Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

xunzhi-ai-runtime — measured AI agent skill | DecimalAI

lishuangqiang/xunzhi-ai-runtime

NewSkillScoreMaintained

AI-Meeting 运行时与配置知识 Skill。用于处理面试与 Agent 链路中的限流、AI Guard、singleflight、线程池、场景绑定、长会话恢复配置和关键 `application.yaml` 语义；当需求涉及超时、重试、并发、回放、接管、缓存 TTL 或运行参数时使用。

agent-toolingby lishuangqiang

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

AI Engineer pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/lishuangqiang-xunzhi-ai-runtime/SKILL.md

decimalai skills pull lishuangqiang-xunzhi-ai-runtime --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/lishuangqiang-xunzhi-ai-runtime/SKILL.md https://app.decimal.ai/s/lishuangqiang-xunzhi-ai-runtime/SKILL.md

Raw file: /s/lishuangqiang-xunzhi-ai-runtime/SKILL.md ↗ · machine-readable summary at /s/lishuangqiang-xunzhi-ai-runtime.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

23 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 3h ago

Yes — +57% lift-21% tokens0% turns96% of whole cases pass with skill (22/23)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	-10%	0%
case-02	✗→✓	▲ Improved	-15%	0%
case-03	✗→✓	▲ Improved	-17%	0%
case-05	✗→✓	▲ Improved	-44%	0%
case-08	✗→✓	▲ Improved	-31%	0%

Showing 5 of 23 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

lishuangqiang / ai-meeting

skills/xunzhi-ai-runtime/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1Bundled 9 files · scripts

Skill contents

scripts/

references/

SKILL.md

xunzhi-ai-runtime

这层不直接定义业务结果，但决定业务链路的稳定性和并发表现。

使用顺序

再看 references/object-dictionary.md，确认运行时语义对象和配置对象的真相源。
先看 references/generated-config-index.md，确认真实配置值。
再看 references/ai-guard.md 和 references/ai-singleflight.md，判断链路受哪层保护。
涉及限流时看 references/flow-limit.md。
涉及线程池与异步任务时看 references/thread-pool.md。
涉及场景与 Agent 名称绑定时看 references/agent-binding.md。

这层管什么

面试评分、追问、提题、神态分析的超时、重试、并发上限。
同一输入的重复 AI 调用去重与结果复用。
热点请求的跟随者等待、接管和结果缓存策略。
线程池资源划分和调度资源。
业务场景到 Agent 配置的绑定。

必守约束

stage 名称必须在配置、调用点和监控语义上保持一致。
loadMode、confidence 不是展示字段，而是恢复是否可写的正式业务语义。
singleflight key 的组成不能随意改，否则会直接影响重复调用复用率。
不要只改配置不改说明文档；改完重新生成配置索引。
不要在业务层绕过 Guard 或 SingleFlight，否则等于把保护层拆掉。

参考资料

references/object-dictionary.md
references/generated-config-index.md
references/agent-binding.md
references/ai-guard.md
references/ai-singleflight.md
references/flow-limit.md
references/thread-pool.md
references/gotchas.md
scripts/extract_config_index.py

Want to see if lishuangqiang/xunzhi-ai-runtime would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.