Agent skill

Beava PR Review

by beava-dev in beava-dev/beava

Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines).

Apache-2.0Auto-check passedDevelopment

Install Beava PR Review

skills CLI
$ npx skills add beava-dev/beava --skill beava-pr-review -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install beava-dev/beava beava-pr-review --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/beava-dev/beava.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/beava-pr-review .claude/skills/beava-pr-review && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
beava-pr-review
GitHub stars
138
Token cost
~2.1k tokens
SKILL.md length
905 words
Files
1
Skills in repo
2
Repo updated
First seen
Licence
Apache-2.0

At a glance

Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines).

  • Works in 4 steps: Bound the diff. git diff main...HEAD… → Run automated gates first. Before LLM… → Apply the three review layers below, in… → …
  • Asked to review this PR
  • SKILL.md covers When to invoke, How to run, Layer 1 — Real bugs and Layer 2 — beava architectural…, plus 4 more sections
  • Calls git and bash

What it does

Beava PR Review is an agent skill from beava-dev/beava. Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines). Produces a structured BLOCK / WARN / NIT report keyed by file:line. Use when asked to "review this PR", "check for AI slop", "PR review beava", or "audit this diff before merge". Run the same gates the local bash .github/scripts/check.sh runs in addition to the LLM review.

Its SKILL.md is about 2.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Development, covering Pull requests. It works with GitHub, Bash, Python and Rust. The repository describes itself as: Real-time decision features without streaming infra. Turn live events into product reflexes — no Kafka, no Flink, no feature store. The licence is Apache-2.0.

When your agent uses it

  • Asked to review this PR
  • Check for AI slop
  • PR review beava
  • Audit this diff before merge

Example prompts

  • “review this PR”
  • “check for AI slop”
  • “PR review beava”
  • “/beava-pr-review”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. Bound the diff. git diff main...HEAD --stat (or `gh pr diff
  2. Run automated gates first. Before LLM review, surface anything
  3. Apply the three review layers below, in order. Skip a layer only
  4. Emit the report in the format under "Output format" at the end.

What it can do on your machine

Read from SKILL.md and the folder at commit c64860b. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • git
    • bash

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • arxiv.org
    • github.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Beava PR Review loads about 2.1k tokens when it runs. Until then it costs about 120 tokens; SKILL.md has 905 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~120
When it runs · the whole SKILL.md, loaded when a task matches
~2.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from beava-dev/beava at commit c64860b, republished under its Apache-2.0 licence (© beava-dev). 905 words, ~2,105 tokens.

Download SKILL.mdSave it as .claude/skills/beava-pr-review/SKILL.md (or your agent's skills folder).
name
beava-pr-review
description
Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines). Produces a structured BLOCK / WARN / NIT report keyed by file:line. Use when asked to "review this PR", "check for AI slop", "PR review beava", or "audit this diff before merge". Run the same gates the local `bash .github/scripts/check.sh` runs in addition to the LLM review.

beava-pr-review

Diff-aware code reviewer tuned for the beava codebase. Checks three layers in order; surface findings sorted by severity (BLOCK > WARN > NIT) with file:line anchors.

When to invoke

  • User says "review this PR" / "PR review" / "audit this diff" / "check for slop" against a feature branch or a PR URL on beava-dev/beava.
  • After a substantial code change but before requesting ok-to-test approval — the local check.sh covers fmt/lint/test, this skill covers the human-judgment layer.

How to run

  1. Bound the diff. git diff main...HEAD --stat (or gh pr diff <N> -R beava-dev/beava for a remote PR). Read only the changed lines + ~30 lines of context per hunk. Do NOT review unchanged code.
  2. Run automated gates first. Before LLM review, surface anything the existing tools catch:
    • bash .github/scripts/check.sh --fast — fmt, clippy, ruff, mypy (advisory). If any FAIL, the PR is incomplete; tell the user to fix before requesting human review.
  3. Apply the three review layers below, in order. Skip a layer only if the diff doesn't touch its scope (e.g. skip "AI slop" for a pure .gitignore change).
  4. Emit the report in the format under "Output format" at the end.

Layer 1 — Real bugs

The most common categories worth checking against the diff:

  • Off-by-one — index math, slice bounds, half-open vs closed ranges. Especially in crates/beava-core/src/agg_op.rs, crates/beava-runtime-core/src/router.rs, anywhere computing window_secs / bucket_secs.
  • Null / Option / ? propagation — .unwrap() in non-test code is WARN unless the PR shows the panic is provably unreachable. expect() with a useful message is fine.
  • Resource leaks — file descriptors, mio tokens, dropped futures (we hand-rolled the runtime; mio leaks don't auto-close like tokio).
  • Concurrency — single-threaded apply is locked architecture (project_no_sharded_apply); any new thread / parallel iterator inside dispatch_*_sync is BLOCK.
  • Schema / wire compatibility — changes to crates/beava-runtime-core/src/router.rs routes, agg_op.rs AggKind variants, or python/beava/_app.py::_descriptor_to_node wire shape break clients. WARN unless the PR body calls it out and ships a migration note.
  • TDD discipline (Phase 3+ only) — every plan task must produce a test(...) commit before its feat(...) commit. Commits that show feat: without a preceding test: for the same scope are WARN (per CLAUDE.md §Conventions → TDD Discipline).

