asadullah48

AI Sdr OS — Data skill for Claude Code

Data community

How to install AI Sdr OS

This entry records only its repository, not the path inside it, so there is no exact command to give. Open asadullah48/ai-sdr-os and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Sdr OS does

Production-grade agentic AI SDR (sales development rep) system: a 7-stage pipeline (Find, Score, Research, Personalise, Contact, Follow Up, Route Replies) with a human-in-the-loop approval gate, constitutional-AI guardrails, and a Next.js pipeline board. FastAPI + Postgres backend, Claude API for research/personalisation/reply classification.

Alternatives in Data

Skill Production Pipeline — Claude-skills — Effective: 2026-03-07 Applies to ALL new skills, improvements, and deployments 5.3k ★
Agentlas Storm — Run a force-robust Stormbreaker loop — route to real agents, execute a verified pipeline to completion 1.1k ★
Claude Deep Research Skill — Enterprise-grade deep research skill for Claude Code with 8-phase pipeline, source credibility scoring, and au 1k ★

README

AI SDR Operating System

A production-grade agentic system that automates the seven jobs of a sales development rep - Find, Score, Research, Personalise, Contact, Follow Up, Route Replies - into one pipeline with a human-in-the-loop approval gate before anything leaves the building.

Full design and rationale: [SPEC-AI-SDR-OS.md](./SPEC-AI-SDR-OS.md). This README tracks what is actually built, in what order, and what is deliberately deferred - it is a supplement to the spec, not a replacement.

Status: Session 4 (Validation) complete - v1.0.0

Area	Status
Repo scaffold (backend/frontend/docker-compose)	Done
Postgres schema + Alembic migrations (every table in spec §5, plus `outbox` - see "Known gaps")	Done
FastAPI skeleton, health check, JWT auth (seeded user), CORS	Done
Stage 1 - Find (rule-based ICP filter, mock CSV data source)	Done
Stage 2 - Score (weighted rubric, cutoff, demotion)	Done - see "Known gaps"
Stage 3 - Research (Claude API, `prompts/research.md`)	Done
Stage 4 - Personalise (Claude API, `prompts/personalise.md`)	Done
Guardrail layer (BLOCK/FLAG, deterministic pattern-matching)	Done
Approval queue (enforced at the service layer before Stage 5)	Done - auto-approves by default; see "Known gaps"
Stage 5 - Contact (mock send adapter, writes to `outbox`)	Done
Stage 6 - Follow Up (cadence-driven, cycles through the sequence's channel order)	Done
Stage 7 - Route Replies (Claude classification, rule-based owner routing)	Done
Pipeline orchestrator (Find→Score→Research→Personalise)	Done
Full 7-stage integration test	Done
End-to-end final-state-distribution test (spec §10 Session 4)	Done
Frontend: pipeline board, prospect detail, approval queue view	Done
APScheduler cadence check	Done - off by default (`ENABLE_SCHEDULER`), on in docker-compose
CI (GitHub Actions): lint + test on push	Done - green on GitHub Actions