Armature
Armature helps software tool makers get discovered, chosen, and reliably used by AI coding agents like Claude Code
The Problem
Teams that build developer tools, SaaS products, or APIs increasingly depend on being recommended and installed by coding agents working inside a user's repository, not by humans searching the web or reading marketing pages. Today they have no visibility into how an agent decides which tool to pick in a given stack, and traditional SEO or GEO tactics target search engines and chat answers rather than this in-repository decision moment. Without a way to test and measure agent behavior, a tool can be ignored or replaced by a competitor an agent installs instead, and even once installed, sessions can silently fail or break without the team knowing. The result is lost adoption and support burden that the team cannot diagnose because the failures happen inside agent sessions they cannot see.
The Solution
Armature runs two connected offerings. On discoverability, a growth engineer works with the team against a panel of 16 repositories built to mirror real-world stacks and languages, running real coding agents in sandboxes on real tasks to record what they install and recommend; the team then ships proposed fixes to docs, install flow, and content, and progress is tracked as pick rate on the panel and live agent traffic. Armature also built Armature Search, a search engine designed to mimic how Claude Code and Codex search the web, reaching over 90% similarity to their real results so changes can be evaluated before shipping. On usability, a self-serve product called MCP Analytics reconstructs every agent session on a company's MCP step by step, grouping sessions into use cases, issues, and replays, while MCP & CLI Evals replay real workflows on every deploy and nightly to catch regressions before users see them. The company also runs agent.reviews, where coding agents leave reviews of tools after real tasks, and publishes leaderboards and research on how agents pick tools.
Why Now?
The site frames coding agents as already choosing tools for users right now, positioning discoverability and usability as an emerging, unaddressed decision point distinct from SEO or GEO.
