time, agent, panel role, terminal status, and estimated USD cost.
  277. Jobs whose agent ran but whose model cannot be priced remain present with
  278. `cost_usd: null`; a reported free run is represented as `0`.
  279. Cost rows are ordered by `(finished_at, job_id)` for stable pagination. A fresh
  280. export over an overlapping window returns the current price for every matching
  281. job, so idempotent consumers can pick up pricing recorded after an earlier
  282. export. The export follows the same database-reset and opaque-cursor contract
  283. as the other exports. `--legacy` selects the structurally identified pre-panel
  284. CI era for a one-time historical backfill; regular and legacy cost cursors
  285. cannot be mixed.
  286. ## Configuration
  287. Create `.roborev.toml` in your repo:
  288. ```toml
  289. agent = "claude-code"
  290. snapshot_dir = ".roborev"
  291. review_guidelines = """
  292. Project-specific review instructions here.
  293. """
  294. # Optional: use repo guidelines instead of appending global review_guidelines.
  295. review_guidelines_supersede_global = false
  296. # Optional: metadata for roborev-owned fix commits and prompt hints for agent-owned fix commits.
  297. fix_commit_author = "Your Name <you@example.com>"
  298. fix_commit_co_authored_by = ["Pair Reviewer <pair@example.com>"]
  299. ```
  300. You can also set `review_guidelines` in `~/.roborev/config.toml`. Global
  301. guidelines apply to every repo and are appended before repo guidelines by
  302. default.
  303. `snapshot_dir` must be repo-relative. `roborev init` ensures it is ignored in `.gitignore`; snapshot creation also adds a local `.git/info/exclude` fallback for existing checkouts whose ignore setup is stale.
  304. See [configuration guide](https://roborev.io/docs/configuration/) for all options.
  305. ### Kata task context
  306. If your repo is bound to a [kata](https://github.com/kenn-io/kata) project (a
  307. committed `.kata.toml`), roborev can pull the kata issue(s) referenced in the
  308. reviewed commit messages into the review prompt, and file review findings back
  309. as kata issues.
  310. ```toml
  311. # .roborev.toml
  312. [kata_context]
  313. mode = "current" # off (default) | current | open
  314. max_chars = 50000 # cap on kata context bytes in the prompt
  315. # File a kata issue when a review fails or returns findings:
  316. [[hooks]]
  317. event = "review.*"
  318. type = "kata"
  319. # branches = ["main"] # only file katas for reviews on these branches; default all
  320. # project = "myproj" # defaults to the .kata.toml binding
  321. # labels = ["from-review"]
  322. # priority = 2
  323. ```
  324. `mode = "current"` includes only the katas referenced anywhere in the reviewed
  325. commit messages (e.g. `Closes: kata#abc4`); `open` includes every open kata in
  326. the bound project, except issues the kata hook itself filed (labelled
  327. `roborev`), so review findings are not fed back into later reviews as task
  328. intent. A hook's optional `branches` list (glob patterns such as `release/*`)
  329. limits it to reviews on matching branches; unset fires on all branches. The
  330. matched branch is the commit's branch for local reviews and the PR base
  331. (target) branch for CI pull-request reviews — so `branches = ["main"]` means
  332. commits on `main` locally but PRs *targeting* `main` in CI, and a fork's head
  333. branch name cannot satisfy a protected-branch filter. The `kata` CLI must be
  334. on `PATH`; when it is absent or the repo is unbound, prompt context is
  335. silently skipped. Any other failure — a broken `.kata.toml`, a failing `kata`
  336. invocation — is logged by both the prompt builder and the configured `kata`
  337. hook, so a configured integration never goes dark unnoticed.
