Distill stored observations for one session into repo-local memory candidates. Org/global promotion still requires explicit human review.
AI agents use kage_distill to create or update resources in Kage — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Kage environment.
This tool creates or modifies data (distilled memory candidates stored as git-tracked JSON per server description) in a reversible manner. It does not execute arbitrary code, delete data, move money, or bypass human review for org/global changes. The write is scoped to repo-local candidates and requires human review before wider promotion, making it Write rather than Execute or Destructive.
From the tool's definition Tool description states it 'Distill[es] stored observations for one session into repo-local memory candidates', which creates or modifies memory data structures.
Documented attack patterns abuse exactly the kind of access kage_distill gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Kage, and nothing reaches the server without passing your rules. This is the rule we recommend for kage_distill:
{
"version": "1",
"default": "deny",
"tools": {
"kage_distill": {
"limits": [
{
"counter": "kage_distill_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}kage_distill stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Distill stored observations for one session into repo-local memory candidates. Org/global promotion still requires explicit human review. It is categorised as a Write tool in the Kage MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Kage MCP server in PolicyLayer and add a rule for kage_distill: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Kage. Nothing to install.
kage_distill is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the kage_distill rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for kage_distill. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
kage_distill is provided by the Kage MCP server (@kage-core/kage-graph-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Kage, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
62 Kage tools catalogued and risk-classified — across an index of 43,000+ MCP servers.