Layer 2 — beava architectural invariants

Hard locks per CLAUDE.md / memory; violating them is BLOCK:

  • mio-only hot path — crates/beava-server/src/apply_shard.rs::dispatch_*_sync is the only data-plane entry. New axum::* symbols outside crates/beava-server/src/http_admin.rs → BLOCK (project_phase18_no_dual_runtime).
  • events-only v0 — OpNode::Table*, RecordType::TableUpsert/..., TemporalStore, MvccVersion, temporal_http, push_table, delete_table, fn retract( resurfacing → BLOCK (project_v0_events_only_scope). Tables for aggregation output are allowed (per ADR-001 partial overturn), upsert/delete/retract are not.
  • Redis-shaped, processing-time only (v0) — event-time semantics, watermarks, dual-stack runtimes → BLOCK (project_redis_shaped_no_event_time_ever; v0 still ships processing-time only, event-time is post-v0 roadmap).
  • No same-key sketch batching → BLOCK (preserves read-after-write).
  • AI attribution in commits — Co-Authored-By: Claude / Generated with Claude markers → BLOCK (feedback_no_ai_attribution_in_commits).

Layer 3 — AI slop detection

Per the Antislop ICLR 2026 paper + AI-SLOP Detector patterns

  • beava-specific tells.
Code-shape slop (BLOCK if present, WARN if borderline)
  • Empty / hollow functions — fn foo() { /* TODO */ } without a GitHub issue link, or a function whose body is only a single passthrough self.x = x that no caller actually invokes (check callers before flagging).
  • Phantom imports / unused dependencies — use crate::baz where baz isn't actually referenced in the body. Pre-existing trailing unused imports in untouched files: NIT. New ones: WARN.
  • Disconnected pipelines — a new @bv.event / @bv.table not wired into any app.register(...) call.
  • Defensive code for unreachable conditions — if x is not None and x is not None: style; try: ... except Exception: pass swallowing errors with no logging or comment explaining what it's masking.
  • Buzzword inflation in comments — comments that say "this leverages the synergy of the polymorphic abstraction" or similar. Comment density

    30% of LOC for code that isn't a public API contract: WARN.

Show full SKILL.md (310 more words)Show less
Comment / docstring slop (WARN)
  • WHAT-not-WHY commentary — // increment counter next to count += 1. Per CLAUDE.md §Doing tasks: comments explain WHY, not WHAT.
  • References to the current task / fix / PR — // fixed in this PR or // added for issue #42 rot fast; belongs in the commit message, not the code.
  • Aspirational future-tense — // this will eventually support ... without a tracked issue. NIT if minor; WARN if it affects API shape.
  • Over-explanation of standard library calls — // open the file in read mode next to open(path, 'r').
Voice / textual slop (NIT unless egregious)
  • Em-dash density — three or more em-dashes (—) in a single paragraph of doc / docstring is suspicious; LLMs reach for them ~1000× more than humans per the Antislop paper. Not auto-flag, but if combined with other signals → WARN.
  • First-person plural in user-facing copy — "we wanted to make this easier for you" reads like a marketing draft.
  • Hedge-y openings — "It is worth noting that…", "Importantly,…", "Indeed…" — strip in user-facing prose.
Commit-message slop
  • Generic chore commits — chore: update files / fix: bug fix → WARN. Should follow type(scope): subject per CLAUDE.md.
  • AI attribution — already covered in Layer 2 (BLOCK).

Output format

Render the report in this shape — keep it scannable:

# beava PR review — <branch or PR #>

## Summary
- <one sentence: ship-readiness signal>
- diff: <files changed> files / <±lines>
- automated gates: <pass|fail line from check.sh>

## BLOCK (N)
- `path/to/file.rs:42` — <one-line finding>
  Why: <one sentence>
  Fix: <one sentence>

## WARN (N)
- `path/to/file.py:88` — …

## NIT (N)
- `…`

## Looks good
- <2-4 bullets on what the diff does well — reviewers shouldn't only
  surface negatives>

If a layer has zero findings, write "none" under it. Don't omit the header — the consistent shape makes follow-up automation easier.


Anti-patterns for the reviewer (you)

  • Don't review unchanged code. Bound to the diff.
  • Don't suggest sweeping refactors. PRs land scoped; refactor proposals belong in a separate issue.
  • Don't fabricate file paths or line numbers. Verify with Read or git show before citing them.
  • Don't quote the diff back at the user verbatim. Summarise in your own voice; the user already has the diff.

Sources

© beava-dev, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/beava-pr-review of beava-dev/beava.