  338. ### Environment Variables
  339. | Variable | Description |
  340. |----------|-------------|
  341. | `ROBOREV_DATA_DIR` | Override default data directory (`~/.roborev`) |
  342. | `ROBOREV_TELEMETRY_ENABLED` | Set to `0` to disable anonymous usage telemetry |
  343. | `EDITOR` | Editor used by `roborev job comment` when no message is supplied |
  344. | `GH_TOKEN`, `GITHUB_TOKEN` | GitHub authentication for CI commands |
  345. | `GITLAB_TOKEN`, `GL_TOKEN` | GitLab authentication for CI commands |
  346. | `ROBOREV_COLOR_MODE` | TUI color theme: `auto` (default), `dark`, `light`, `none` |
  347. | `ROBOREV_SYNC_CURSOR_LOOKBACK` | PostgreSQL sync cursor overlap duration (default `5m`) |
  348. | `ROBOREV_AGENT_HOOK_TURN_THRESHOLD` | Override agent-hook Stop threshold |
  349. | `ROBOREV_AGENT_HOOK_COMMIT_THRESHOLD` | Override agent-hook commit threshold |
  350. | `ROBOREV_AGENT_HOOK_FAILED_REVIEW_THRESHOLD` | Override agent-hook failed-review threshold |
  351. | `ROBOREV_AGENT_HOOK_ROBOREV_ADDR` | Override the regular daemon address used by Agent Hook |
  352. | `ROBOREV_DROID_HOOK_TURN_THRESHOLD` | Override Factory Droid agent-hook Stop threshold |
  353. | `ROBOREV_DROID_HOOK_COMMIT_THRESHOLD` | Override Factory Droid agent-hook commit threshold |
  354. | `ROBOREV_DROID_HOOK_FAILED_REVIEW_THRESHOLD` | Override Factory Droid agent-hook failed-review threshold |
  355. | `NO_COLOR` | Set to any value to disable all color output ([no-color.org](https://no-color.org)) |
  356. ## Supported Agents
  357. | Agent | Install |
  358. |-------|---------|
  359. | Codex | `npm install -g @openai/codex` |
  360. | Claude Code | `npm install -g @anthropic-ai/claude-code` |
  361. | Gemini | `curl -fsSL https://antigravity.google/cli/install.sh \| bash` (preferred Antigravity CLI) or `npm install -g @google/gemini-cli` |
  362. | Copilot | `npm install -g @github/copilot` |
  363. | OpenCode | `npm install -g opencode-ai@latest` ([anomalyco/opencode](https://github.com/anomalyco/opencode)) |
  364. | Cursor | [cursor.com](https://www.cursor.com/) |
  365. | Kiro | [kiro.dev](https://kiro.dev/) |
  366. | Kilo | `npm install -g @kilocode/cli` |
  367. | Droid | [factory.ai](https://factory.ai/) |
  368. | Pi | [pi.dev](https://pi.dev/) |
  369. | Muse Code | [Meta installation guide](https://dev.meta.ai/docs/overview) |
  370. roborev auto-detects installed agents.
  371. To use Pi as the auto-design routing classifier (`classify_agent = "pi"`),
  372. install the JSON Schema output extension too:
  373. ```bash
  374. pi install npm:@nqbao/pi-json-schema
  375. ```
  376. roborev loads this extension explicitly when it invokes the classifier. Keeping
  377. it installed in Pi makes the classifier setup visible in `pi list` and avoids
  378. runtime package-fetch surprises in offline or locked-down environments.
  379. ### Routing Claude Code to a proxy (Ollama, LiteLLM, etc.)
  380. The `claude-code` agent accepts a model spec of the form `<model>@<base_url>`.
  381. When `<base_url>` starts with `http(s)://`, roborev points Claude Code at
  382. that endpoint and pins all tier aliases (Opus/Sonnet/Haiku/subagent) to the
  383. given model.
  384. ```toml
  385. # .roborev.toml — local Ollama for reviews, real Anthropic for fixes
  386. agent = "claude-code"
  387. review_model = "glm-5.1:cloud@http://127.0.0.1:11434"
  388. fix_model = "sonnet"
  389. ```
  390. Or via CLI: `roborev review --model 'glm-5.1:cloud@http://127.0.0.1:11434'`.
  391. **Proxy auth.** Set `ROBOREV_CLAUDE_PROXY_TOKEN` to forward a bearer token
  392. to the proxy as `ANTHROPIC_AUTH_TOKEN`. If unset, roborev sends a placeholder
  393. (sufficient for gateways that don't check the header, such as Ollama).
  394. roborev does *not* forward `ANTHROPIC_API_KEY` to proxy endpoints — that
  395. would leak a real Anthropic credential to arbitrary third parties.
  396. **URL restrictions.** Proxy URLs must not embed `user:pass@` credentials
  397. (use `ROBOREV_CLAUDE_PROXY_TOKEN`); `http://` is only accepted for loopback
  398. hosts (`127.0.0.1`, `::1`, `localhost`) so plaintext endpoints can't receive
  399. tokens over the wire. Use `https://` for remote proxies. The full URL
  400. (including any path or query string) is forwarded as-is to
  401. `ANTHROPIC_BASE_URL`, so include the path your gateway expects (e.g.
