// MCP TOOL REFERENCE

CONTEXT ENGINEERING TOOLS

4 tools from the Context Engineering MCP Server, categorised by risk level.

READ TOOLS

get_workspace_state Generate a snapshot of the current workspace state (goals, file structure, git status) read_observation Read a previously saved observation file 2/5

WRITE TOOLS

save_observation Save large tool output or observation to a file in .agent_memory 3/5

EXECUTE TOOLS

initialize_context_system Initialize the Context Engineering directory structure (.ai/skills, .agent_memory) 3/5

// FAQ

How many tools does the Context Engineering MCP server have? +

The Context Engineering MCP server exposes 4 tools across 3 categories: Read, Write, Execute.

How do I enforce policies on Context Engineering tools? +

Use Intercept, the open-source MCP proxy. Write YAML rules for each tool — rate limits, argument validation, or deny rules — then run Intercept in front of the Context Engineering server.

What risk categories do Context Engineering tools fall into? +

Context Engineering tools are categorised as Read (2), Write (1), Execute (1). Each category has a recommended default policy.

Enforce policies on Context Engineering

Open source. One binary. Zero dependencies.

npx -y @policylayer/intercept

github.com/policylayer/intercept →