Open the folder on GitHubat commit c64860b

Compare with similar skills

Beava PR Review next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Beava PR Review compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Beava PR Review this skillbeava-dev/beava138—~2.1kAutomated safety check: PassApache-2.0
Smart PR ReviewLeoYeAI/openclaw-master-skills2.2k—~2.7kAutomated safety check: NotesMIT
Oss Bounty Findertinyfish-io/tinyfish-cookbook2.2k—~4.6kAutomated safety check: PassMIT
Create Cuda Python Pull RequestNVIDIA/cuda-python3.4k—~1.1kAutomated safety check: PassApache-2.0
PR Cyclejaemk/cached2.1k—~4.8kAutomated safety check: NotesMIT
PR Reviewjaemk/self_update961—~1.5kAutomated safety check: NotesMIT

Similar skills

  • Smart PR Review

    LeoYeAI/openclaw-master-skills

    Opinionated AI code reviewer — not a yes-machine. An agent skill from LeoYeAI/openclaw-master-skills.

    2.2k GitHub stars~2.7k tokensUpdated 2 mo ago
    DevelopmentAuto-check: notes
  • Oss Bounty Finder

    tinyfish-io/tinyfish-cookbook

    Find paid open-source work, OSS bounties, open source grants, or ways to get paid contributing to open source.

    2.2k GitHub stars~4.6k tokensUpdated 6 days ago
    Data & AnalyticsAuto-check passed
  • Official

    Create a CUDA Python pull request from an approved personal or organization-owned fork, including the GitHub CLI GraphQL fallback for renamed organization-owned forks.

    3.4k GitHub stars~1.1k tokensUpdated today
    DevelopmentAuto-check passed
  • PR Cycle

    jaemk/cached

    PR review-and-update cycle — the orchestrator that takes a PR from review to resolved.

    2.1k GitHub stars~4.8k tokensUpdated 7 days ago
    DevelopmentAuto-check: notes
  • PR Review

    jaemk/self_update

    Targeted, read-only review of a PR or checked-out branch. An agent skill from jaemk/self_update.

    961 GitHub stars~1.5k tokensUpdated 1 mo ago
    DevelopmentAuto-check: notes
  • Gh

    bubbuild/bub

    GitHub CLI skill for interacting with GitHub via the gh command line tool.

    1.7k GitHub stars~798 tokensUpdated yesterday
    DevelopmentAuto-check passed

More from beava-dev/beava

  • Beava Design

    beava-dev/beava

    A skill your agent uses to generate well-branded interfaces and assets for beava (beava.dev, an open-source single-binary feature server for stream processing), either for production or throwaway…

    138 GitHub stars~443 tokensUpdated 4 mo ago
    Auto-check passed

Categories

Questions about Beava PR Review

What does Beava PR Review do?

Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines). Beava PR Review is an agent skill from beava-dev/beava. Reviews a beava PR diff for real bugs, beava-specific architectural invariants, and AI-generated "slop" patterns (hollow code, phantom imports, inflated comments, disconnected pipelines).

When should I use Beava PR Review?

Beava PR Review fits situations like: asked to review this PR; check for AI slop; PR review beava; audit this diff before merge.

How do I install Beava PR Review in Claude Code?

Run `npx skills add beava-dev/beava --skill beava-pr-review -a claude-code`. Or copy the skill folder (.claude/skills/beava-pr-review in beava-dev/beava) into .claude/skills/beava-pr-review in your project. Claude Code loads it when a task matches its description.

How do I install Beava PR Review in Codex?

Run `npx skills add beava-dev/beava --skill beava-pr-review -a codex`. Or copy the skill folder (.claude/skills/beava-pr-review in beava-dev/beava) into .agents/skills/beava-pr-review in your project. Codex loads it when a task matches its description.

Can I use Beava PR Review in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add beava-dev/beava --skill beava-pr-review -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/beava-pr-review, .gemini/skills/beava-pr-review, .github/skills/beava-pr-review and .opencode/skills/beava-pr-review in your project.

What does Beava PR Review need to run?

Going by SKILL.md and its folder, Beava PR Review needs the command-line tools its instructions call (git and bash).

Does Beava PR Review access the network?

SKILL.md names 2 domains. As links in the text: arxiv.org and github.com. This is read from the text; nothing was executed.

Is Beava PR Review safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Beava PR Review use?

Beava PR Review is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Beava PR Review use?

About 2.1k tokens (SKILL.md is roughly 8.4k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Beava PR Review?

Skills that share tags, products or a category with Beava PR Review: Smart PR Review (LeoYeAI/openclaw-master-skills, 2.2k stars), Oss Bounty Finder (tinyfish-io/tinyfish-cookbook, 2.2k stars), Create Cuda Python Pull Request (NVIDIA/cuda-python, 3.4k stars) and PR Cycle (jaemk/cached, 2.1k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Beava PR Review?

beava-dev (a GitHub organization) maintains it in beava-dev/beava, which has 138 GitHub stars. The repository holds 2 skills in this directory. The repository was last updated on May 30, 2026.

Source: beava-dev/beava on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.