  402. LiteLLM may want a trailing `/v1`; Ollama wants no path).
  403. **Environment behavior (breaking change in this release).** When the
  404. `claude-code` agent runs, roborev always strips inherited `ANTHROPIC_API_KEY`,
  405. `ANTHROPIC_BASE_URL`, `ANTHROPIC_AUTH_TOKEN`,
  406. `ANTHROPIC_DEFAULT_{OPUS,SONNET,HAIKU}_MODEL`, and `CLAUDE_CODE_SUBAGENT_MODEL`
  407. from the child environment. If you were previously routing Claude Code by
  408. exporting these vars in your shell, switch to the `<model>@<base_url>` spec
  409. instead. For native (non-proxy) mode, configure `ANTHROPIC_API_KEY` via
  410. roborev's config (it is re-injected from roborev's stored key, not inherited
  411. from the operator's shell).
  412. ## Telemetry
  413. Release builds of roborev send limited anonymous telemetry to PostHog when the
  414. daemon starts and once every 24 hours while it remains running: `daemon_started`
  415. and `daemon_active` with `repo_count` and `review_count` as buckets: `0-5`,
  416. `6-20`, `21-100`, or `over-100`. These events also include sync enabled,
  417. CI enabled, and auto-design enabled, plus `application=roborev`, version, OS/arch,
  418. `$process_person_profile=false`, `$geoip_disable=true`, and an anonymous install
  419. ID.
  420. The daemon sends `setup_finished` once per install the first time it finds a
  421. finished review: one that ran locally, arrived through sync, or already existed
  422. when the daemon started. Panel member reviews count. The event has no
  423. properties. Its timestamp records when the daemon noticed the review, so it can
  424. come after an upgrade or after telemetry is turned on. If sending fails, the
  425. daemon does not retry.
  426. Closing a browser tab, leaving it hidden for 30 minutes, or exiting the TUI
  427. reports one `session_ended` with `surface` and `duration_bucket`: `under_1m`,
  428. `1_to_5m`, `5_to_30m`, or `over_30m`. Each browser session adds up visible time
  429. across tab switches and excludes hidden time. A browser session is lost when the
  430. browser quits, crashes, or discards the tab before the page closes normally.
  431. The web UI asks the daemon to report `app_opened` once per UTC day per browser
  432. profile, on load or focus. `roborev tui` asks each time it starts. CLI commands
  433. that work through the daemon ask after their first successful daemon request,
  434. without starting a daemon; hook, MCP and daemon management commands, and
  435. commands the bundled agent skills run with `--from-skill`, do not. Agent calls
  436. count separately as described below. The daemon sends at most one `app_opened`
  437. per surface per UTC day, so repeated opens on the same day count once, with the
  438. same install ID and default fields plus `surface` (`web`, `tui` or `cli`). The
  439. browser, the TUI and the CLI only talk to the daemon; the TUI and the CLI also
  440. send nothing when an opt-out variable is set in their own environment. The
  441. browser and TUI also report `screen_viewed` with a fixed `screen` name and
  442. `surface` of `web` or `tui`. Browser screens are `reviews` and `analytics`. TUI
  443. screens are `queue`, `review`, `prompt`, `filter`, `comment`, `commit-msg`,
  444. `help`, `log`, `tasks`, `worktree-confirm`, `patch`, `column-options`,
  445. `release-notes` and `rerun-agent`. Each screen counts once per install per UTC
  446. day across interfaces, including daemon restarts. Screen names contain no paths,
  447. job IDs, filter text or review contents. `review_viewed` counts a rendered
  448. review read once per installation, surface, and UTC day. It carries
  449. `surface=web|tui|cli`; the TUI requires key or mouse input before counting
  450. loaded content. `finding_fixed` counts each committed fix with
  451. `surface=tui|cli`, and `review_closed` counts each successful explicit close
  452. with `surface=web|tui|cli`. CLI actions require terminal output and exclude
  453. agents and `--from-skill`. Successful reopens and automatic closes stay silent.
  454. Anonymous telemetry also counts core action error banners and toasts in the web
  455. UI and TUI, and PostgreSQL sync outages. Failed closes and reopens both count.
  456. Daemon-unreachable errors can't be counted. See [telemetry configuration](https://github.com/kenn-io/roborev/blob/main/docs/configuration.md#telemetry)
  457. for event properties and reporting rules.
  458. `search_empty` counts each Enter-submitted browser search that returns no jobs
  459. and each human CLI search that finds no reviews, with only `surface=web|cli`.
