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.
git clone --depth 1 https://github.com/LeanAndMean/scramjetWrote 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.
[](https://agentmods.dev/commands/leanandmean/scramjet/mach12-gh-pr-read)<a href="https://agentmods.dev/commands/leanandmean/scramjet/mach12-gh-pr-read"><img src="https://agentmods.dev/badge/commands/leanandmean/scramjet/mach12-gh-pr-read/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.
<a href="https://agentmods.dev/commands/leanandmean/scramjet/mach12-gh-pr-read"><img src="https://agentmods.dev/badge/commands/leanandmean/scramjet/mach12-gh-pr-read.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00025 | $0.00756 |
| Opus 5 | $0.00013 | $0.00378 |
| Sonnet 5 | $0.00005 | $0.00151 |
| Haiku 4.5 | $0.00003 | $0.00076 |
Grade A, and why
mach12:gh-pr-read 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Read GitHub Pull Request
Goals
- Return the caller an authoritative pull-request snapshot containing the parent metadata and complete chronological top-level conversation.
- When requested, identify the most recent matching marker comment and its numeric ID without weakening completeness checks.
- Make incomplete or malformed history explicit rather than presenting partial evidence as authoritative.
This subroutine is gh-specific. A future forge-agnostic command set would substitute an equivalent glab-pr-read (or similar); the marker-hunt logic stays the same.
Step 1: Parse input
Extract:
- The PR number (required, first token).
- An optional
--marker <html-marker>flag naming an HTML comment marker to locate (e.g.,--marker mach12-review,--marker mach12-assessment).
If no PR number is present, return an error to the caller and stop.
Step 2: Read the PR and all top-level PR conversation comments
Resolve the canonical owner/name with gh repo view --json nameWithOwner. Query the PR through gh api graphql --paginate with explicit variables for owner, name, PR number, and $endCursor. Request parent title, body, createdAt, updatedAt, and:
comments(first: 100, after: $endCursor) {
totalCount
nodes { databaseId body author { login } authorAssociation createdAt url }
pageInfo { hasNextPage endCursor }
}
The query must declare $endCursor: String and pass pageInfo.hasNextPage plus pageInfo.endCursor so --paginate follows every page. Accumulate comment nodes in chronological page order. Verify the accumulated node count exactly equals totalCount; reject duplicate database IDs. If pagination stops early, a page is malformed, the count differs, or any command fails, surface the full error and report that authoritative history is incomplete; do not return a partial array as complete.
Step 3: Locate the marker comment (if requested)
If --marker <html-marker> was provided, parse the comments array and scan from the end (most recent first) for the first comment whose body contains the literal HTML marker <!-- <html-marker> -->. If multiple comments contain the marker, use the most recent one.
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.
- 4d ago Changed · +4 lines 5ef5bc42fba6
- 9d ago First seen · 55 lines · 25 tokens per session scan A bea07eb7f04e
mach12:gh-pr-read is a command published in the GitHub repository LeanAndMean/scramjet (6 stars, last pushed today), licensed Apache-2.0. It adds 25 tokens to every session and 756 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 2026-08-31.
Other commands, from other repositories
afrv
Revise the current feature worktree after code review — decide accept/revert/modify (shortcut for feature-code-revise).
aigon-feature-code-revise
A reviewing agent has just committed fixes (or notes) on this feature branch. Your job — as the implementing agent — is to read what the reviewer did, then decide how to respond: accept, revert, or modify.
feature-code-review
Review feature - code review with fixes by a different agent.
aigon-feature-code-review
Perform a code review on another agent's implementation, making targeted fixes where needed. Use a different model than the implementer for best results.
feature-code-revise
Revise the current feature worktree after code review — decide accept/revert/modify.
done
Finish a task - document, create PR or merge, close.