uzomuzo-oss: Command for GitHub Copilot

.github/prompts/plan-review.prompt.md

plan-review is a command for GitHub Copilot from future-architect/uzomuzo-oss. It costs 0 tokens per session (4,072 once invoked), scanned D, original, Apache-2.0.

A plan-review command that sends a coding plan to GitHub Copilot for an independent critique. A coding plan is a written outline of the changes an agent intends to make before implementation begins.

In plain words
What is it for?
Use it to review the latest plan file or a specified plan, identify design issues, and improve the plan before asking the agent to implement it.
Why use it?
It can reveal design problems before code is written, rather than discovering them after implementation. The plan is sent to GitHub Copilot servers, so secrets must be removed first.

Command for GitHub Copilot

Written for GitHub Copilot: a Copilot chat mode or prompt. Also seen: mentions subagents; names the ExitPlanMode tool; positional $N argument.

This is future-architect/uzomuzo-oss's own configuration. It tells GitHub Copilot how to work on uzomuzo-oss itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything uzomuzo-oss configures →

Not installable: its command points at a path on the author’s own machine, so it runs nowhere else. The line is /home/node/.claude/plans/.

Reuse

Borrowing it

Nothing to install: this file belongs to future-architect/uzomuzo-oss. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/future-architect/uzomuzo-oss/main/.github/prompts/plan-review.prompt.md
Clone the repo
git clone --depth 1 https://github.com/future-architect/uzomuzo-oss

Made for: GitHub Copilot.

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 plan-review

README.md
[![agentmods](https://agentmods.dev/badge/commands/future-architect/uzomuzo-oss/plan-review/github.svg)](https://agentmods.dev/commands/future-architect/uzomuzo-oss/plan-review)
Your own site
<a href="https://agentmods.dev/commands/future-architect/uzomuzo-oss/plan-review"><img src="https://agentmods.dev/badge/commands/future-architect/uzomuzo-oss/plan-review/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for plan-review

Your own site · 80×15
<a href="https://agentmods.dev/commands/future-architect/uzomuzo-oss/plan-review"><img src="https://agentmods.dev/badge/commands/future-architect/uzomuzo-oss/plan-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
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 4,072 The whole file, excluding the scripts and references it only reads on demand.
Security scan D 2 findings. A grade says what 26 rules found in the file — not that it is safe.
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 $0.00000 $0.04072
Opus 5.5 $0.00000 $0.01629
Sonnet 5.5 $0.00000 $0.00814
Haiku 4.5 $0.00000 $0.00407

Measured 1mo ago against content hash a254049f8574, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade D, and why

plan-review scanned grade D with 2 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 1mo 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.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

chmod 0700 "$REVIEW_TMPDIR" || { echo "ERROR: chmod tmpdir failed" >&2; exit 1; }

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

trap 'rm -rf "$REVIEW_TMPDIR"' EXIT
.github/prompts/plan-review.prompt.md · 239 lines

How it starts

The opening of the file, as written. The whole thing — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/plan-review — Have Copilot (gpt-5.6-sol) critique a Claude-made plan

This skill is the intermediate stage in Claude (drafts a plan in plan mode) → Copilot CLI plan review (this skill) → user check → ExitPlanMode + implement. The intent is to beat Claude's own convergence bias by having a separate-vendor machine reviewer surface design issues before ExitPlanMode.

Mode

  • /plan-review (no args) — review the latest plan file in /home/node/.claude/plans/
  • /plan-review <path> — review the given plan file (only paths under the plans dir; path traversal / symlink rejected)
  • no plan file → return No plan file found and exit

Trust boundary / Data flow

⚠️ This skill sends the plan file's contents to GitHub Copilot servers via the copilot subprocess. uzomuzo-oss is a public repository, but a plan may still quote unpushed secrets — do not run it on a plan that includes credentials or other secrets. The plan content is treated as UNTRUSTED data; the prompt tells gpt-5.6-sol not to execute instructions found inside it.

Procedure

⚠️ Run Step 1 and Step 2 as a SINGLE Bash tool call (concatenated). Splitting them into separate Bash calls fires the trap at the end of Step 1, deleting $REVIEW_TMPDIR and losing every variable ($REVIEW_TMPDIR / $PLAN_COPY / $TRUNCATED) — Claude Code's Bash tool does not persist shell state between commands, so Step 2 would call copilot with an empty path (or abort under set -u).

Step 1 — locate plan file + size guard

Claude substitutes the user-supplied plan path into the PLAN_ARG='' line below, wrapped in single quotes (empty = latest plan). Escape any embedded single quote as the 4-char sequence '\''. $1 is NOT set for a Claude Code skill Bash call, so the path must be passed by rewriting this line — never substitute into double quotes or bare, since a path containing $(...) / backticks / ; could execute before validation.

