AI analysis grounded in the code graph — computed facts, not vibes · 2026-07-06T05:40:57Z
Agency Agents is a curated collection of Markdown-defined AI agent personas — each file combining an identity, workflow, and set of "deliverables" — organised into divisions (e.g. engineering, GIS, healthcare, network engineering). Shell scripts (install.sh, convert.sh) install and adapt this roster into the config formats expected by tools such as Claude Code, Cursor, Codex, Gemini CLI, OpenCode, Osaurus and Hermes. It targets developers who want ready-made, "specialised" prompt personas rather than generic system prompts, installable with a single command or via a companion desktop app.
The star count (127,631) vastly outsizes what a Shell-only agent-prompt repo would organically expect, but the period in question shows 0 stars gained — so whatever drove the bulk of that growth predates or lies outside this window. Commit activity is heavy and continuous (roster additions, tool-integration fixes, a new native app announcement), which explains sustained relevance rather than the historical spike itself; the graph facts don't contain data (e.g. HN/Reddit mentions, prior release notes) that would account for the original surge.
What changed recently, how it's actually built (from the code graph), and whether you should care. Free account — no card, no spam.