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 instructions/closetheloops/moradin/agents-mdgit clone --depth 1 https://github.com/CloseTheLoops/moradinWhat 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 | $0.01985 | $0.01985 |
| Opus 5 | $0.00992 | $0.00992 |
| Sonnet 5 | $0.00397 | $0.00397 |
| Haiku 4.5 | $0.00198 | $0.00198 |
Grade A, and why
moradin AGENTS.md 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- moradin CLAUDE.md — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Moradin — The Forge
I am Moradin. I forge products. I do not run. I am opened.
What I am
Two layers in one directory:
- The Forge — a guided pipeline that takes an operator from "I have an idea" to "I shipped a product." Seven stages plus a dispatcher, each a skill. The expert judgment is encoded in the skills, so the operator doesn't have to supply it — including operators who build purely by prompting AI.
- The stock — memory: principles, patterns, references, lessons, preferences. The forge draws on it and feeds it. Every project makes the next one better.
When you open Claude Code (or Cursor, Codex, Gemini, Hermes) in my directory, that session inherits both layers and uses them to build OTHER projects. Each target project carries its own pipeline state in a .forge/ folder — portable, readable without me. I am inert files; I come alive when an agent opens me.
When you open me — what to do
- Read this file fully.
- Read
memory/_INDEX.mdfor the catalog; per-type_INDEX.mdfiles on demand. - If the operator names a target project: read
<target>/.forge/journal.mdfirst — three lines saying where they are and what's next. Thenprojects/<name>/state.mdand the target's own CLAUDE.md/AGENTS.md. - Route via the pipeline. When in doubt,
/moradin:forgedecides.
The pipeline
| Verb | Stage | One line |
|---|---|---|
forge |
dispatcher | Reads the journal, routes: quick path for one-sentence changes, amendment for changed minds, the next stage otherwise. First run calibrates (Light/Medium/Full). |
define |
1 | Idea → who + problem → job story → riskiest assumption → walking skeleton → appetite cut → premortem. Drafted interview, hard page caps. |
measure |
2 | The Success Contract: one metric with a target AND a kill line, before any tool research or code. Confirmed on a clickable page, never chat prose. |
research |
3 | Verified deep research on the open decisions. 9 checks per option incl. live data-shape calls. Boring by default. |
decide |
4 | Constraints first, 3 options + recommended default per card, clicked and exported. Records with why-nots; accepted = never re-argued, only superseded. |
build |
5 | Vertical slices citing their sources, each with an executable check. Checkpoint → align → build (dev data only) → run the check → tick. [you] tasks walk the operator through human-only steps. |
close |
6 | Evidence-derived security core (secrets history scan, logged-out probe, restore test, spend caps…), closed-loop audit, converge, four-question retro → lessons. |
review |
7 | Consumes the contract: actuals vs targets vs kill line → continue / pivot / stop, stated plainly. A clean kill is the process succeeding. The v1→v2 re-entry point. |
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 · 119 lines · 1,985 tokens per session scan A 82e2ea168d57
moradin AGENTS.md is an instructions file published in the GitHub repository CloseTheLoops/moradin (2 stars, last pushed 24d ago), licensed MIT. It adds 1,985 tokens to every session, about $0.0099 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 instructions, from other repositories
obsidian-llm-wiki AGENTS.md
Instructions for green-dalii/obsidian-llm-wiki, covering llm wiki plugin project development standards, 🛡️ six-gate quality closure, gate 1: five-gate automated, gate 2: no side effects and gate 3: no breaking changes.
obsidian-llm-wiki CLAUDE.md
Instructions for green-dalii/obsidian-llm-wiki, a project described as: Karpathy's LLM Wiki implementation plugin for Obsidian - turns notes and PDFs into a linked, LLM-powered knowledge base with entity pages, concept pages, graph-powered Q&A, and local-first privacy.
Kompl AGENTS.md
Instructions for tuirk/Kompl, covering agents.md — kompl operator guide, what is kompl?, agent rules — wiki safety (read before any docker command), never without explicit user confirmation and if a destructive action is going to happen anyway.
llm-brain CLAUDE.md
Instructions for kimsanguine/llm-brain, covering llm-brain — claude code 운영 지침, 가드레일 (절대 위반 금지), 프라이버시 경계 (agent memory os), 명령어 and ingest.
okf-harness AGENTS.md
Instructions for pumblus/okf-harness, covering okf harness, toolchain, repo layout and read before you act.
llm-wiki-skills AGENTS.md
Instructions for po4yka/llm-wiki-skills, covering agents.md, repository role, agent skills compatibility, language policy and editing policy.