A tool for reviewing code changes, files, and documents with inline notes in a text-based interface. It also includes automatic review of work plans.

Annotating diffs, files, and documents, and reviewing plans automatically.

It makes review comments visible alongside the material being checked and helps catch issues in proposed plans.

revdiff — Plugin by umputun

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.

Claude Code

/plugin marketplace add umputun/revdiff

agentmods

npx agentmods add plugins/umputun/revdiff/marketplace

Clone the repo

git clone --depth 1 https://github.com/umputun/revdiff

Made for: Claude Code.

Per session not measured What this adds to a session before it is invoked.

When invoked not measured Not applicable: nothing here is loaded into a session.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Security

Grade A, and why

revdiff 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.

The scan reads marketplace.json. This mod also ships 5 executable files (skills/revdiff/scripts/agentdeck-window.sh, skills/revdiff/scripts/detect-ref.sh, skills/revdiff/scripts/launch-revdiff.sh, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

.claude-plugin/marketplace.json · 28 lines

What it actually says

{
  "name": "revdiff",
  "owner": {
    "name": "umputun",
    "email": "[email protected]"
  },
  "plugins": [
    {
      "name": "revdiff",
      "source": "./",
      "description": "Review diffs, files, and documents with inline annotations in a TUI overlay",
      "version": "0.8.23",
      "author": {
        "name": "umputun"
      }
    },
    {
      "name": "revdiff-planning",
      "source": "./plugins/revdiff-planning",
      "description": "Automatic plan review with revdiff",
      "version": "0.3.9",
      "author": {
        "name": "umputun"
      }
    }
  ]
}

Files

What ships with it

10 files beside marketplace.json in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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.

yesterday First seen · 28 lines scan A ad89acc89271

Subscribe to this mod's changes

revdiff is a plugin published in the GitHub repository umputun/revdiff (826 stars, last pushed 3d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. 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 2026-08-30.

Code review Planning Productivity agentic-workflow claude-code diff tui

claude-code-tips

Developer experience essentials: GitHub Actions debugging, conversation half/quarter-cloning, context handoffs, research (Reddit and Hacker News), private GitHub search, and Claude Code version checks.

ykdojo/claude-code-tips · not measured

kiln

Multi-model software creation pipeline. A conversation becomes running, tested software — an interactive brainstorm, then an autonomous build: GPT-5.6 writes the code, Claude shapes the plans and interfaces, and each family inspects the work of the other.

Fredasterehub/kiln · not measured

sound

Code taste as installable rules. Probe a repo's stack, select only the rules its code shows a surface for (evidence-cited per rule), confirm, and scaffold a browsable .claude/sound/ rule map (surfaced by the sound:prime skill). Installed rules are user-owned; re-runs reconcile.

saadshahd/moo.md · not measured

revdiff

Getting it into your agent

Grade A, and why

Nothing flagged

What it actually says

What ships with it

What this file has done since we first saw it

Other plugins, from other repositories

claude-code-tips

kiln

sound

superpipelines-marketplace

promptlog

forwward-teams