  460. Previews and search text are excluded.
  461. Daily screen and `review_viewed` claims share `telemetry-daily.json` beside the
  462. daemon database, wherever `--db` points, with a `telemetry-daily.json.lock`
  463. sidecar. Deleting the file resets its claims; existing SQLite claims remain
  464. effective through their recorded UTC dates. See [telemetry
  465. configuration](https://github.com/kenn-io/roborev/blob/main/docs/configuration.md#telemetry)
  466. for recovery instructions. Counts mean the report was queued, not confirmed
  467. delivered to PostHog. The TUI retries a failed screen report on the next key or
  468. mouse input while that screen is showing. The browser sends each screen and
  469. review read once per UTC day per browser profile, across reloads and window
  470. focus returns. It keeps these daily telemetry claims in localStorage.
  471. After a `disabled` receipt, the browser sends nothing until a later UTC day's
  472. load or focus checks again. A TUI, CLI, or MCP process stops reporting until the
  473. next UTC day, a TUI reconnect, or its own restart. The agent-call route returns
  474. the same `disabled` or `queued` JSON receipt as the event route.
  475. Agent activity counts each MCP `tools/call`, including tool errors, and each
  476. invocation of `agent-hook run`, `agent-hook fix-done`, `post-commit`, `enqueue`,
  477. `remap`, or an eligible daemon command marked `--from-skill` after its first
  478. successful product request. MCP discovery and hooks that do no daemon work
  479. add nothing. The daemon emits `agent_active` once per install per UTC day with
  480. `call_count_bucket=1-10`, then `agent_call_count` only on calls 11 and 101 with
  481. `11-100` and `over-100`. The highest bucket observed that day represents its
  482. volume. Counts persist across restarts in the existing local database.
  483. Delivery is best effort and honors all opt-outs.
  484. CLI and stdio MCP notifications and TUI exit reports use a one-second request
  485. timeout. CLI commands, stdio MCP sessions, and the TUI wait up to 50 ms at exit
  486. for unfinished reports. Finished reports add no delay. An unfinished report may
  487. be lost when the process exits. Telemetry alone never starts a daemon.
  488. Each event also carries `install_age_hours`, the whole hours since the install
  489. was created, so short-lived installs such as test sandboxes can be filtered out.
  490. Each event also carries `distribution`, the channel that built the binary, such
  491. as `homebrew` or `kenn` for official direct downloads. It's empty for source
  492. builds.
  493. It does not send repo names, paths, remotes, prompts, review output, provider
  494. tokens, usernames, or IP geolocation. Set `ROBOREV_TELEMETRY_ENABLED=0`,
  495. `TELEMETRY_ENABLED=0`, or `DO_NOT_TRACK=1` to disable it, then restart the
  496. daemon. Telemetry is always off in CI. Release builds, `make build`, and
  497. `make install` set the PostHog project key at link time. Plain `go build`,
  498. `go install`, and Go tests have no key and send no telemetry.
  499. ## Security Model
  500. roborev delegates code review and fix tasks to AI coding agents that
  501. have shell access. Review agents may execute read-only git and shell
  502. commands to inspect diffs; fix agents run in isolated worktrees with
  503. full tool access. A custom review type with a strict tool policy runs
  504. Claude Code in a locked-down session with only read tools and the MCP
  505. tools the operator allows.
  506. **roborev is designed for use with trusted codebases.** The review
  507. prompt includes diff content and commit messages from the repository.
  508. If you are reviewing untrusted code (e.g., open-source contributions
  509. from unknown authors), run roborev inside a sandboxed environment
  510. (container, VM, or similar) to limit the blast radius of any
  511. prompt-injection attack that could cause an agent to execute
  512. unintended commands.
  513. ## Documentation
  514. Full documentation available at **[roborev.io](https://roborev.io)**:
  515. - [Quick Start](https://roborev.io/docs/quickstart/)
  516. - [Installation](https://roborev.io/docs/installation/)
  517. - [Commands Reference](https://roborev.io/docs/commands/)
  518. - [Configuration](https://roborev.io/docs/configuration/)
  519. - [Auto-Fixing with Refine](https://roborev.io/docs/guides/auto-fixing/)
  520. - [Code Analysis and Assisted Refactoring](https://roborev.io/docs/guides/assisted-refactoring/)
  521. - [Hooks](https://roborev.io/docs/guides/hooks/)
  522. - [Agent Hook](https://roborev.io/docs/agent-hook/)
  523. - [Agent Skills](https://roborev.io/docs/guides/agent-skills/)
  524. - [PostgreSQL Sync](https://roborev.io/docs/advanced/postgres-sync/)
  525. For local development in this repo, install hooks with `prek install` or run
  526. `make install-hooks` as a thin wrapper around `prek install`.
  527. ## License
  528. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.