Tool reference · Errors, rules & self-improvement
self_rules_context
Returns the active rules that apply to the current project, optionally narrowed to rules for the current task phase.
local stdio server read-only
Caution. Not purely read-only: each returned rule has its fire_count incremented, which feeds its success rate.
When to use
- At the start of a session, to load the rules the agent should follow.
- You are entering a specific phase (plan, build, reflect) and only want rules for it.
- You are working on a task that matches certain error categories and want their rules too.
Parameters
| Name | Type | Default | Description |
|---|---|---|---|
project | string | "general" | — |
categories | string[] | — | Error categories relevant to current task |
phase | "van" | "plan" | "creative" | "build" | "reflect" | "archive" | — | Optional: lazy-load only rules relevant to this phase (core + phase-specific). Omit to get all rules. |
Example
Arguments
{
"project": "my-api",
"phase": "build"
}Result shape
{
"rules_count": 1,
"rules": [
{
"id": 19,
"content": "Always run the migration in a transaction on staging first.",
"category": "config_error",
"scope": "project:my-api",
"priority": 7,
"success_rate": 0.8,
"tags": "[\"phase:build\"]"
}
],
"phase_filter": "build"
}Returns at most 20 rules: global ones plus those scoped to the project or to the given categories.
Values are illustrative; the keys follow the server's handler. MCP clients receive the result as JSON text content.
Server description
The description the server sends to your agent in tools/list, captured from the v14.7.0 source:
Get active behavioral rules for current session. Call at SESSION START to load rules. Returns rules filtered by project and scope. v8.0: pass `phase` to lazy-load rules relevant to current task phase — core rules (no phase tag) + rules tagged phase:<X>. Cuts prompt tokens ~70%. After task completion, rate rules: self_rules(action='rate', id=X, success=true/false).