@eos-ai/sdk

TypeScript client for the EOS Memory API — semantic search over your engineering team's knowledge graph.

API surface evolves until Loop 3. Breaking changes in minor versions until 1.0.0.

Install

npm install @eos-ai/sdk

Quick start

import { EosClient } from "@eos-ai/sdk";

const client = new EosClient({ apiKey: "eos_..." });

const r = await client.search("what changed in auth last week");
for (const node of r.data.results) {
  console.log(`[${node.node_type}] ${node.content}  score=${node.score}`);
}

API

new EosClient(opts?)

| Option | Type | Default | Description | |-----------|----------|--------------------------|--------------------------| | apiKey | string | process.env.EOS_API_KEY| EOS API key (eos_…) | | baseUrl | string | https://eos.dev | Override for local dev |

client.search(query, opts?)

| Option | Type | Default | Description | |-----------------------|-----------|---------|----------------------------------| | limit | number | 5 | Max knowledge nodes (1–20) | | includeConversations| boolean | true | Include conversation summaries |

Returns SearchResponse:

{
  data: {
    query:         string;
    results:       SearchResult[];      // ranked knowledge nodes
    conversations: ConversationResult[]; // conversation summaries
    total:         number;
  };
  meta: {
    request_id: string;
    took_ms:    number;
    cached:     boolean;
  };
  errors: null | string[];
}

Insights & recommendations

Structured, actionable engineering intelligence — the same signals EOS delivers in Slack, typed for your own dashboards. Each item carries a severity and, where relevant, a suggested_action. All return the { data, meta } envelope.

// Sprint on-time probability (Pro+)
const dc = await client.insights.deliveryConfidence();
console.log(dc.data.score, dc.data.status); // 58 "at_risk"

// Per-developer burnout (Starter+). A member key sees only its own.
const burnout = await client.insights.burnout({ developer: "dan" });

// Codebase risk (Starter+)
const hot   = await client.insights.hotspots({ days: 90, limit: 10 });
const silos = await client.insights.knowledgeRisk();

// Flow anomalies vs the team baseline (Starter+)
const anomalies = await client.insights.anomalies({ severity: "critical" });

// Who should review the stalled PRs (Starter+)
const reviewers = await client.recommendations.reviewers();
const forPr     = await client.recommendations.reviewers({ repo: "acme/api", prNumber: 482 });

| Method | Endpoint | Plan | |---|---|---| | insights.deliveryConfidence(opts?) | /v1/insights/delivery-confidence | Pro+ | | insights.burnout(opts?) | /v1/insights/burnout | Starter+ | | insights.hotspots(opts?) | /v1/insights/hotspots | Starter+ | | insights.knowledgeRisk(opts?) | /v1/insights/knowledge-risk | Starter+ | | insights.anomalies(opts?) | /v1/insights/anomalies | Starter+ | | recommendations.reviewers(opts?) | /v1/recommendations/reviewers | Starter+ |

Get an API key

eos.dev/settings/api — Starter plan or higher.