AI Labo
PulseAugur coverage of AI Labo — every cluster mentioning AI Labo across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Claude Code /advisor pattern halves AI coding agent LLM calls
A benchmark study compared two AI coding agent patterns for fixing multi-module bugs. The "Solo Executor Loop" pattern involved iterative editing and retrying, while the "Advisor Blueprint Routing" pattern used an initi…
-
AI coding agents use "Save-Game" context compaction to cut tokens and boost performance
A new technique called "Save-Game Context Compaction" has been developed to address input token inflation and "lost-in-the-middle" spec drift in multi-step AI coding agents. This method, tested on a three-milestone powe…
-
AI Labo experiment shows error pruning boosts coding agent efficiency
A new experiment from AI Labo demonstrates that retaining failed code attempts and error traces in an LLM's chat history can lead to "context poisoning," causing the model to anchor on incorrect solutions. By implementi…
-
AI coding agents slash token use with structured Pytest log filtering
A new method for processing Pytest logs significantly reduces the number of input tokens required by AI coding agents. By filtering out passing tests and ANSI escape codes, and instead providing a structured JSON summar…
-
AI coding agents perform better with interface skeletons than full dependency files
A new benchmark study reveals that providing AI coding agents with full dependency files, rather than just interface skeletons, can significantly reduce accuracy. The experiment, conducted within isolated Proxmox VMs, s…
-
MCP Tools: Managing Quantity, Security, and Context for AI Agents
The Model Context Protocol (MCP) is facing challenges related to the number and management of tools it integrates. Too many tools can increase prompt size, leading to higher costs and potential confusion for AI agents. …
-
AI coding agents improve efficiency with "Episodic Gotchas Memory"
A new benchmark series from AI Labo, titled Experiment E20, has demonstrated the effectiveness of an "Episodic Gotchas Memory" (MEMORY.md) for AI coding agents. This method involves injecting a small, distilled memory o…