# Abo Fles (abofles, ابو فليس) > A free, MIT-licensed plugin for Claude Code. It picks the model (Haiku, Sonnet, Opus, Fable) and effort (low … max) for every subagent from a topic × side matrix, so cheap jobs stop running on expensive settings. Independent open-source project, not made or endorsed by Anthropic. Latest: abofles 2.13.1 and the companion mod abofles-guide 0.15.1 (2026-11-21). 1,317 topics in 31 domains for developers, designers and marketers; 23 agents including 13 specialist reviewers. Install in Claude Code: /plugin marketplace add Sajadh92/abofles /plugin install abofles@sajadh92 /plugin install abofles-guide@sajadh92 (optional: the Abo Fles pane) ## Pages - [Home](https://abofles.com/): what it does, for whom, features, FAQ - [Install](https://abofles.com/install/): install, auto-update, update by hand, requirements, uninstall - [The pane](https://abofles.com/pane/): real screenshots of every page of the Abo Fles pane - [Agents](https://abofles.com/agents/): the 23 agents and the model × effort matrix per tier - [Config](https://abofles.com/config/): every override key with values and defaults - [Changelog](https://abofles.com/changelog/): all 31 versions - [Arabic home](https://abofles.com/ar/) ## Key facts - Routing: each task maps to a topic with a risk tier T1 Minor, T2 Normal, T3 Sensitive, T4 Critical. Each tier × side (Read, Analyse, Plan, Work, Verify) cell sets model, effort and a reply cap. Default matrix "quality": Opus writes, plans and analyses at every tier; Haiku and Sonnet read and verify. - Hooks enforce it: an agent without model and effort is refused; Fable only on Analyse and Plan. - Receipt at the end of a task: agents, real tokens, real dollars and the saving versus the main session's model. - Question gate: a Judge agent tries to answer a question from the repo before the user is asked. - Specialist reviewers after Verify: Security, Finance, Compliance, Tester, QA, UIUX, Clean, DBA, Ops, Perf, Data, Design, Marketing. - Git guard and a Git agent, a work memo with resume, context watch, Prompt checkup, sub-session routing, a Scrum agent for Jira/Linear/ClickUp and more. - Requires Python 3.8+ (macOS, Linux, Windows with Git Bash). ## Links - Source: https://github.com/Sajadh92/abofles - Releases: https://github.com/Sajadh92/abofles/releases - Instagram: https://www.instagram.com/abofles/ - Full text for LLMs: https://abofles.com/llms-full.txt