Convert LLM task execution into evidence-driven workflows with strict closure.
npx skills add https://github.com/menghun3-cn/linshigong --skill linshigong
Please help me install this Agent Skill. Skill: linshigong Source: https://github.com/menghun3-cn/linshigong/tree/main/linshigong Command: npx skills add https://github.com/menghun3-cn/linshigong --skill linshigong
linshigong solves inconsistent agent outputs by enforcing “temporary worker” thinking: uncover what’s missing, tighten granularity, and close the loop with validation evidence rather than claims.
Use case examples:
Ask your AI assistant to follow the linshigong behavior protocol, then solve the current task with explicit assumptions and tool-verified results, including the required status/display output format.
High-intent search queries and answers about installing and using this skill.
Enforcing evidence-based reasoning in LLM agent workflows requires behavioral guardrails that mandate explicit assumption validation, tool-verified hypotheses, and measurable success criteria to prevent unproven task completion.
Preventing agents from claiming completion without proof requires an end-to-end closure protocol that mandates searching original context, validating hypotheses with tools, and reporting verification outputs before marking a task complete.
Debugging a regression with AI agents involves reproducing the issue, searching related patterns, validating assumptions with tools, and producing evidence-based confirmation of the fix rather than relying on unverified claims.
Implementing end-to-end delivery guardrails for multi-step coding tasks involves defining precise minimal edits, preventing over-engineering, setting success standards, running checks, and closing impacts across the chain with verification evidence.
The linshigong protocol requires no external dependencies, relying solely on following its behavior protocol in SKILL.md and using referenced display and supporting documentation to report progress and verification evidence.
Reasoning guardrails for agent workflow validation should be used in coding, debugging, refactoring, and multi-step delivery scenarios where agents must validate assumptions and avoid unverified completion outcomes.
LerianStudioGenerate a complete PRD with problem discovery, user stories, and success metrics.
0xUrsanomicsEnforce a mandatory reply tool call at the end of every interactive chat turn.
SkastVnTManage tool IDs, bindings, and auto-trigger rules in the chatbot search cascade.
creeveliuAnalyze product or organizational problems with Marxist structural analysis.
