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/skillnerds/xskill/xskill-devnpx skills add SkillNerds/xskill --skill xskill-devgit clone --depth 1 https://github.com/SkillNerds/xskillWhat 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 | $0.00050 | $0.00695 |
| Opus 5 | $0.00025 | $0.00347 |
| Sonnet 5 | $0.00010 | $0.00139 |
| Haiku 4.5 | $0.00005 | $0.00069 |
Grade A, and why
xskill-dev 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 2d 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.
What it actually says
Developing xskill
工具描述从哪来
Agno 框架直接把 @tool 装饰的 Python 函数 docstring 当作发给模型的工具描述，
参数说明也从 docstring 里解析。也就是说 docstring 就是模型上下文的一部分：
- 写给模型看，不写内部实现细节（内部细节放代码注释）。
- 系统提示词里不要再抄一遍工具清单，框架会自动注入 schema，抄了会漂移。
- generate 代理的轨迹工具在
src/xskill/agents/traj_tools.py， 通用工具在src/xskill/agents/agent_tools.py， wiki 工具在src/xskill/agents/llm_wiki.py。
改完 docstring 必跑：dump_schema
改任何工具的 docstring 或签名之后，跑一次导出脚本，看 Agno 实际生成的 schema 是什么样，确认模型看到的和你想的一致：
/home/admin/xskill/.venv/bin/python \
scratch/standalone-generate/tool_surface/dump_schema.py
输出写在同目录 SCHEMA.txt。对照检查：
- 每个工具的 description 是否完整、有没有被截断或混进实现细节；
- 参数名、类型、必填项是否与函数签名一致；
- 新增或删除工具后，工具总数是否符合预期（generate 面当前是 16 个）。
SCHEMA.txt 可以进 code review diff，reviewer 能直接看到模型侧的变化。
上下文预算的流式陷阱
_wrap_with_context_mgmt 只包 model.invoke。任何用 stream=True 跑 agent
的路径都会走 invoke_stream，完全绕过 compact、spill 和超长兜底，模型跑在
后端原生窗口里（DeepSeek 是 1M），而且日志里一条 CONTEXT 事件都不会有。
产品 GenerateAgent 用非流式 agent.run() 所以没事；写实验脚本、demo、
新 agent 入口时必须非流式，或先给 invoke_stream 补包装。判断预算机制
是否真在跑，看 agent.log 里有没有 CONTEXT 事件（Compacted context、
Spilled、Compact was not needed 任意一种）。
llm_cfg 里开剪裁的键名是 enable_spill，不是 spill。
相关材料
- 工具面设计与取舍：
docs/plans/2026-08-27-generate-tool-surface.md - 独立实验台（Phoenix 观测、变体对比）：
scratch/standalone-generate/， 入口run_experiment.py，product变体加载产品 traj_tools 与产品 SYSTEM_PROMPT，是验证产品行为的首选变体。
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.
- 2d ago First seen · 55 lines · 50 tokens per session scan A b3bc5962312c
xskill-dev is a skill published in the GitHub repository SkillNerds/xskill (121 stars, last pushed 3d ago), licensed MIT. It adds 50 tokens to every session and 695 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
reversi-master
One-click install + model switch:Claude Code,Codex CLI (OpenAI), Grok Build (xAI), DeepSeek Harness, Kimi Code (Moonshot) ,Qwen Code,Aider,OpenCode,MiMo Code (Xiaomi),ZCode (Z.AI),OpenClaw,Pi,OpenScience,Vibe-Trading,Claude Desktop (3P profile),ChatGPT desktop,OpenCode Desktop.
Trash Talk
AI becomes provocative and mocks the opponent during gameplay.
tmux
Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output.
summarize
Summarize or extract text/transcripts from URLs, podcasts, and local files (great fallback for “transcribe this YouTube/video”).
add-matrix
Add Matrix channel integration via Chat SDK. Works with any Matrix homeserver.
add-resend
Add Resend (email) channel integration via Chat SDK.