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/hkust-knowcomp/deeprefine-skill/claude_skillnpx skills add HKUST-KnowComp/DeepRefine-Skill --skill claude_skillgit clone --depth 1 https://github.com/HKUST-KnowComp/DeepRefine-SkillWrote 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/hkust-knowcomp/deeprefine-skill/claude_skill)<a href="https://agentmods.dev/skills/hkust-knowcomp/deeprefine-skill/claude_skill"><img src="https://agentmods.dev/badge/skills/hkust-knowcomp/deeprefine-skill/claude_skill.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 5.1 | $0.00068 | $0.01318 |
| Opus 5 | $0.00034 | $0.00659 |
| Sonnet 5 | $0.00014 | $0.00264 |
| Haiku 4.5 | $0.00007 | $0.00132 |
Grade A, and why
deeprefine 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 6d 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
2 near-identical copies found in the catalogue:
- deeprefine — 92% identical, 34 lines differ
- deeprefine — 89% identical, 47 lines differ
How it starts
The opening of the file, as written. The whole thing — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DeepRefine - Claude Code Adapter
This file is the Claude Code-specific entrypoint. It keeps the platform rules small and loads longer DeepRefine procedure details only when needed:
- Full workflow, queue selection, refinement branch logic, and review rules: references/deeprefine-workflow.md
- Verbatim judgement, abduction, and refinement prompts: references/llm-prompts.md
- Checklist, command sequence, trace schema, paths, and CLI mode: references/trace-and-commands.md
Do not reimplement or shorten the algorithm from memory. Load the relevant reference file before executing that part of the workflow.
Claude Code Invocation
Trigger this skill when the user:
- explicitly invokes
/deeprefine; - asks to refine, improve, diagnose, repair, inspect, or review a Graphify knowledge graph;
- asks to apply a previously reviewed DeepRefine refinement.
Run from the knowledge-base project root, where graphify-out/graph.json
exists. If the user is planning or asking how DeepRefine works, explain the
workflow and do not mutate files.
If deeprefine is unavailable, tell the user to install it:
pip install deeprefine-cli
For source development:
pip install -e /path/to/DeepRefine-Skill
Hard Safety Policy
A normal /deeprefine invocation is dry-run only and MUST NEVER call
deeprefine apply.
The default workflow must stop after:
deeprefine loop validatedeeprefine review- showing the proposed actions and HIGH/MEDIUM/LOW review report to the user
Then ask for explicit approval.
Only if the user's next message explicitly says to approve/apply/write the graph may you run:
deeprefine apply --trace-file ... --refinement-file ...
deeprefine loop finish --trace-file ... --refinement-file ...
Do not treat any of these as approval:
- generation of a
<refinement>block; - a valid
loop_trace_<query_id>.json; - a prior user message;
- a successful
deeprefine review.
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.
- 6d ago First seen · 169 lines · 68 tokens per session scan A 8e1927da2c59
deeprefine is a skill published in the GitHub repository HKUST-KnowComp/DeepRefine-Skill (92 stars, last pushed 13d ago), licensed MIT. It adds 68 tokens to every session and 1,318 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 2026-08-30.
Other skills, from other repositories
llm-wiki
Build and maintain a persistent, interlinked Obsidian-compatible markdown wiki using Karpathy's LLM Wiki pattern. Extension-backed with auto-generated metadata, guardrails, and 14 custom tools (+3 opt-in agent-trajectory tools).
binder-import
Import external data into a Binder workspace. Handles CSV, JSON, YAML, Markdown files, and directories of Markdown. Use when asked to "import data", "load records from a file", "ingest documents", "migrate data into binder", or bulk-create records from an external source.
customer_support_agent
You have access to Leeroopedia, a curated ML/AI knowledge base, via MCP tools. These are real MCP tools registered in your environment -- call them directly like any other tool. They contain framework-specific docs, code examples, API references, and best practices.
ml_inference_optimization
This document describes the Leeroopedia MCP tools available during the with-KB benchmark run. It is a standalone reference and is not fed to the agents automatically.
leeroopedia-mcp
Use Leeroopedia MCP to fetch grounded ML/AI best practices, build and review ML plans, debug failures, verify code/math correctness, and expand KB citations via getpage.
llm_post_training
You have access to the Leeroopedia MCP tools. Use them throughout this pipeline to make informed decisions. Specifically.