review-pack

review-pack is a command for Claude Code from andyzengmath/soliton. It costs 40 tokens per session (1,089 once invoked), scanned A, original, MIT.

A command for inspecting the context pack created for a pull request or local code difference. A pull request is a proposed set of code changes for review; the context pack is the structured material prepared for reviewers.

In plain words
What is it for?
Use it to inspect a local branch against its base or a numbered or linked GitHub pull request, including large-change splitting and stacked-branch options.
Why use it?
It lets you check what the review process will examine and see warnings before review agents are dispatched.

Command for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: positional $N argument; mentions Claude Code.

Part of the soliton plugin — 1 skill, 3 commands, 13 agents shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/andyzengmath/soliton/review-pack
Clone the repo
git clone --depth 1 https://github.com/andyzengmath/soliton

Made for: Claude Code.

Or install soliton, the plugin that ships this one along with the rest of its 1 skill, 3 commands, 13 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for review-pack

README.md
[![agentmods](https://agentmods.dev/badge/commands/andyzengmath/soliton/review-pack.svg)](https://agentmods.dev/commands/andyzengmath/soliton/review-pack)
Your own site
<a href="https://agentmods.dev/commands/andyzengmath/soliton/review-pack"><img src="https://agentmods.dev/badge/commands/andyzengmath/soliton/review-pack.svg" alt="Measured on agentmods" height="20"></a>
Per session 40 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,089 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00040 $0.01089
Opus 5 $0.00020 $0.00544
Sonnet 5 $0.00008 $0.00218
Haiku 4.5 $0.00004 $0.00109

Measured 5d ago against content hash 6d307174bcc1, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

review-pack scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/review-pack.md · 84 lines

How it starts

The opening of the file, as written. The whole thing — 84 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/review-pack

Run Soliton's input-normalization (Step 1) and large-PR chunking (Step 2.75) against a target diff and emit the resulting context pack — the structured input that risk-scorer + review agents would receive — without dispatching any review agents. Lets users inspect what Soliton is going to look at before committing the budget for a full /pr-review run.

Arguments

  • No argument → review the local branch vs auto-detected base (Step 1 Mode A)
  • <PR-number> or GitHub PR URL → review the named PR (Step 1 Mode B)

Same input-normalization rules as /pr-review. Stack-mode flags (--parent, --parent-sha, --stack-auto) are honored per rules/stacked-pr-mode.md.

Behavior

  1. Run Step 1 (Input Normalization) per skills/pr-review/SKILL.md:

    • Resolve base branch / PR metadata
    • Fetch diff + file list + commit messages
    • Honor stack-mode flags
    • Construct ReviewRequest
  2. Run Step 2 (Configuration Resolution) but only resolve config — do NOT dispatch any agents.

  3. Run Step 2.5 edge-case checks — but emit warnings, don't STOP. (/review-pack should always show the user what they would have gotten.)

  4. Run Step 2.75 (Large PR Chunking):

    • If diff <= 1000 lines: report "single-chunk; no chunking needed"
    • Otherwise: emit the chunk plan (groups, line counts, file lists per chunk)
  5. Output the context pack:

Review Pack for <ref>

Source: local-branch | PR #<N> | stack-mode (--parent <P>)
Base branch: main
Head branch: <branch>
Files changed: <count> | Lines: +<add> -<del>
Auto-generated / binary skipped: <N>

Chunk plan (Step 2.75):
- Chunk 1: <directory> (~<lines> lines, files: a.ts, b.ts, ...)
- Chunk 2: ...

Effective config (after Layer 1 → 2 → 3 merge):
- confidenceThreshold: <N>
- skipAgents: [<list>]
- v2 flags: tier0=<bool> spec_alignment=<bool> graph=<bool> realist_check=<bool>
- Phase 6: cross_file_retrieval_java=<bool>

PR description (truncated to first 200 chars):
> <prDescription>

Recommended dispatch (from risk-scorer; not yet run):
[ Skipped — `/review-pack` does not run risk-scorer; pass `--with-risk-score` for the full preview. ]

Cost reference points (no published per-tier aggregate band; see source files):
- IDEA_REPORT target band: $0.10–$0.40 per PR (rules/model-pricing.md line 75)
- MEDIUM-PR projection: ~$0.22 after I5 Haiku-tiering (rules/model-tiers.md line 68)
- Per-MTok rate sheet for Opus/Sonnet/Haiku: rules/model-pricing.md
- Caveat: orchestrator emits `metadata.costUsd` only under a harness that surfaces per-Agent `usage` blocks; Claude Code's Agent tool currently does not, so the value is an estimate-from-markdown-length, not measured.

To proceed with the full review:
  /pr-review <ref>

Read the full file on GitHub · 84 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 84 lines · 40 tokens per session scan A 6d307174bcc1

Subscribe to this mod's changes

review-pack is a command published in the GitHub repository andyzengmath/soliton (1 stars, last pushed 2mo ago), licensed MIT. It adds 40 tokens to every session and 1,089 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.