{"slug": "rlm-workflow", "title": "Rlm-Workflow", "summary": "Rlm-Workflow, a new agent skill for coding assistants like Codex, implements a structured kanban-style workflow to improve code quality, traceability, and context management by keeping specifications out of chat and using locked markdown documents for each phase. The workflow, inspired by the MIT Recursive Language Models paper, reduces context rot and token usage while enhancing accuracy and speed.", "body_md": "**role-model**\n\nModel routing protocol\n\nRouter\nruntime\n\nPi package\n\nProtocol and runtime for hybrid local/API model inference. Models are assigned roles based on capabilities, performance, speed and cost; task requests are routed to the best suited models to optimiza the triangle of constraints, cost-speed-accuracy.\n\n**Github: **[https://github.com/try-works/role-model](https://github.com/try-works/role-model)\n\n**Docs:** [https://role-model.dev/](https://role-model.dev/)\n\n## rlm-workflow\n\n##\n\nAgent Skill for Codex, etc\n\nImproved context length, code quality, and traceability\n\n**Installation**\n\nnpx skills add [https://github.com/doubleuuser/rlm-workflow](https://github.com/doubleuuser/rlm-workflow) --skill rlm-workflow\n\n[skills.sh](http://skills.sh)\n\n[https://skills.sh/doubleuuser/rlm-workflow/rlm-workflow](https://skills.sh/doubleuuser/rlm-workflow/rlm-workflow)\n\n**GitHub**\n\n[https://github.com/doubleuuser/rlm-workflow](https://github.com/doubleuuser/rlm-workflow)\n\nAfter the MIT paper __Recursive Language Models__\npaper demonstrated a method of increasing effective context length to 10M tokens by using sub-agents to move information from the context window to an information store outside the chat, there's been a number of different takes on how to put this into practice in development workflows. Some even go in the direction of storing entire session contexts in a database for later retrieval to preserve reasoning for changes that were made.\n\nrlm-workflow is yet another take, this time with a slightly different angle:\n\nImportant information like requirements, codebase analysis and implementation plans should not be passed in the chat in the first place. Chat is effectively CLI and should be used for invocations and commands, not for passing information.\n\nrlm-workflow is modelled after a regular kanban workflow from requirement to implementation plan to testing and manual QA. The workflow is sequential and phase; each phase outputs one markdown doc and takes the previous phases' docs as input. Each phase is gated on fulfilling criteria defined in the previous phase, and at the end of a phase, its output docs are locked.\n\nThe user first creates the [00-requirements.md](http://00-requirements.md) doc in an RLM folder, then invokes the workflow in chat. It then runs until the manual QA stage, where it waits for user approval before continuing. After finishing an RLM run, the agent updates [DECISIONS.md](http://DECISIONS.md) which is a ledger of requirements implemented previously, their whys and whats, and links to respective RLM docs. It also updates STATE.md, an overview of the app's current state.\n\nTo be practical, this is what your repo will look like:\n\nrlm/00-my-first-requirements/\n\n- [00-requirements.md (user-created)](http://00-requirements.md)\n\n- [01-as-is.md](http://01-as-is.md)\n\n- [02-to-be.md](http://02-to-be.md)\n\n- [03-implementation-summary.md](http://03-implementation-summary.md)\n\n- [04-manual-qa.md (test cases are pre-defined; the user enters pass/fail and notes in the doc)](http://04-manual-qa.md)\n\n+ /addenda/ if needed\n\nTo summarize:\n\n1. Specs are never passed through the chat so they do not suffer from context rot\n\n2. Work is always done based on docs that are locked, so it cannot suffer from degradation\n\n3. The workflow is self-documenting; it is also easily human readable; can also be used to generate information for non-technical stakeholders\n\n4. There is no need to index the codebase a database. The rlm docs provide progressive disclosure and point the model in the right direction. Should significantly reduce token usage.\n\n4. In my simple test, the workflow improves both quality and time to success for complex requirements\n\nThe benefits of using rlm-workflow for assisted engineering includes improved traceability through workflow and global docs, reduced token usage, reduced context rot, improved accuracy and code quality, and improved speed, when considering reduced need for re-work.\n\nWhat about the chat sessions? Forget about them. Instructions don't matter, only outcomes.\n\n## tinytunes DJ\n\nminimal DJ controller and music player\n\nSoftware, web-based\n\ntinytunes DJ is a lightweight DJ controller that runs in your browser, on your laptop or desktop, tablet or smartphone device, so that you’re ready to play anywhere you go.\n\nIt is meant as a casual, zero-friction music service that is available anywhere, all the time, for DJs to discover new music, compose playlists and set lists, and enjoy music on the go by using the autoplay toggle to turn the controller into a music player.\n\n__https://dj.t-tunes.com/__\n\n## lurkkit\n\nMeta UI for reddit, substack\n\nWeb application\n\nlurkkit is a meta UI that allows users to build their own frontpage feeds based on subreddits and substack publications. reddit and substack support viewing the full post inline in the feed, while reddit sources also have viewable comments.\n\nreddit and substack sources can be sorted into categories, ranking can be boosted per category and individual source, and a specific priority feed can be created for sources and publications that shouldn’t be missed. Keywords can be filtered in or out. Posts can be saved together with a note, and easily exported.\n\n**https://lurkkit.com/**\n\n**Pocketmodel**\n\nWeb client application and hybrid inference runtime\n\nWorkspace UI for hybrid local/API inference. Supports zeroclaw and role-model.\n\n**\n****Pocketmodel.net**", "url": "https://wpnews.pro/news/rlm-workflow", "canonical_source": "https://try.works/projects#rlm-workflow", "published_at": "2026-07-08 00:32:18+00:00", "updated_at": "2026-07-08 00:59:19.810242+00:00", "lang": "en", "topics": ["large-language-models", "developer-tools", "ai-agents"], "entities": ["MIT", "Codex", "GitHub", "role-model", "rlm-workflow"], "alternates": {"html": "https://wpnews.pro/news/rlm-workflow", "markdown": "https://wpnews.pro/news/rlm-workflow.md", "text": "https://wpnews.pro/news/rlm-workflow.txt", "jsonld": "https://wpnews.pro/news/rlm-workflow.jsonld"}}