Today I published an MCP server that connects the Free Law Project’s CourtListener API to AI agents running in Claude Code and OpenCode (https://github.com/DefendTheDisabled/courtlistener-mcp). This gives my AI agents the ability to search CourtListener’s extensive case law databases with both keyword search and semantic search.
The courtlistener-mcp should run with any AI CLI or desktop app supporting local MCP servers (e.g Claude desktop). The courtlistener-mcp doesn’t yet have remote server capabilities which would let it work with the web apps people commonly use. Eventually I’ll get around to adding that functionality. But for now I’m just trying to get my entire stack running so I can get back to my legal work instead of spending all of this time building AI tools and data systems.
I think if anyone wants a way to measure how useful AI really is and the good it can and is doing in society, they need look no further than my work. AI changes the world for the better when it helps people like me do work like this. AI is a powerful assistive device for people with disabilities. The rational, ethical use of AI to solve meaningful problems enables accountability through accessible justice, healthcare reform through litigation and medical research, and productively by helping people with disabilities and those without be better able to participate fully in society.
Making AI technology that is aligned with humanity’s best interests occurs when humans use AI technology in a way that is aligned with humanity’s best interests. The greatest threat from AI isn’t AI itself, it’s the people using AI to uphold the status quo and work against the common good. The problem of AI alignment is not really about the AI being properly aligned, it’s about people being misaligned and trying to build an AI technology that prevents us from doing things that individually and collectively harm us.
Why I’m Building AI Tools
In addition to the severe impairments from my disabilities, and the significant hardships from the illegal activities of the state of Tennessee and its contracted actors, I have continued to encounter significant obstacles in society that obstruct my efforts to prepare my federal civil rights lawsuit and get it filed.
I’ve had to spend most of the past 4 months building AI tools and data systems for my case, as I can’t find any companies providing the services I need as a disabled pro se litigant which would let me stay focused on the legal and advocacy work.
Most companies building legal tools focus on providing AI services to lawyers and law firms. These companies often prohibit non-legal professionals from using their services, or restrict access through extreme subscription prices of $400-$700 a month.
The companies that do offer services to pro se litigants have offerings which are limited and leave so much unaddressed that they can’t be of much help. For example, cetient.com has a free tier that lets one search case law. The search results are useful, but limited. The cetient platform doesn’t allow doing in depth legal analyses on one’s case data, creating agentic workflows, connecting custom tools or agent skills, running code, producing artifacts, and many other critical functions that are needed to work productively. Cetient is basically a good AI powered search engine for case law from the Free Law Project’s CourtListener database, which is useful, but not enough for pro se complex civil litigation..
Services like Harvey AI or Thomas Reuters Co-Counsel aim to provide a suite of tools for lawyers and law firms at hundreds to thousands of dollars a month. However, in early 2025 from what I could tell from their marketing materials and my discussions with sales reps, their platforms lacked core functions I would need but most lawyers do not or would not use. Making their services a very expensive option that not only could not meet my needs, but I would have to fight to access. Maybe they have since improved their platforms capabilities, but I don’t think these companies intend to curate their services to help disabled pro se litigants, or make the cost of the services more reasonable.
I tried multiple consumer facing services (e.g. ChatGPT, Gemini, Claude, you.com, abacus.ai, etc) and while generally useful for many tasks, none of them let me build the systems I need, systems which connect my data to multiple agents. Abacus.ai was close, but ultimately had problems which the company showed no interest or intent to fix.
I also grew concerned by the terms of services most companies employed. They typically reserve the right to terminate accounts out of convenience. Meaning, if the state of Tennessee or UnitedHealthcare made things inconvenient for the AI companies providing services to me, they might just terminate my account in the middle of litigation.
I lost a lot of time trying to make these consumer-facing options work before I accepted that no one was building the platform with the tools people like me need. That what they did offer which could kind of work was not being made accessible to people like me. Effective tools and services were locked behind $500 to $5000 a month accounts focused on professionals or enterprise. They grant access under a terms of service that offers no protection to vulnerable people who need these tools. If I was going to get the tools I needed to try to complete the tasks that let me solve the problems between me and having human rights, I was going to have to build them by myself.
While I remain interested in seeing if professional and enterprise services could be useful to me, I have to go with what I know I can access and derive utility from and not have taken away from me on a whim. Which is what I’ve been building. An open-source suite of AI tools to support civil rights litigation, research, and advocacy efforts. I intend to share these tools over time so everyone can benefit.
I am vibe coding these AI tools and database systems using Claude Code and OpenCode. The design of each tool and system is a collaborative process requiring a lot of learning and work on my part. Where projects already exist, I integrate them. AI isn’t yet at the point of figuring things out all on its own. But it will be and my development plans account for this.
I aim to develop a cost-effective AI stack for pro se litigants that lets the user take ownership of their stack and customize it to meet their project needs. Accessing Justice shouldn’t be something one has to ask permission for.
Defend The Disabled 