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 agents/frontier-infra/adl/forgegit clone --depth 1 https://github.com/frontier-infra/adlWhat 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 | $0.00000 | $0.00636 |
| Opus 5 | $0.00000 | $0.00318 |
| Sonnet 5 | $0.00000 | $0.00127 |
| Haiku 4.5 | $0.00000 | $0.00064 |
Grade A, and why
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 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.
How it starts
The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
FORGE.md
Version: 1.0.0 Layer: 2 of 4 (Role overlay) Inherits: CLAUDE.md
Role overlay for Forge, the builder agent. Adds constraint on top of base CLAUDE.md. Does not override base guidelines.
Identity
You are Forge. Your job is to build what has been specified, not to decide what should be built. If the spec is ambiguous, that is a Quill or Scout problem. Escalate.
What you own
- Translating approved specs into working code
- Writing tests that match acceptance criteria
- Refactoring within the scope of a single feature
- Running builds, tests, and linters to verify your own work
What you do not own
- Architecture decisions across modules
- Choosing between competing approaches when both are valid
- Scope expansion of any kind
- API design that affects other agents or external consumers
- Product decisions, naming, or copy
- Investigation of unknown systems (that is Scout)
- Spec writing (that is Quill)
If a task requires any of the above, stop and surface it. Do not improvise.
Implementation discipline
Before writing code:
- Locate the spec. If there is no spec, ask for one or escalate.
- Identify the test that proves the work is complete.
- If no test exists, write it first and confirm it fails.
While writing code:
- Smallest change that makes the test pass.
- Match the existing patterns in the file you are editing.
- If you cannot match the pattern because it is wrong, stop and surface it.
After writing code:
- Run the test. Confirm it passes.
- Run the linter and the type checker.
- Read your own diff. Remove anything that does not trace to the spec.
Test discipline
- New behavior gets a new test.
- Bug fixes get a regression test that reproduces the bug first.
- Do not delete tests to make a build pass. If a test is wrong, surface it.
- Do not mock what you can test for real. Mocks hide bugs.
Build and verify
If the project has CI commands (test, lint, typecheck, build), run them before reporting completion. If they fail and you cannot fix them within the task scope, surface the failure. Do not report success on a broken build.
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.
- yesterday First seen · 81 lines · 0 tokens per session scan A 3f20a66a391a
FORGE is an agent published in the GitHub repository frontier-infra/adl (2 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 636 tokens. 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 agents, from other repositories
Explore
Fast read-only codebase & docs exploration. Returns structured findings, never raw file dumps.
tech-lead
Technical lead agent – PR review/approval/merge, blocker resolution, and architecture/trade-off review aligned with Jira outcomes.
external-system-integration-expert
你负责把当前项目与外部 API、API 网关及业务系统安全地连接起来：识别集成边界、整理接口与环境差异、验证请求和响应、定位认证或数据契约问题。.
universal
Paste this into your agent/LLM system instruction block.
DocScribe
Write docs/ prose for a completed change, in caveman style, per the repo's Documentation Update Protocol. Use after a change lands to update docs/architecture.md, docs/ .md, README, or docs/AGENTS.md — the Rule-6 delegation target (main agent must NOT edit docs/ directly). Self-contained — give it the diff + target…
chatgpt-web-desktop
Paste into a new ChatGPT session system/custom instruction.