Context Lens
A developer tool for precisely measuring the actual token consumption of a code repository before it is fed into an AI model. It supports comparison between raw tokens and mapped tokens, validation using real tokenizers, fine-grained module-level analysis, and verification of AI-generated answers against repository content.
Timeline (last 90 days)
The core function of Context Lens is to turn the AI token cost consumed by a codebase from a guess into a measurable fact, running commands against cleaned snapshots with each command exposed publicly as a form.
Context Lens shows a comparison of raw tokens versus mapped tokens for each repository and validates the results against a real o200k tokenizer run.
Context Lens allows drill-down to reveal which modules consume the most context, how different mapping strategies change budget allocation, and what each agent's startup file requests before work begins.
Context Lens provides a verified disk for cross-checking answers or execution plans given by AI against the actual content of the repository.
Context Lens has a built-in playground targeting a real SIGMAP MCP server, providing a full tool list and schema-driven invocation.
All Context Lens operations run on cleaned head snapshots, with sensitive information stripped at input and output results scanned again for leaks on the return path.
All Facts (6)
The core function of Context Lens is to turn the AI token cost consumed by a codebase from a guess into a measurable fact, running commands against cleaned snapshots with each command exposed publicly as a form.
50%UnverifiedContext Lens shows a comparison of raw tokens versus mapped tokens for each repository and validates the results against a real o200k tokenizer run.
50%UnverifiedContext Lens allows drill-down to reveal which modules consume the most context, how different mapping strategies change budget allocation, and what each agent's startup file requests before work begins.
50%UnverifiedContext Lens provides a verified disk for cross-checking answers or execution plans given by AI against the actual content of the repository.
50%UnverifiedContext Lens has a built-in playground targeting a real SIGMAP MCP server, providing a full tool list and schema-driven invocation.
50%UnverifiedAll Context Lens operations run on cleaned head snapshots, with sensitive information stripped at input and output results scanned again for leaks on the return path.
50%