dw

A command wrapper for the Diffwarden review skill. It lets you choose actions such as reviewing, fixing and verifying in a loop, checking status, or writing a short pull-request comment.

In plain words
What is it for?
Use it with a workspace, git changes, pull request, or document to review findings, run fixes, check scores, or prepare a review comment.
Why use it?
It provides a consistent shortcut for applying Diffwarden's review rules from supported coding agents. A pull-request comment is a concise review message attached to a proposed code change.

Command

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/jperocho/diffwarden/dw
Clone the repo
git clone --depth 1 https://github.com/jperocho/diffwarden
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 650 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin 92% copy Near-identical to another mod 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 $0.00000 $0.00650
Opus 5 $0.00000 $0.00325
Sonnet 5 $0.00000 $0.00130
Haiku 4.5 $0.00000 $0.00065

Measured yesterday against content hash f81e1990d719, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dw 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 yesterday.

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.

Origin

This is a copy

92% identical to diffwarden — 2 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

skills/diffwarden/commands/dw.md · 24 lines

What it actually says

Diffwarden (/dw)

Read and follow the diffwarden skill (SKILL.md).

The user text after this command (if any) is a Diffwarden invocation. Parse per the skill Slash Commands section:

  • No args → help
  • Else parse subcommand (review, loop, status, comment, help), an optional target, and flags
  • Primary subcommands: review (read-only), loop (review-fix-verify until c5/5), status (score only), comment (short PR review comment), help
  • Compatibility aliases (accept, don't advertise): fixloop; prepareloop --push; securityreview --security; review-plan <file>review <file> --as-plan; fix-plan <file>loop <file> --as-plan
  • Targets: workspace (current folder, git not required), local (git working tree), staged (git staged), PR (#123, URL), or file path (plan/docs/guides/tutorials)
  • Flags: --mvp, --verbose, --orchestrate, --commit, --push, --as-code, --as-plan, --web (alias --research), --reply, --resolve, --delegate, --dry-run, --max N, --review-model, --fix-code-model, --fix-text-model, --security, --go (Go profile for code targets), --lang <name> (language profile for code targets)
  • review/loop auto-detect code vs document mode per skill Target Auto-Detection; --as-code/--as-plan override. comment is PR-only.
  • --web is opt-in per-finding consent; valid on review/loop (code targets); rejected on status and document mode.
  • --orchestrate enables optional reviewer/fixer role split; off by default. Model flags override config when set.
  • Default output is lean (cN/5 loop lines plus final Status:/Level: lines); --verbose restores full report.
  • Final-line contract: every review/loop/status/comment must end with exactly Status: then Level: as the last two output lines. Pi/master/orchestration wrappers do not replace them.

Examples: loop workspace, review #123, comment #123, status local, loop docs/install.md --as-plan, review --security, loop --orchestrate, loop --mvp, review #123 --web, review local --lang go, loop staged --go.

If the caveman skill is loaded, run in caveman mode — see the skill Caveman Mode section.

Expand to skill flags, run full Diffwarden loop. Do not rephrase unless parse fails.

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. yesterday First seen · 24 lines · 0 tokens per session scan A f81e1990d719

Subscribe to this mod's changes

dw is a command published in the GitHub repository jperocho/diffwarden (2 stars, last pushed 21d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 650 tokens. A static security scan graded it A with 0 findings. It is 92% identical to diffwarden, differing in 2 lines, and is treated as a copy.