TalentOS

One place for your hiring pipeline. Talk to the copilot in Telegram, WhatsApp, or the web—add candidates, log interviews, track follow-ups, get a daily brief, and send emails. Same memory everywhere.

Built as an OpenClaw plugin with Mem0: structured hiring memory (candidates, interactions, promises), auto-recall and auto-capture, and 10 tools so the agent can run the pipeline without a separate ATS.

Note: This project does not use the official OpenClaw × Mem0 plugin (@mem0/openclaw-mem0). It talks to the Mem0 API directly from the TalentOS plugin. Reason: maximum customization (structured hiring schema, 10 domain tools, custom recall/capture) so that all memory logic lives in this repo and using the official plugin would mean having very less code here to review!!

---

Architecture

TalentOS runs inside OpenClaw as a plugin; Mem0 is the single source of truth. All hiring flows (add candidate, log interaction, daily brief, etc.) go through the plugin; an optional view server provides a read-only dashboard from the same Mem0 data.

flowchart TB
  subgraph entry["Entry points"]
    Chat[Telegram / WhatsApp / Web]
    Webhook[Webhook]
    Cron[Cron]
  end

  subgraph openclaw["OpenClaw"]
    Gateway[Gateway]
    subgraph plugin["TalentOS Plugin"]
      Hooks[Hooks: auto-recall + auto-capture]
      Skill[Skill: SKILL.md]
      Tools[10 tools]
    end
  end

  Mem0[(Mem0)]
  ViewServer[View server]
  Dashboard[Dashboard /view]

  entry --> Gateway
  Gateway --> plugin
  Hooks --> Mem0
  Tools --> Mem0
  ViewServer -->|read-only| Mem0
  Dashboard --> ViewServer

Loading

→ Full architecture — request flow, hooks, tools, memory schema, cron, webhook.

---

What you can do

• Add candidates: Name, role, contact, referrer. Stored in Mem0, not lost in chat.
• Log interactions: Screening notes, stage updates, strengths/concerns. Timeline per candidate.
• Track follow-ups: “Schedule call with X tomorrow 3pm.” Surfaces in the daily brief and follow-ups list.
• Query pipeline: “Who’s in the pipeline for Staff Engineer?” “Who did Ankit refer?” “Pipeline health?” “Any candidates who expressed concerns about compensation?”
• Daily brief: Proactive summary: follow-ups due, at-risk candidates, pipeline snapshot. On demand or on a schedule (cron).
• Send email: “Send [candidate] an email to schedule a call.” Uses Resend; one message, no copy-paste.
• Same memory, all channels: Add in Telegram, ask in WhatsApp or from a webhook. One user_id, one pipeline.

---

Quick start

1. Repo: npm install. Copy .env.example → .env, set MEM0_API_KEY (for the optional dashboard).
2. OpenClaw: Install the TalentOS plugin into ~/.openclaw/extensions/talentos-hiring, set apiKey (or mem0ApiKey) and userId in config, allow the TalentOS’ tools, restart the gateway. → Setup
3. Optional dashboard: npm run dev → http://localhost:3010/view (read-only pipeline view).

---

Repo layout

Part	What it does
openclaw-plugin/	Skill, auto-recall/capture hooks, 10 tools + optional email. All hiring runs here; no separate API.
View server	Optional. GET /view, GET /api/v1/talent/view — dashboard and read-only API from Mem0.

Scripts

Command	Purpose
npm run dev	Run view server (tsx).
npm run build	Compile TypeScript.
npm start	Run view server (node).
npm run test	Unit tests.
npm run smoke	Smoke test (server must be running).

---

Screenshots

Screenshots are in docs/screenshots/. Some flows use two images to show the full conversation or view.

Add candidate — Add a candidate in chat → agent confirmation.

Full pipeline — "Who's in the pipeline for [role]?" → table of candidates.

Email to schedule call — Send email → agent sends via Resend, confirmation.

Follow-ups or daily brief — "What follow-ups are due?" or "Give me the daily brief."

Memory dashboard — http://localhost:3010/view

---

Docs

• ARCHITECTURE — Diagram, tools, hooks, skill, view server, cron, webhook, memory schema.
• SETUP — Install, config, tools allowlist, view server, LLM (Ollama/Anthropic/OpenAI), cron, webhook, email.

License: ISC. See LICENSE, CONTRIBUTING, SECURITY.