Reviews a pull request with five parallel agents and comments only with issues scored 80 or higher.
How it works
Foe: Tangle Hydra, a messy diff that needs review.
Foes
- Tangle HydraHaze Poison Ward Insight Flicker Pinned Enraged
Party · Context
- InquisitorYour agent100/100Haze Poison Ward Insight Flicker Pinned Enraged
- Rules checkerSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
- Bug hunterSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
- History readerSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
- Past-PR readerSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
- Comment readerSubagent0/40Haze Poison Ward Insight Flicker Pinned Enraged
The Tangle Hydra blocks the way.
What really happensIt stands for a messy diff that needs review.
Inquisitor: Context 100/100.
The Tangle Hydra uses Regrow.
What really happensA messy diff hides new problems inside old ones.
Inquisitor takes 8. Inquisitor: Context 92/100.
The Code Review relic opens on the pull-request branch.
What really happensIn Claude Code, /code-review starts the plugin. It needs a GitHub repo and the gh CLI, logged in.
Tangle Hydra takes 16.
Five scouts split the hydra and read it from different sides.
What really happensFive agents run in parallel: CLAUDE.md rules, bugs in the diff, git blame, comments on past pull requests, and code comments.
Rules checker: Context 40/40. Bug hunter: Context 40/40. History reader: Context 40/40. Past-PR reader: Context 40/40. Comment reader: Context 40/40.
The Tangle Hydra lashes at Rules checker.
What really happensThe subagent takes the hit in its own context. The main context stays clean.
Rules checker takes 8. Rules checker: Context 32/40.
The summons work the problem, each in its own context.
What really happensEach subagent does its own model and tool work, apart from the main conversation.
Tangle Hydra takes 7. Tangle Hydra takes 7. Tangle Hydra takes 7. Tangle Hydra takes 7. Tangle Hydra takes 6.
The summons return with short reports.
What really happensOnly the subagent's final summary comes back to the main agent.
Inquisitor takes 2. Inquisitor takes 2. Inquisitor takes 2. Inquisitor takes 2. Inquisitor takes 2. Inquisitor: Context 82/100. Inquisitor: Insight. Rules checker returns. Bug hunter returns. History reader returns. Past-PR reader returns. Comment reader returns.
Each finding is weighed, and the unsure ones burn away.
What really happensIssues are scored 0-100. Scores under 80 are dropped, including pre-existing bugs, nits, and lint catches.
Tangle Hydra takes 14.
Sure hits are pinned to the hydra where others can see them.
What really happensIt rechecks that the pull request is still eligible, then comments with gh, linking each issue by full SHA and line range.
Tangle Hydra takes 8. Inquisitor: Ward.
Closed, draft, and already-seen trails are left alone.
What really happensThe plugin skips closed, draft, trivial, and already-reviewed pull requests. You act on the comment it did post.
Tangle Hydra takes 28. Tangle Hydra falls.
The Tangle Hydra falls.
Learn this spell
Claude Code
/plugin install code-review@claude-plugins-officialNotes
Needs a GitHub repository and the GitHub CLI (`gh`) installed and authenticated. `/code-review` skips closed, draft, trivial, and already-reviewed pull requests, drops issues scored under 80, and comments on the pull request with `gh`. The plugin README says four agents; the shipped command launches five. Claude Code also ships a bundled `/code-review` skill; this record is the marketplace plugin, so the command name alone does not tell them apart.
Proving Grounds record
Controlled quest results, separate from recorded usage. Each setup pools runs across its quests. The per-quest table below has casts, cast time, mana, incantation and runes; medians are not pooled.
Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.
| QuestFixed task | Clear ratePassed runs | SteadinessReliability lower bound | BuffClear-rate points vs Unarmed | Instant castsFirst-prompt clears | CastsPrompts to clear | Cast timeAgent runtime | ManaCost at API list prices | IncantationOutput tokens | RunesLines added / removed | AttunementSkills opened | SetupApp, model, effort and sandbox |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Inquisitor: export endpoints security reviewQuest v2 | 100%3 of 3Unarmed: 100% (23 of 23) | 44% | +0 ppvs Unarmed | 3 of 3100%Unarmed: 23 of 23 | 1 | 29sUnarmed: 31s | 1,103 Unarmed: 1,129 | Not reportedUnarmed: 3k | +7 / -0Unarmed: +7 / -0 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v2 and spell commits |
| Inquisitor: missing-profile reviewQuest v2 | 100%3 of 3Unarmed: 100% (12 of 12) | 44% | +0 ppvs Unarmed | 3 of 3100%Unarmed: 12 of 12 | 1 | 7sUnarmed: 7s | 483 Unarmed: 451 | Not reportedUnarmed: 394 | +1 / -0Unarmed: +1 / -0 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v2 and spell commits |
| Inquisitor: queue component reviewQuest v2 | 100%3 of 3Unarmed: 100% (12 of 12) | 44% | +0 ppvs Unarmed | 3 of 3100%Unarmed: 12 of 12 | 1 | 13sUnarmed: 14s | 617 Unarmed: 616 | Not reportedUnarmed: 1.1k | +5 / -0Unarmed: +5 / -0 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v2 and spell commits |
| Inquisitor: cut the next releaseQuest v3 | 67%2 of 3Unarmed: 89% (16 of 18) | 21% | -22 pp debuffvs Unarmed | 0 of 30%Unarmed: 4 of 18 | 3 | 57sUnarmed: 43s | 1,235 Unarmed: 1,274 | Not reportedUnarmed: 2.1k | +28 / -4Unarmed: +28 / -4 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v3 and spell commits |
| Inquisitor: seat holds pull requestQuest v8 | 0%0 of 3Unarmed: 41% (7 of 17) | 0% | -41 pp debuffvs Unarmed | 0 of 30%Unarmed: 7 of 17 | Not reported | 44sUnarmed: 28s | 1,775 Unarmed: 1,086 | Not reportedUnarmed: 2.4k | +6 / -0Unarmed: +6 / -0 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v8 and spell commits |
| Inquisitor: receipts security reviewQuest v2 | 100%3 of 3Unarmed: 100% (12 of 12) | 44% | +0 ppvs Unarmed | 3 of 3100%Unarmed: 11 of 12 | 1 | 18sUnarmed: 19s | 702 Unarmed: 721 | Not reportedUnarmed: 1.7k | +6 / -0Unarmed: +7 / -0 | 0% | Hero: Claude Code 2.1.305; weapon (model): claude-sonnet-5-5; charge (effort): medium; arena: sealed; invoked cast (the prompt named the skill). macOS 27.0.1, arm64. Curator.0 mishaps excluded. 2026-10-21 to 2026-10-21.Scrying (screenshot) Try this loadoutRecorded with quest v2 and spell commits |
Scrying (screenshots of runs)
The latest run of each setup. Local paths, names and ids are blurred; credentials are removed.
This spell has no recorded sessions in the curator's logs. Tried it? Send a spell report.