Tool execution
From-scratch agent runtime with task graphs, multi-index RAG, memory, typed tools, reusable skills, handoffs, traces, and smoke tests.
npx -y skills add QiRebecca/Agentdemo --skill tool_executionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
0.4 KB, 78 tokens by cl100k_base, as published. Nobody here has run it
Tool Execution
Purpose
Call registered typed tools and record observations.
Steps
- Select the tool or tool sequence by name.
- Validate required input keys.
- Execute the tool.
- Normalize success and error outputs.
- Record the tool call and observation in the trace.
Validation
- Tool result includes a status.
- Tool call is traced.