Magnitude is an open-source inference server, meaning software that runs language models and answers requests on a user's own computer. It profiles available hardware, recommends suitable local models, and connects them to coding agents for private and offline use on macOS, Linux, or Windows through WSL. Its catalogue entries help agents set up and use Magnitude.
Borrowing it
Nothing to install: this file belongs to magnitudedev/magnitude. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/magnitudedev/magnitude/main/.agents/skills/refine/SKILL.mdgit clone --depth 1 https://github.com/magnitudedev/magnitudeWrote 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/magnitudedev/magnitude/refine)<a href="https://agentmods.dev/skills/magnitudedev/magnitude/refine"><img src="https://agentmods.dev/badge/skills/magnitudedev/magnitude/refine/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.
<a href="https://agentmods.dev/skills/magnitudedev/magnitude/refine"><img src="https://agentmods.dev/badge/skills/magnitudedev/magnitude/refine.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 5.1 | $0.00013 | $0.01162 |
| Opus 5 | $0.00006 | $0.00581 |
| Sonnet 5 | $0.00003 | $0.00232 |
| Haiku 4.5 | $0.00001 | $0.00116 |
Grade A, and why
refine 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 10d 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 — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Overview
This skill documents a generic process for iterative code refinement. It is intented to be used when code changes are in place, but their correctness and quality is not yet guaranteed.
The objective of this skill's use is to prescribe an iterative loop that can be followed to ensure correctness and optimize for code quality.
Phase 1: Grounding
Before reviewing the code, it is important to understand the changes, the context behind them, and any requirements or constraints they are intended to follow.
If you are the agent which made these changes, you may already have some of this context, however you still may be missing some.
Gather the following information:
- The actual code changes, staged and unstaged (use git diff and/or read files directly as appropriate)
- Any plan the changes are based on
- Any relevant design docs
- Related code and architecture
Use the code changes to inform what other code and design docs you need to read to get the full picture.
Plan documents
Plan documents represent some point-in-time intended delta. Consider that a plan may prescribe changes that were adapted in-flight. The plan may not serve as an up-to-date source of truth for what was meant to be implemented, and may be missing information or contain outdated information. The plan should serve as one piece of evidence in your understanding of the changes being undergone and what they may or may not be doing correctly, rather than some ultimate source of truth.
Design documents
Design documents serve as a semantic basis for the intented project state. These documents, however, may not be fully up to date. Relevant design documents should be read, but their content should be treated as evidence rather than ultimate truth.
Related code
Use findings from the code changes and design documents to identify related code - code that touches or integrates with the changes, as well as code which follows similar patterns or is architectually related. This will give you a better understanding of whether the current code is following established practices, whether there are duplicated patterns, or whether architectural changes or new abstractions may be necessary.
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.
- 10d ago First seen · 117 lines · 13 tokens per session scan A b492acc46905
refine is a skill published in the GitHub repository magnitudedev/magnitude (4,229 stars, last pushed today), licensed Apache-2.0. It adds 13 tokens to every session and 1,162 once invoked, about $0.0001 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
autoreview
Pre-commit/ship code review: Codex default; optional Claude or Pi.
rework-rate
Measure and interpret PR rework rate — the emerging 5th DORA metric.
omh-code-review
This is a Hermes-native code-review workflow skill.
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
code-reviewer
Code review specialist focused on patterns, bugs, security, and performance.
agent-teams-simplify-and-harden
Implementation + audit loop using parallel agent teams with structured simplify, harden, and document passes. Spawns implementation agents to do the work, then audit agents to find complexity, security gaps, and spec deviations, then loops until code compiles cleanly, all tests pass, and auditors find zero issues or…