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/closetheloops/moradin/decidenpx skills add CloseTheLoops/moradin --skill decidegit clone --depth 1 https://github.com/CloseTheLoops/moradinWrote 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/skills/closetheloops/moradin/decide)<a href="https://agentmods.dev/skills/closetheloops/moradin/decide"><img src="https://agentmods.dev/badge/skills/closetheloops/moradin/decide.svg" alt="Measured on agentmods" 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 | $0.00065 | $0.00830 |
| Opus 5 | $0.00032 | $0.00415 |
| Sonnet 5 | $0.00013 | $0.00166 |
| Haiku 4.5 | $0.00006 | $0.00083 |
Grade A, and why
moradin-decide 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 3d 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 — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
moradin:decide
Choices get made once, with reasons, and stay made — so the build never re-argues the database at midnight. The operator clicks; the records write themselves.
When invoked
/moradin:decide — requires .forge/research/ output. If the research stage didn't already produce the options page, build it now from the research files.
Step 1 — Constraints before options
Before showing anything, confirm the operator's constraints in one short exchange: monthly budget ceiling, expected scale ("just me" is a fine answer), and what they already know how to use. Criteria stated after seeing options get bent to justify a favorite — locking them first is the whole discipline.
Step 2 — The options page
Generate (or reuse) the single-file interactive HTML report and publish it for the operator:
- One card per decision. 2–3 options; the recommended one pre-selected and tagged. Never more — choice overload hits non-experts under exactly these conditions.
- Per option: one-line what-it-is · cost now and at 10× scale · spend cap yes/no · exit cost in days · a "pick this if…" sentence · risk lines (repricing history, single-maintainer, UNVERIFIED flags) in a warning treatment.
- Non-dev triage on every card: accepting the default is always a first-class choice; "explain this in plain language" is a legitimate export value that routes back to a conversation, not a failure.
- One export button for the whole page → JSON blob pasted back into chat.
Style/mechanics follow the project's existing review pages (dark, single file, no external requests, clipboard export with textarea fallback).
Step 3 — Lock the records
For each exported choice, append a record to .forge/decisions.md (~15 lines, from templates/forge/decisions.md):
- Numbered title · date ·
Status: accepted· door type with concrete exit ("two-way, exit ≈ 1 day: pg_dump → any host") · innovation token yes/no - Context (2–3 sentences: what forced this, the operator's stated constraints)
- Decision (one sentence, active voice) · Why (2–3 bullets tied to those constraints)
- Rejected — every option not chosen, one line each on why not. The why-nots are what prevent relitigating.
- Verified — source URL + fetch date, spend-cap status
- Revisit when — event triggers only ("pricing changes", ">10k users", "PumpPortal goes down"). Never calendar dates; stale decision logs die of scheduled reviews nobody does.
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.
- 3d ago First seen · 47 lines · 65 tokens per session scan A 587ce77afc51
moradin-decide is a skill published in the GitHub repository CloseTheLoops/moradin (2 stars, last pushed 25d ago), licensed MIT. It adds 65 tokens to every session and 830 once invoked, about $0.0003 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 skills, from other repositories
okf-harness-bootstrap
This unified entrypoint routes setup and workspace maintenance without making the user choose a mode.
okf-harness
One Door routes OKF Harness workspace requests to exactly one internal workflow at a time.
llm-wiki-faq
Explain LLM-Wiki adoption and answer stakeholder objections. Use when the user wants a concise FAQ response, evidence summary, ROI framing, migration answer, or plain-language rebuttal; route adversarial risk registers and bad-fit audits to llm-wiki-critique-audit.
llm-wiki-mcp-integration
Design or review MCP/API integration for an LLM-Wiki. Use when the user wants agents such as Claude Code, Codex, Cursor, ChatGPT, VS Code, GitHub Copilot, LangGraph or other MCP clients to search, read, lint, rescan, propose, review or export wiki knowledge through MCP or a local/remote API.
llm-wiki-retrieval-architect
Design the retrieval/indexing layer for an LLM-Wiki. Use for lexical/FTS search, BM25, hybrid retrieval, vector databases, rerankers, metadata filters, or GraphRAG, or when rg/index.md no longer retrieves enough context; route whole-stack local-first tooling choices to llm-wiki-local-first-stack.
llm-wiki-eval-tooling
Select evaluation tools and CI gates for LLM-Wiki measurement. Use when the user asks which eval framework, dataset format, scorecard file, red-team config, prompt regression test, hosted/self-hosted service, or human calibration workflow to use; route actual pilot runs to llm-wiki-eval.