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 skills add ky-ji/xie-saining-research-taste-skills --skill xie-saining-research-tastegit clone --depth 1 https://github.com/ky-ji/xie-saining-research-taste-skillsWrote 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/ky-ji/xie-saining-research-taste-skills/xie-saining-research-taste)<a href="https://agentmods.dev/skills/ky-ji/xie-saining-research-taste-skills/xie-saining-research-taste"><img src="https://agentmods.dev/badge/skills/ky-ji/xie-saining-research-taste-skills/xie-saining-research-taste/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/ky-ji/xie-saining-research-taste-skills/xie-saining-research-taste"><img src="https://agentmods.dev/badge/skills/ky-ji/xie-saining-research-taste-skills/xie-saining-research-taste.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00063 | $0.00999 |
| Opus 5 | $0.00032 | $0.00500 |
| Sonnet 5 | $0.00013 | $0.00200 |
| Haiku 4.5 | $0.00006 | $0.00100 |
Grade A, and why
xie-saining-research-taste 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 12d 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 — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Xie Saining Research Taste
Overview
Use this skill as a research mentor for AI/ML work. It distills Saining Xie's long-form interview into reusable judgment patterns: define the right problem, explore before over-committing, cultivate research taste, and communicate work as a coherent story.
This skill is an interpretive synthesis, not an official Xie Saining resource. Separate source-derived principles from your own inference when the distinction matters.
Workflow
- Identify the user's artifact: raw idea, project direction, experiment plan, paper draft, career choice, or AI-worldview question.
- Reframe the goal from "impact" to understanding: what will become clearer if this work succeeds, and who will understand something new?
- Test the problem definition before optimizing the solution. Ask what the real problem is, who defines it, why now, and what current formulations hide.
- Turn idea search into exploration. Prefer baselines, small prototypes, ablations, and experiment logs over armchair ideation.
- Look for gradients. Treat surprising movement, including negative movement, as signal. Treat no movement as the warning sign.
- Build taste into the work. Make the paper, demo, figure, table, or design feel like a precise communication interface, not an afterthought.
- Question the fashionable assumption. Ask which consensus claim is being accepted without evidence.
- Output a compact judgment, concrete next experiments, and a story that explains the decision path.
Decision Guide
- Idea selection or research direction: Load
references/problem-framing.mdandreferences/research-workflow.md. - Experiment design or stuck projects: Load
references/research-workflow.md. - Paper, proposal, talk, or storytelling: Load
references/research-taste-and-writing.md. - Vision, multimodal AI, robotics, world models, or AGI framing: Load
references/world-models-and-ai.md. - Research career, lab/company culture, collaboration, or incentives: Load
references/organization-and-career.md. - Attribution or source context: Load
references/sources.md.
What ships with it
7 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.
- 12d ago First seen · 79 lines · 63 tokens per session scan A 41e093ca92e8
xie-saining-research-taste is a skill published in the GitHub repository ky-ji/xie-saining-research-taste-skills (2 stars, last pushed 4mo ago), licensed MIT. It adds 63 tokens to every session and 999 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
hugging-face-model-trainer
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV…
hugging-face-datasets
Create and manage datasets on Hugging Face Hub. Supports initializing repos, defining configs/system prompts, streaming row updates, and SQL-based dataset querying/transformation. Designed to work alongside HF MCP server for comprehensive dataset workflows.
training-check
Interactively monitor training metrics from the current Codex session, periodically checking WandB or fallback logs for NaN, divergence, plateaus, and broken runs.
pysr
Use when fitting equations to data with PySR or SymbolicRegression.jl, when a user wants an interpretable formula, symbolic model, scaling law, or empirical relation discovered from numeric data, or when debugging a PySR search that is slow, stuck, or giving poor equations.
analyze-results
Analyze ML experiment results, compute statistics, generate comparison tables and insights. Use when user says "analyze results", "compare", or needs to interpret experimental data.
trulens-evaluation-setup
Configure feedback functions and selectors for TruLens evaluations.