MCP Tools
In normal use, you should not need to choose low-level tools by hand. The agent starts with agent_workspace or guide_next_steps, while LINZA keeps the working route: check the folder, index, show a map, build review cards, learn from accepted examples, and propose the next step in preview.
Default Tool Set
By default, LINZA exposes a compact MCP tool set of 15 tools.
| Tool | Purpose |
|---|---|
agent_workspace | One entry point for diagnostics, import, map, review, teaching, growth, relations, memory search, and context export |
guide_next_steps | Show the next safe step in plain language |
index_all | Index the Markdown folder into .linza/linza.db |
search | Semantic and lexical search |
read_file | Read an exact file from the working folder |
get_stats | Quick local database counters |
scan_vault | Read-only folder diagnostic |
build_review_apply_queue | Build review cards with stable rq-* IDs |
approve_review_queue_items | Preview or apply selected cards |
list_approved_items | Show accepted local records |
explain_node | Explain one node: links, bridges, context |
explain_relationship | Explain a possible relation between two nodes |
who_depends | Show dependencies and neighbors |
show_flow | Find a route or flow between nodes |
create_context_pack | Build a compact context package for an agent |
agent_workspace
This is the main agent entry point. It contains working actions such as doctor, map, review, teach, grow, connect, and calibr. You should see a short status, evidence, next steps, and a clear explanation of what could change; raw JSON stays in the technical layer.
Review And Apply
LINZA uses stable rq-* IDs. That makes application exact: only cards you have seen should be accepted. If IDs do not match, stop and rebuild the preview.
Advanced Tool Set
For development and debugging, advanced tools can be enabled:
$env:LINZA_TOOL_SURFACE="advanced"Normal work does not need this. Keep the path short: diagnostics, index, map, review cards, teaching, and the next preview.