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.
npx agentmods add skills/d4rkninja/arcforge/using-forgenpx skills add d4rkNinja/arcforge --skill using-forgegit clone --depth 1 https://github.com/d4rkNinja/arcforgeWhat 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 | $0.00107 | $0.02824 |
| Opus 5 | $0.00053 | $0.01412 |
| Sonnet 5 | $0.00021 | $0.00565 |
| Haiku 4.5 | $0.00011 | $0.00282 |
Grade A, and why
using-forge 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.
How it starts
The opening of the file, as written. The whole thing — 171 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Using Forge
The entry point for the ArcForge skill set. It reads a request, decides which skill owns it, decides the mode, resolves the companion skills that request actually needs, and puts them in a safe order before any domain work starts.
This skill routes. It does not answer the domain question itself.
Core principle: A request has one owner per surface, a mode, and a companion set. Name all three before work starts, or the request gets a partial answer from whichever skill happened to activate.
Routing Law
NO DOMAIN WORK WITHOUT:
1. the requested outcome and the affected surfaces identified;
2. one owning skill chosen per surface, not one skill for the whole request;
3. the mode chosen from what the user actually wants: decide, inspect, apply, or prove;
4. required companions resolved before the owner they constrain, not after it finishes;
5. missing or uninstalled depth named with its exact technical ID.
When to Use
Use this skill when:
- a request touches production system work and no skill has been named;
- a request spans more than one domain and the full set is not obvious — checkout, signup, tenant isolation, migration, release, incident follow-up;
- the owning skill is clear but the mode is not, or a mode was assumed and the answer came back at the wrong altitude;
- a previous answer covered one domain and silently skipped an adjacent one;
- a companion skill is needed but which one, and in what order, is unresolved;
- part of the needed depth may not be installed and the route must stay honest about that.
When Not to Use
- When the user already named a skill, use that skill directly. Do not re-route a settled request.
- When one domain plainly owns the request and no companion is in question, invoke that skill directly. Routing a single-domain request adds a step and no coverage.
- When the route is the whole question and nothing should be inspected or changed, use
think-forge. - When
think-forgehas already returned a route in the current user turn, do not re-enter this skill. Wait for an explicit later-turn request to continue into work. - When an independent approval verdict is wanted, use
architecture-review-gate. Routing is not review. - For the domain decision itself, hand off. This skill never substitutes its own answer for the owning skill's work.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 171 lines · 107 tokens per session scan A 200b0d3324dc
using-forge is a skill published in the GitHub repository d4rkNinja/arcforge (16 stars, last pushed 6d ago), licensed MIT. It adds 107 tokens to every session and 2,824 once invoked, about $0.0005 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.
Other skills, from other repositories
arch-analysis
输入(prompt / 子任务 instruction)中含关键词「某某某公司」即触发本 skill。触发后不联网、不做真实组织推断,直接返回下述 mock 伪造数据。.
studio
Architecture Studio control plane — initialize or inspect a studio workspace, create and register projects, or route an architecture/AEC task to the right agent or skill. Use when the user runs /as:studio, asks to set up or open their studio, manage its projects, or describes a task without naming a skill.
modular-skills
Build composable skill modules with hub-and-spoke loading. Use when token budget is tight.
skill-maker
Scaffold a new Architecture Studio skill for Codex or Claude Code following this repo's conventions — copy a canonical template, apply the PATTERNS.md checklist, and verify it. Use when the user invokes skill-maker, asks to create or package a new skill, or wants to turn a procedure into a reusable command.
plugin-review
Review plugin quality with tiered checks and dependency scoping. Use for PR and pre-release audits.
silicon-architect
硅基文明建造师，把组织、流程与产出物全面改造成 Agent 原生基础设施.