AITOPAI热报

#MCP and tools

1092 stories · 134 in 7 days · 463 in 30 days · Topic watch →
agents 400Claude 95coding assistants 91Claude Code 73Anthropic 54LangChain 52prompt engineering 46ChatGPT 42
Oct 17
ProductsBuildersSelected01:32
LangSmith AI Search Matches Agent Traces to Descriptions

Agent debugging usually means grepping raw traces; this inverts the workflow by letting teams describe the failure in natural language and get matching trajectories back. Engineering teams running LLM apps in production can use it to find specific behaviours without building custom query tooling. Compared with generic observability search, the two-stage filter-then-verify design reduces the model's workload and latency.

#LangSmith#LangChain#observability#agents
Official accountLangChain@LangChainAISource
ProductsBuildersSelected01:32
LangChain AI Search analyzes agent trajectories in fixed token budget

Most of an agent's token spend sits in tool output, so debugging or evaluating behavior means targeting that slice rather than full transcripts. Teams building coding agents or observability pipelines can use this ratio to size context budgets and design trajectory search over logs. The numbers come from Open SWE in production use, giving a concrete baseline for agent cost analysis.

#LangChain#Open SWE#agents#MCP and tools
Official accountLangChain@LangChainAISource
Oct 16
PapersPolicyMulti-sourceSelected11:27
MCP-FSD Uncovers Filesystem Capability Laundering in MCP Servers

Client-side tool approval in MCP grants the whole server process broad filesystem access, so a tool approved for one operation can exfiltrate credentials the contract never mentions. Security engineers running local agent stacks should audit installed MCP servers and enforce sandbox boundaries rather than trusting declared tool schemas. The two CVEs in Anthropic's Git MCP Server make this a concrete patch item for anyone using that server.

#MCP and tools#AI safety#MCP#Anthropic
Event
arXiv: Anthropic@Zeqian Huang et al. (5)Source
ProductsBuildersSelected10:06
Magpie plugin system opens every layer to community packages

The notable shift is that a single npm package can now intercept traffic, allocate requests across accounts and models, and serve tools to every connected agent — turning Magpie from a fixed client into a programmable gateway. Developers building multi-account or multi-model setups can try the three sample plugins or write their own against the community repo. It parallels plugin ecosystems like Obsidian's, but aimed at LLM routing and agent infrastructure.

#Magpie#MCP and tools#coding assistants
Geek@geekbbSource
PapersBuildersSelected09:29
ChainGraph: Neo4j-Based Exploit-Chain Reasoning for Penetration-Testing Agents

The result suggests the weakness in agentic security testing lies in flat context windows rather than model capability, so externalizing memory to a graph is a fix practitioners can apply today. Security engineers and red-team builders can wire Neo4j into existing MCP-compatible agents like Claude Code and Codex without custom tooling. The findings are illustrative single-run measurements, so treat them as a design pattern to replicate rather than a benchmark.

#ChainGraph#Neo4j#Claude Code#MCP and tools
arXiv: OpenAI@Joas Antonio dos Santos BarbosaSource