Working Notes on Agent Systems/Brad Zhang

@teach_fireworks / X longform

After the launch of Claude Skill, Claude took advantage of the situation to optimize the original

After the launch of Claude Skill, Claude took advantage of the situation to optimize the original shortcomings and limitations of MCP: A large number of inefficient...

November 10, 2025 · 1 min read

After the launch of Claude Skill, Claude took advantage of the situation to optimize the original
Figure 1 / source image

After the launch of Claude Skill, Claude took advantage of the situation to optimize the original shortcomings and limitations of MCP: A large number of inefficient llm requests consume a lot of tokens and the accuracy cannot be guaranteed.

The essential reason is that the mcp client at that time essentially used the prompt word engineering idea to guide the interaction between mcp and llm, while claude skill It is based on the idea of context engineering to interact with large models.

One of the most effective techniques is to offload the context to the file system.

The context here can be a specific executable scirpt file.

This is also This is the essential reason why the first mcp client can save a huge amount of tokens and greatly improve the success rate of mcp.

Visual summary

Article argument map

Generated from the post's content graph

FORMATTOPICCAPABILITYMARKETcoverscoverscoverscoverssignalssignalssignalssignalsFORMATlongform noteTOPICharness engineeringTOPICai workbenchesTOPICmemoryTOPICtoolingCAPABILITYagent workflowCAPABILITYharness engineeringCAPABILITYAI-native workbenchCAPABILITYdeveloper toolingCAPABILITYtechnical distribution
Mermaid outline
flowchart LR
  format-long_post["longform note"]
  topic-harness-engineering["harness engineering"]
  topic-ai-workbenches["ai workbenches"]
  topic-memory["memory"]
  topic-tooling["tooling"]
  capability-agent-workflow["agent workflow"]
  capability-harness-engineering["harness engineering"]
  capability-ai-native-workbench["AI-native workbench"]
  capability-developer-tooling["developer tooling"]
  capability-technical-distribution["technical distribution"]
  format-long_post -->|covers| topic-harness-engineering
  format-long_post -->|covers| topic-ai-workbenches
  format-long_post -->|covers| topic-memory
  format-long_post -->|covers| topic-tooling
  format-long_post -->|signals| capability-agent-workflow
  format-long_post -->|signals| capability-harness-engineering
  format-long_post -->|signals| capability-ai-native-workbench
  format-long_post -->|signals| capability-developer-tooling

Visual structure

Essay structure map

Built from summary and key paragraph positions

After the launch of Claude Skill, Claude took advantage of the situation to optimize...THESISAfter the launch ofClaude Skill, Claudetook advantage of thesituation to optimizeSIGNALAfter the launch ofClaude Skill, Claudetook advantage of thesituation to optimizeOPERATOROne of the mosteffective techniquesis to offload thecontext to the fileIMPLICATIONThis is also This isthe essential reasonwhy the first mcpclient can save a huge
Mermaid outline
flowchart LR
  thesis["After the launch of Claude Skill, Claude took advantage of the situation to optimize the original shortcomi..."]
  signal["After the launch of Claude Skill, Claude took advantage of the situation to optimize the original shortcomi..."]
  operator["One of the most effective techniques is to offload the context to the file system."]
  implication["This is also This is the essential reason why the first mcp client can save a huge amount of tokens and gre..."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post