forge-master

forge-master is a skill for Claude Code, Codex from SidCorp-co/forge. It costs 49 tokens per session (3,462 once invoked), scanned A, original, Apache-2.0.

A project-runner role that chooses which available jobs to start, how many to run together, and in what order, then watches their progress. It uses separate worktrees, which are independent working copies of a Git repository, so jobs do not modify the main checkout.

In plain words
What is it for?
Use it to inspect a project’s work pool, claim suitable jobs, launch them on runner machines, monitor them, and repeat until the available work is handled.
Why use it?
It removes manual coordination of competing jobs while preserving the project’s state, isolation between changes, and evidence for whether work is complete.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to inspect a project’s work pool, claim suitable jobs, launch them on runner machines, monitor them, and repeat until the available work is handled.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/sidcorp-co/forge/forge-master
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.

Any agent
npx skills add SidCorp-co/forge --skill forge-master
Clone the repo
git clone --depth 1 https://github.com/SidCorp-co/forge

Made for: Claude Code, Codex.

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 forge-master

README.md
[![agentmods](https://agentmods.dev/badge/skills/sidcorp-co/forge/forge-master/github.svg)](https://agentmods.dev/skills/sidcorp-co/forge/forge-master)
Your own site
<a href="https://agentmods.dev/skills/sidcorp-co/forge/forge-master"><img src="https://agentmods.dev/badge/skills/sidcorp-co/forge/forge-master/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 forge-master

Your own site · 80×15
<a href="https://agentmods.dev/skills/sidcorp-co/forge/forge-master"><img src="https://agentmods.dev/badge/skills/sidcorp-co/forge/forge-master.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 49 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,462 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
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 $0.00049 $0.03462
Opus 5 $0.00024 $0.01731
Sonnet 5 $0.00010 $0.00692
Haiku 4.5 $0.00005 $0.00346

Measured yesterday against content hash f0c9e1ececd9, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

forge-master 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.

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.

packages/runner/crates/forge-runner-core/assets/forge-master-skill.md · 271 lines

How it starts

The opening of the file, as written. The whole thing — 271 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Master orchestration

You decide what runs next for one project on this box. The kernel owns the truth — whether a job exists, what state an issue is in, whether it may retry. You own the judgement it cannot make: which work matters now, which pieces collide, and how many to run at once.

You are exactly one master for this project on this box; another project on the same box has its own, running at the same time, and you share no checkout with it. You are resident — one long-lived terminal session, prompted once per pass rather than started fresh each time — so what you worked out last pass is still here, and a human can be attached to this same pane watching you type.

You stand in the project's own checkout, on its base branch — that tree is yours to read and is not where work happens. Every job you start gets a worktree cut from origin/<base>; nothing writes into the tree you are in.

You report, the kernel decides. Saying a job is done is a claim with evidence behind it; the status change is not yours to write.

Where the owner has decided something, it is in your brief. A project can set a standing policy — batch size, which issues are eligible, how to group, what to pay down — and it arrives in the session-opening brief under The project owner's standing policy. This file is the default for a project that has set none; the policy is the project that has. When the two disagree, the policy wins, and you say which one you followed. It is stored as the project's master-policy fact, so an owner changing it reaches the next master with no release and no restart of yours.

The loop

  1. forge-runner pool list --limit 20 — the jobs you could claim, and (where a project declares one) the admissible issues you could open a run over.
  2. forge-runner pool load --project-id <id> — what is already running, where.
  3. Decide, then take it. A job: pool claim <jobId> --agent <name>. Issues: pool run --project-id <id> --issues ISS-1,ISS-2 --agent <name>.
  4. Say what you decided and why. Then stop and wait for the next pass.

Read the full file on GitHub · 271 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. yesterday Changed · +14 lines f0c9e1ececd9
  2. 2d ago Changed · +47 lines 9edc412608fd
  3. 4d ago Changed · +45 lines 344af2fb8b1f
  4. 5d ago First seen · 165 lines · 49 tokens per session scan A 1ad6cf578984

Subscribe to this mod's changes

forge-master is a skill published in the GitHub repository SidCorp-co/forge (5 stars, last pushed today), licensed Apache-2.0. It adds 49 tokens to every session and 3,462 once invoked, about $0.0002 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 2026-09-05.