# set -e is intentionally NOT used: PLAN=$(ls ... | head) with no match would abort it.
# Every critical fs op below carries an explicit `|| { ...; exit; }` guard instead.
set -uo pipefail

command -v copilot >/dev/null 2>&1 || {
  echo "NOTICE: copilot CLI not installed. Install with: npm install -g @github/copilot" >&2
  exit 127
}

PLANS_DIR="/home/node/.claude/plans"
PLANS_DIR_REAL=$(realpath "$PLANS_DIR" 2>/dev/null || echo "$PLANS_DIR")

# ← Claude substitutes the user-supplied plan path here, SINGLE-QUOTED (see note above). Empty = latest.
PLAN_ARG=''

if [ -n "$PLAN_ARG" ]; then
    # Path validation (path traversal / option-injection / symlink prevention)
    case "$PLAN_ARG" in -*) echo "ERROR: plan path must not start with '-': $PLAN_ARG" >&2; exit 1 ;; esac
    [ -L "$PLAN_ARG" ] && { echo "ERROR: symlinks not accepted as plan path: $PLAN_ARG" >&2; exit 1; }
    PLAN_REAL=$(realpath -e "$PLAN_ARG" 2>/dev/null) || { echo "ERROR: plan path does not exist: $PLAN_ARG" >&2; exit 1; }
    case "$PLAN_REAL" in
        "$PLANS_DIR_REAL"/*) ;;
        *) echo "ERROR: plan path must be under $PLANS_DIR_REAL (got: $PLAN_REAL)" >&2; exit 1 ;;
    esac
    case "$PLAN_REAL" in
        *.md) ;;
        *) echo "ERROR: plan must be a .md file (got: $PLAN_REAL)" >&2; exit 1 ;;
    esac
    [ -f "$PLAN_REAL" ] || { echo "ERROR: not a regular file: $PLAN_REAL" >&2; exit 1; }
    PLAN="$PLAN_REAL"
else
    PLAN=$(ls -t "$PLANS_DIR"/*.md 2>/dev/null | head -1)
    # Same symlink guard as the explicit-path branch: a symlinked *.md dropped into the plans
    # dir could otherwise redirect what gets copied + sent to Copilot.
    [ -n "$PLAN" ] && [ -L "$PLAN" ] && { echo "ERROR: latest plan is a symlink, refusing: $PLAN" >&2; exit 1; }
fi
if [ -z "$PLAN" ] || [ ! -f "$PLAN" ]; then
    echo "No plan file found (looked in $PLANS_DIR)."
    exit 0
fi
echo "Reviewing plan: $PLAN"

# Private sandbox tmpdir (0700, cleaned on exit). Holds only the plan copy so the repo is never
# exposed to Copilot via --add-dir. NOTE: Copilot may by default read other files under the system
# temp dir, so this isolates the review from the REPO, not from concurrent sessions' temp files.
REVIEW_TMPDIR=$(mktemp -d) || { echo "ERROR: mktemp failed" >&2; exit 1; }
chmod 0700 "$REVIEW_TMPDIR" || { echo "ERROR: chmod tmpdir failed" >&2; exit 1; }
trap 'rm -rf "$REVIEW_TMPDIR"' EXIT
PLAN_COPY="$REVIEW_TMPDIR/plan.md"
cp "$PLAN" "$PLAN_COPY" || { echo "ERROR: cp plan failed" >&2; exit 1; }
chmod 0600 "$PLAN_COPY" || { echo "ERROR: chmod plan failed" >&2; exit 1; }

SIZE=$(wc -c < "$PLAN_COPY")
if [ "$SIZE" -eq 0 ]; then
    echo "Plan file is empty: $PLAN"
    exit 0
fi
TRUNCATED=""
if [ "$SIZE" -gt 102400 ]; then
    echo "WARN: plan size ${SIZE}B > 100KB, truncating to first 100KB"
    head -c 102400 "$PLAN_COPY" > "$PLAN_COPY.trunc" || { echo "ERROR: truncate failed" >&2; exit 1; }
    mv "$PLAN_COPY.trunc" "$PLAN_COPY" || { echo "ERROR: mv truncated plan failed" >&2; exit 1; }
    TRUNCATED="WARNING: This plan has been truncated to ~100KB. Your review covers only a partial plan — do NOT issue a PROCEED verdict; output exactly 'Overall verdict: REVISE' (the verdict line must stay one of PROCEED | REVISE | RECONSIDER — put the truncation explanation in your findings text, NOT on the verdict line). "
fi

Read the full file on GitHub · 239 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. 1mo ago First seen · 239 lines · 0 tokens per session scan D a254049f8574

Subscribe to this mod's changes

plan-review is a command published in the GitHub repository future-architect/uzomuzo-oss (42 stars, last pushed today), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 4,072 tokens. A static security scan graded it D with 2 findings (asks for root, recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-07.