t0ddc3by/claude-for-customer-success

This repository is a reference implementation of what an AI built workflow for Customer Success might look like when it is grounded in domain knowledge rather than generic chat. The author is clear that it is not a deployment ready product, only a working demonstration. It packages six plugins that target every function inside a typical Customer Success organization: customer success managers, CS operations, renewals, onboarding, revenue operations for CS, and a small infrastructure plugin called auq resilience that adds fallback handling for interactive prompts. The project ships as a Claude Code or Claude Cowork plugin and can also be deployed through the Claude Managed Agents API behind your own workflow engine. The same system prompts and skills are used in either path, so you choose where the agent runs. The README points new readers to a QUICKSTART file that claims a 60 second install. Distribution is via pre built .plugin files inside the dist folder, and each plugin can be installed on its own. The README lists what each plugin covers. The csm plugin handles account research, QBR prep, success plans, health review, risk flags, and value statements. The cs ops plugin focuses on portfolio analytics, segmentation, capacity planning, and data quality. The renewals plugin covers renewal forecasting, expansion signals, churn analysis, and contract review. The onboarding plugin focuses on kickoff prep, milestone tracking, and handoff documentation. The rev ops plugin is the largest, with 34 of the 81 total skills, and covers CRM data quality, forecasting, quota and incentive planning, deal desk work, and revenue leakage scanning. A large section of the document argues that the methodology behind the skills is what makes the suite useful, not the agents themselves. The author describes the customer lifecycle as a complex adaptive system where value moves from theoretical at sale, to demonstrated through adoption, to networked over time. The suite encodes the SuccessCOACHING methodology while staying tool agnostic, meaning it does not require a specific CRM or success platform. The README addresses three audiences explicitly: CS leaders looking for a reference of what AI enabled CS can cover, operators and developers who want to extend it, and architects evaluating it as a pattern for building multi plugin suites. License is Apache 2.0. The full README is longer than what was shown.