ctxlens: Open-Source CLI Tool That Measures Codebase Token Usage Across AI Models

·1 views

A developer built ctxlens, a command-line utility that analyzes how many tokens a codebase consumes relative to the context limits of 23 AI models across 7 providers. The tool, inspired by the Unix 'du' disk-usage command, ranks files and directories by token count, helping developers identify which files consume the most context budget. Additional features include a diff command to estimate token savings from stripping comments, a CI flag for automated budget gating, and a watch mode for real-time monitoring. The creator developed it in early 2025 while working with local Ollama models and Codex, where hitting context limits was a frequent problem. Though the author notes reduced personal use since frontier models now offer larger context windows, the tool remains potentially useful for developers running local LLMs with limited context capacity.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Event-Driven Architecture: Key Patterns, Benefits, and Common Pitfalls

Event-Driven Architecture (EDA) is a software design pattern that enables loosely coupled, highly scalable systems by having components communicate through asynchronous events rather than direct calls. Core benefits include easier scalability, component flexibility, and a complete audit trail when events are stored as an immutable sequence. Developers are advised to separate read and write models, design events to be immutable, and implement robust error handling with retry mechanisms. Common mistakes include over-engineering solutions, introducing unnecessary complexity, and misusing events for synchronous communication. In cloud-native environments, EDA pairs well with services like AWS Lambda and Google Cloud Functions to handle real-time event spikes at scale.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Why Structured Workflows Often Outperform Autonomous AI Agents in Practice

A developer and AI builder argues that structured workflows deserve more attention than autonomous agents, which have become the default recommendation across the AI community. While agents offer flexibility, they introduce compounding complexity — more prompts, APIs, failure points, and harder-to-trace errors — that many projects do not actually require. Predictable, step-by-step workflows are easier to test, debug, monitor, and scale over time, since each component holds a clearly defined responsibility. The author also emphasizes that integration with external systems like GitHub, databases, and APIs often delivers greater business value than sophisticated but isolated agent architectures. The core advice is to first ask what the simplest workflow is that solves a problem, rather than defaulting to an autonomous agent from the outset.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Engineers Design Fail-Closed WORM Architecture for Multi-Agent AI Coordination

A development team has published details of a 10-tuple canonical envelope architecture designed to ensure reliable state management across asynchronous, multi-agent AI systems. The approach enforces five strict invariants, including transactional ingestion boundaries using PostgreSQL ACID transactions and cryptographic HMAC witness seals on all inter-agent messages. A fail-closed default principle means any unverified or unwitnessed claim is automatically placed on hold, preventing unauthorized state mutations. Chaos testing across 82 continuous integration cycles reportedly achieved a 100% pass rate for single-effect-per-event enforcement with zero duplicate state transitions. The team recommends that system architects avoid unauthenticated webhooks, isolate secrets outside cloud workspaces, and use autonomous cleanup agents to manage expired claims.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Dev team fixes RPC desync errors during Polygon hard forks with block-padding buffer

A development team encountered real-time transaction tracking failures caused by RPC endpoint desyncs during heavy network loads and blockchain upgrades. Public shared RPC nodes were dropping events due to indexing propagation lags, triggering 'invalid block range' errors when querying the chain tip. To resolve this, the team implemented a programmatic block-padding delay that queries logs three blocks behind the current chain tip, creating roughly a six-second safety buffer. This approach prevents polling of unfinalized blocks and stabilizes the asynchronous reward voucher pipeline. The fix ensures full data accuracy for user claim balances across the multi-tier infrastructure.

0 comments Read more at DEV Community