Save the steps behind a support resolution or weekly analysis. Procedural memory keeps that playbook available so your application can find the method, supply new inputs, and carry the work forward.
Procedural memory
A focused start for the design team.
Begin with design. Share a written summary before proposing a call.
Confirm the pilot approver, then prepare the invitation for the first group.
Review the completed steps and their output. Your application decides what to do next and when to refine the method.
Example customer-briefing workflow, with actions and execution supplied by the application.
Once you have a useful way to investigate a problem, give it a name and save its steps. Procedural memory makes that process available the next time the same kind of work arrives.
Give a workflow a recognizable name and description, so your application can find the right process.
Store actions and their inputs in order. A playbook keeps the intended process visible.
Pair the saved workflow with parameters for the current customer, project, or reporting period in your application.
Describe which steps depend on earlier work when defining procedures in the engine.
Use past episodes to identify recurring routines worth saving as procedures.
Read the saved steps, dependencies, and validation criteria before your application carries out the routine.
Search the stored procedures for an earlier revenue investigation. Review the method and connect its steps to the actions your application supports.
// Run server-side with credentials scoped to this Mind.
const endpoint = process.env.AKASHA_URL;
const capability = process.env.AKASHA_CAPABILITY;
if (!endpoint || !capability) throw new Error("Configure your Mind connection");
async function request(method: string, path: string, body?: unknown) {
const headers: Record<string, string> = {
"Content-Type": "application/json",
"x-akasha-capability": capability!,
};
if (process.env.AKASHA_TOKEN) {
headers.Authorization = "Bearer " + process.env.AKASHA_TOKEN;
}
const response = await fetch(new URL(path, endpoint), {
method, headers,
body: body === undefined ? undefined : JSON.stringify(body),
});
if (!response.ok) throw new Error("Mind request failed: " + response.status);
return response.json();
}
const procedures = await request("POST", "/v1/memory/procedures/search", {
query: "revenue investigation",
});
// Review the saved procedure and its dependencies.
// Your application provides the execution of its steps.Build a library of useful routines that your agent can return to as new tasks arrive.