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/deeprefinenpx skills add HKUST-KnowComp/DeepRefine-Skill --skill deeprefinegit 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/deeprefine)<a href="https://agentmods.dev/skills/hkust-knowcomp/deeprefine-skill/deeprefine"><img src="https://agentmods.dev/badge/skills/hkust-knowcomp/deeprefine-skill/deeprefine.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 | $0.00042 | $0.04737 |
| Opus 5 | $0.00021 | $0.02368 |
| Sonnet 5 | $0.00008 | $0.00947 |
| Haiku 4.5 | $0.00004 | $0.00474 |
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 4d 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.
This is a copy
100% identical to deeprefine — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 477 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DeepRefine — Agent refinement loop (strict)
Default safety policy: dry-run only
A normal /deeprefine invocation MUST NEVER call deeprefine apply.
The default /deeprefine workflow must stop after:
deeprefine loop validatedeeprefine review- showing the proposed actions and HIGH/MEDIUM/LOW review report to the user
Then ask the user for explicit approval.
Only if the user's next message explicitly says to approve/apply/write the graph may you run:
deeprefine apply --refresh-wiki --trace-file ... --refinement-file ...
deeprefine loop finish --trace-file ... --refinement-file ...
Do not treat generation of <refinement> actions as approval. Do not treat a valid trace as approval. Do not apply in the same /deeprefine turn.
You MUST implement the same control flow as DeepRefine.refine() in DeepRefine (autorefiner/src/deeprefine.py).
| Component | Agent mode | CLI deeprefine refine |
|---|---|---|
| Retrieval | graphify query + k-hop from graph.json |
FAISS retriever |
| LLM | Your session model | External API / vLLM |
| Graph writes | Dry-run proposal + deeprefine review; deeprefine apply only after user approval |
Dry-run by default; --apply persists |
FORBIDDEN (hard stop)
Do NOT:
- Run
deeprefine refine(unless the user explicitly asks for CLI/FAISS mode). - Call
deeprefine applywithout a validloop_trace_<query_id>.json(CLI will reject). - Call
deeprefine applybefore runningdeeprefine reviewand receiving explicit user approval. - Ignore LOW-confidence review warnings unless the user explicitly requests
--allow-low-confidence. - Skip any hop’s
<judge>Yes</judge>/<judge>No</judge>judgement. - Skip error abduction when
len(interaction_history) > 1. - Write
<refinement>before abduction when refinement is required. - Hand-edit
graph.jsonwith Python or ad-hoc JSON patches. - Ignore pending history and refine only one latest query when unrefined queries already exist.
- Invent a shorter pipeline (“read file → write refinement → apply”).
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.
- 4d ago First seen · 477 lines · 42 tokens per session scan A 2d3bf1019872
deeprefine is a skill published in the GitHub repository HKUST-KnowComp/DeepRefine-Skill (93 stars, last pushed 11d ago), licensed MIT. It adds 42 tokens to every session and 4,737 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to deeprefine, differing in 0 lines, and is treated as a copy.
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).
braindb
Memory recall and save. Use at every conversation start and whenever the user shares personal information, expertise, project context, preferences, or decisions worth remembering long-term.
braindb-agent
Persistent memory across sessions via the BrainDB agent. Use at conversation start and whenever you need to recall what you know about the user or save new information to long-term memory.
braindb-custom-profile
How to author a BrainDB custom profile — prompt add/replace fragments and an optional keyless ingestor — that shapes wiki naming/structure and feeds a custom ingestion source, with zero effect on defaults when inactive.
mindbase
Maintain a persistent AI-tended research wiki. Activates when the user wants to ingest a source (PDF, URL, article, paste), capture a thought, build the wiki context, query what's already known, lint for orphans/contradictions, or maintain page conventions. Phrases include "add this to my wiki", "today I decided…"…
run-wenlan-app
Build, launch, screenshot, and drive the wenlan-app Tauri desktop app in dev mode on macOS (WKWebView) or Windows (WebView2). Use when asked to run or start the app, verify a UI change in the real running app (not just tests), or take a screenshot of it.