Detailed Analysis
A developer in India has successfully automated the end-to-end filing of eight Right to Information (RTI) applications across multiple central institutions in Karnataka using a combination of Anthropic's Claude Code and Microsoft's Playwright CLI. The workflow handled the full submission lifecycle — including form filling, UPI payment processing, and confirmation tracking — without manual intervention. The developer has published the setup and workflow publicly on GitHub, inviting community contributions and improvements. The project represents a practical, real-world deployment of agentic AI tooling applied to civic participation, demonstrating that the same automation frameworks used in software testing and web scraping can be redirected toward legally significant government interactions.
The technical stack at the core of this project pairs Claude Code, Anthropic's command-line coding agent, with Playwright CLI, a standalone npm package released by Microsoft in early 2026 specifically designed for AI agent-driven browser automation. Unlike Playwright's Model Context Protocol (MCP) integration — which passes verbose JSON payloads and full accessibility trees into an agent's context window — the CLI variant uses terse, Bash-like shell commands such as `playwright-cli fill`, `playwright-cli click`, and `playwright-cli eval`. This architecture dramatically reduces token consumption, making it well-suited for repetitive, multi-step form workflows like those found on India's RTI Online Portal. Claude Code interprets natural language instructions and translates them into sequences of CLI commands, handling navigation, field population, state extraction, and session management in an agentic loop. The inclusion of UPI payment automation is particularly noteworthy, as payment flows typically represent the most fragile and complex stage in government portal interactions.
The broader significance of this project lies in what it demonstrates about the maturity of agentic AI tooling. Web automation has existed for years via tools like Selenium and early Playwright, but the bottleneck was always the programming effort required to script each unique workflow. Claude Code eliminates much of that friction by generating command sequences from high-level intent, effectively lowering the barrier for non-engineers to build capable automation pipelines. The token-efficient CLI design also signals a meaningful architectural evolution: as AI agents are increasingly deployed for long-horizon, multi-step tasks, the overhead cost of each tool call becomes a real constraint, and purpose-built interfaces like Playwright CLI reflect an industry-wide reckoning with that reality. Comparable agentic pipelines have been documented for software QA, involving multi-agent flows covering exploration, test generation, execution, and maintenance — a structure directly applicable to bulk civic filings.
From a civic technology and policy perspective, the project occupies interesting and somewhat uncharted territory. RTI laws in India exist precisely to empower citizens to demand transparency from public institutions, and automation lowers the friction of exercising that right at scale. A single motivated citizen can now file dozens of targeted information requests with minimal effort, potentially accelerating accountability journalism, legal research, or activist campaigns. At the same time, the same capability could be misused to flood government portals with low-quality or vexatious requests, straining administrative resources. The developer's own ambivalence — describing the combination as "insanely powerful" and "a little scary" — reflects a tension that regulators and platform administrators will increasingly need to confront as agentic AI tools become accessible to general audiences. The open-source nature of the published workflow ensures the question is no longer hypothetical.
Read original article →