melech-plan-review
Review a solution before approving it, like a GitHub changes tab: a file tree of red and green pseudo-code the user can comment on inline, then answer…
Skill by AdirD
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/adird-melech-plan-review
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/melech-plan-review
What's inside
- Skill name
melech-plan-review- Description
- Review a solution before approving it, like a GitHub changes tab: a file tree of red and green pseudo-code the user can comment on inline, then answer Approve, Commented, or Rejected. Use when the user wants to see what would change, a git-like diff, or the PR as if it were ready.
- License
- MIT
- Instructions
- 49 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Files (9)
- README.md
- SKILL.md
- melech-plan-review-tip-diagram.jpg · assets/
- evals.json · evals/
- PLAN.md · evals/fixtures/invoice-reminders/
- invoices.ts · evals/fixtures/invoice-reminders/apps/api/src/billing/
- invoice-reminders-plan-review.json · evals/fixtures/invoice-reminders/
- review_server.py · scripts/
- template.html
SKILL.md (first lines)
# Plan review
Foresee this solution before they approve it. Show what would change, and what
they should expect, as a file tree of git-like red and green lines, as if the
PR is ready and they are reviewing the changes tab.
The page is for them or a product reader. They are not looking to deep-dive
the code.
Mark each part of a file with a blue `@@ section @@` line. Indent the way a
diff would. Write a sentence when the behavior is the point. Write a short
code line when that is easier to read than the sentence. Do not paste the
surrounding implementation.
## Open it for review
Copy [template.html](template.html) to `<plan-name>-plan-review.html` next to
the plan file, or in the repo root when the plan lives only in the chat. Fill
every `{{TOKEN}}`.
Serve it in the background with `scripts/review_server.py` beside this file:
```bash
python3 <skill-dir>/scripts/review_server.py <plan-name>-plan-review.html
```
It opens the page and prints the URL and the review file path. Comments save
to `<plan-name>-plan-review.json` next to the page as the reader writes them.
## Wait for the verdict
Ask one question with three options: Approve, Commented, Rejected. Use your
question tool if you have one. Otherwise ask in chat and end the turn.
When they answer, read the review file. It has an overall `summary` and
`comments`. Each comment has a `body` and the `places` it points at: `file`,
`lines`, and the `quote` they selected. One comment can point at several
lines, or at places in several files. A `file` of `null` is the page header.
- Approve: the plan stands. Fold in any comments, then stop the server.
Do not start implementing unless they ask.
- Commented: answer each comment in chat, change the plan where a comment
asks, and rewrite the page. Rename the review file to
`<plan-name>-plan-review-<round>.json`, ask them to refresh the tab, and ask
again.
- Rejected: read the comments for why. If there are none, ask what is wrong.
Do not rewrite the page until the direction is clear. Stop the server.From AdirD/agent-shell-hamelech, MIT.
More from AdirD/agent-shell-hamelech
- 0.662score
melech-cost
AdirD
Show what coding-agent sessions cost for every worktree of the current repo — Cursor, Claude Code, and Codex — as handoff-style cards with dollars and tokens.
SkillAFreshness: A★ 45Claude Code +9 - 0.662score
melech-live-browser
AdirD
Operate the user's already-open, logged-in Chrome tabs through Chrome DevTools MCP.
SkillAFreshness: A★ 45Claude Code +9 - 0.662score
melech-handoff
AdirD
List and continue coding-agent transcripts across Cursor, Claude Code, Codex, Gemini, and other local agents.
SkillAFreshness: A★ 45Claude Code +9 - 0.659score
melech-tidy
AdirD
Adaptive diff reduction — prune dead code residue, shake prompt bloat, remove cosmetic diff noise, and propose edge-case trades that drop rare-case handling…
SkillAFreshness: A★ 45Claude Code +9 - 0.659score
melech-debug-mode
AdirD
Debug or verify a local UI end to end in the user's existing Chrome tab, with temporary runtime probes and a manual browser fallback.
SkillAFreshness: A★ 45Claude Code +9 - 0.656score
melech-pr-gardener
AdirD
Tend one explicit GitHub author's open PRs and keep them green — one stateless pass per scheduled run.
SkillAFreshness: A★ 45Claude Code +9 - 0.651score
melech-8020
AdirD
Find the smallest useful path to a product or engineering outcome.
SkillAFreshness: A★ 45Claude Code +9 - 0.651score
melech-smart-comments
AdirD
Preserve code intent with selective comments and protect meaningful existing comments.
SkillAFreshness: A★ 45Claude Code +9 - 0.622score
melech-roi
AdirD
Grade whether a change's complexity and blast radius are actually worth the benefit it delivers, using an itemized ROI receipt and rankings among solutions…
SkillBFreshness: B★ 45Claude Code +9 - 0.619score
melech-make-it-click
AdirD
Recover when the agent's explanation did not land by finding the smallest thing blocking understanding and replacing the failed mental model with a concrete…
SkillBFreshness: B★ 45Claude Code +9 - 0.617score
melech-consult
AdirD
Get an independent verdict on an AI-proposed plan, fix, architecture, or claim by verifying its premises against the real artifact, then dispatching a panel…
SkillBFreshness: B★ 45Claude Code +9 - 0.616score
melech-buy-vs-build
AdirD
Decide whether to adopt an existing tool or build a capability yourself, grounded in verified research.
SkillBFreshness: B★ 45Claude Code +9
Signals
- Stars
- 45
- Forks
- 5
- Open issues
- 0
- Last commit (this folder)
- Oct 5, 2026 (22 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/AdirD/agent-shell-hamelech
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from AdirD
- 0.662score
melech-cost
AdirD
Show what coding-agent sessions cost for every worktree of the current repo — Cursor, Claude Code, and Codex — as handoff-style cards with dollars and tokens.
SkillAFreshness: A★ 45Claude Code +9 - 0.662score
melech-handoff
AdirD
List and continue coding-agent transcripts across Cursor, Claude Code, Codex, Gemini, and other local agents.
SkillAFreshness: A★ 45Claude Code +9 - 0.662score
melech-live-browser
AdirD
Operate the user's already-open, logged-in Chrome tabs through Chrome DevTools MCP.
SkillAFreshness: A★ 45Claude Code +9 - 0.659score
melech-debug-mode
AdirD
Debug or verify a local UI end to end in the user's existing Chrome tab, with temporary runtime probes and a manual browser fallback.
SkillAFreshness: A★ 45Claude Code +9 - 0.659score
melech-tidy
AdirD
Adaptive diff reduction — prune dead code residue, shake prompt bloat, remove cosmetic diff noise, and propose edge-case trades that drop rare-case handling…
SkillAFreshness: A★ 45Claude Code +9 - 0.656score
melech-pr-gardener
AdirD
Tend one explicit GitHub author's open PRs and keep them green — one stateless pass per scheduled run.
SkillAFreshness: A★ 45Claude Code +9
Cite this page
melech-plan-review skill. Agent Mart. https://agentmart.dev/skills/adird-melech-plan-review (last updated 2026-10-24)