copilot-leak

A check for issues created by GitHub Copilot’s automatic code reviews, using repository review records and labels or author details to identify them.

In plain words
What is it for?
Search a chosen time window, classify detected review issues, sanitise them, and report clearly when GitHub access or checking was unavailable.
Why use it?
It finds bugs that passed the project’s own checks and turns them into lessons instead of leaving them hidden in review history.

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/azrtydxb/procoder/copilot-leak
Clone the repo
git clone --depth 1 https://github.com/azrtydxb/procoder
Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 725 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 $0.00026 $0.00725
Opus 5 $0.00013 $0.00362
Sonnet 5 $0.00005 $0.00145
Haiku 4.5 $0.00003 $0.00072

Measured 2d ago against content hash ea4c786b751e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

copilot-leak 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 2d 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.

.kilo/commands/copilot-leak.md · 66 lines

How it starts

The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.

The user invoked /procoder:copilot-leak with arguments: $ARGUMENTS

Run:

procoder copilot-leak $ARGUMENTS

(With no arguments it looks back 24 hours. --since 6h narrows the window to any duration; --quiet reports the count and never prompts; --from-copilot reads the captured ledger back instead of going to look for more, and exits 1 while any finding is still unclassified.)

The command reads this repository's Copilot auto-review issues through gh: the ones carrying the auto-copilot label, opened by a copilot*[bot] author, or carrying Copilot's review quote block. A finding there is a bug that got past our own gates, which makes it a lesson, not just a fix.

Read the exit, not the mood

  • Exit 0 with no findings means the window is genuinely empty — or that this repository has no GitHub remote, in which case there were never any auto-reviews to ask about.
  • Exit 2 means the answer was no — either the prompt was declined, or the window was never queried at all.

A check that could not run is never reported as clean. gh missing, gh unauthenticated, a network failure, or output the parser does not recognise all end as NOT checked and exit 2. Fix the cause and re-run: procoder init installs gh, and gh auth login authenticates it. Do not describe the repository as free of Copilot findings on the strength of an exit 2.

When stdin is not a terminal there is nobody to answer, so the answer is no and the run exits 2. That is a report, not an error — re-run it in a terminal when you want the capture.

Privacy is the command's job, and yours

Nothing reaches GitHub or the ledger unsanitised: fenced code is stripped, secret-shaped strings are redacted, paths are made relative to the project. What survives is the defect description, the file and line, and the fix Copilot proposed. If sanitisation leaves a body empty, that finding is skipped and said so — an empty body is never padded with the original text.

Hold the same line yourself. Never paste a raw Copilot body into an issue, a commit message, or the chat to "add context"; quote the sanitised text the command printed.

Read the full file on GitHub · 66 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. 2d ago First seen · 66 lines · 26 tokens per session scan A ea4c786b751e

Subscribe to this mod's changes

copilot-leak is a command published in the GitHub repository azrtydxb/procoder (196 stars, last pushed 3d ago), licensed Apache-2.0. It adds 26 tokens to every session and 725 once invoked, about $0.0001 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 2026-08-30.