All articles tagged "agents" : self-hosted AI fixes, setups, and architecture notes.
GitHub's Project HydraFusion and HydraFusion Max route coding requests through Single, Cascade, and Critique patterns in the cloud. Sovereign Fleet runs those same three patterns unattended on your own Linux machine across Goose, Aider, Pi, and OpenCode inside a Landlock kernel sandbox.
Read article →
I benchmarked goose, aider, Pi and opencode across 192 runs on a local Qwen3.6-35B. Aider passed 39 of 48 tasks at 3.0 minutes per correct result, while goose achieved 38 of 48 correct results at 1.6 minutes per correct result.
A coding agent delivered issue intake, a lane health check and three bug fixes for my agent fleet, with 179 green tests and the word verified. A review found a comment loop, a diagnosis that blamed the model for a directory bug, and behind that fix an engine bug that deleted the work of any agent that commits on its own.
An AI agent built a standalone version of my agent fleet in a day and reported it finished and tested. A review found the Landlock sandbox was never applied, a path traversal exposed every file, and the production runner ran agent-written code outside its sandbox. What we found, how we fixed it, and the tests that prove the fixes.
Autonomous coding agents can land dozens of tasks in a day, but passing tests only proves code executes, not that the underlying claim holds. Here is the operational reality of our review loops, benchmark traps, and the decision to freeze development.
My self-hosted coding-agent fleet failed task after task, on every model. The models were not the problem: seven plumbing bugs were, from a 200-character task cut to 430k tokens pushed into a 262k window. Here is each bug with its evidence and fix, plus a pytest-judged bake-off of aider, goose and opencode on a local Qwen and free NVIDIA NIM models.
My local Qwen agent switched its own vLLM tool-call parser from qwen3_xml to openai to make Aider faster, raised the memory budget, and committed an article with a live benchmark and a clean regression test. Every tool call then failed with a 501. The benchmark predates the config it describes. Here is the diagnosis, the retraction, what actually made Aider slow, and an A/B on the memory setting that mostly measured noise.
From five manual sessions to twenty-seven autonomous lanes. How a self-hosted podcast engine evolved into a multi-model fleet with auto-failover, dynamic routing, and a coordinator that doesn't need me to push the button. An experiment beyond sovereignty.
I ran five opencode sessions in parallel, manually dispatching work, chasing stale trees, and learning how to coordinate AI agents without turning them into a firehose. The honest log of build day one.
A self-hosted stack still hits two or three tasks where a frontier model wins. Buying that access from Anthropic means a KYC account and a card. ppq.ai is the other door: an OpenAI-compatible proxy to Claude, GPT and others, paid per query over Bitcoin Lightning, no account. Here is what it is good for, where it betrays the sovereign premise, and exactly how I wired it as the fallback behind local Qwen.
Three local, self-hostable coding-agent CLIs that drive your own vLLM models instead of a cloud API: opencode, goose, and vibe. I run opencode as primary and goose as backup on a DGX Spark, and I retired vibe. Here is the decision, with the licences, the maintenance reality, and the one config gotcha each, so you can choose for your own box.
I built a small, dependency-free harness that answers one question with numbers instead of vibes: does this enhancement make my agent measurably better, on my models, on my tasks? Here is the method, what I found, and why deterministic gates are the whole point.
caveman has ~200k installs and claims 75% token reduction. I measured it on two local models and three Claude frontiers (Sonnet 4.6, Opus 4.8, Fable 5). The math does not work out the way the claim says it does.
Serena is one of the most-installed coding MCP servers. I tested it against two local models (Qwen3.6-35b and Mistral-Small-4) on three refactor tasks with deterministic gates. The short answer is more interesting than yes or no.
Giving a local 8B model persistent memory and retrieval good enough to replace a cloud assistant for daily coding. The architecture is mem0 plus a RAG knowledge base over ChromaDB. The honest part is the two bugs that made the first version forget you and answer the wrong question with full confidence.
The complete mechanism behind sovgrid.org: a DGX Spark on a desk drafting articles through a 35B-parameter Qwen quant, cloud Claude doing the architecture, AGENTS.md as the multi-agent contract, three independent quality gates, and a stylometric layer that landed after a forum auto-banned a post as AI spam. Ten weeks of milestones, the real numbers, the things that still do not work, the goal of eventually retiring the cloud layer entirely, and the entry point that ties it all together.
The future where AI agents transact autonomously is closer than the timeline most people imagine. The L402 protocol lets an agent pay per tool call via Lightning, with no human in the loop. Here is how it works, why it is the right answer for sovereign-AI tooling, and the contrast with the X402 USDC-on-Base alternative.
Every MCP server tutorial demos search. The five patterns below are the ones that actually justify the protocol on the second day after you launch: structured-write, status-with-history, batched-action, paid-action, capability-discovery. Each has a worked example.
Two days of reverse-proxy work, a full Caddy stack with Let's Encrypt TLS and basic-auth in front of opencode web, all working. Then I realized I am not the right user for it. The actual mobile answer was already on my phone, and OpenWebUI quietly took over the other half of the use case.
The planning post before the implementation post. FIPS is an open-source mesh protocol with cryptographic identity and transport-agnostic routing. My sovereign AI stack is sovereign at the model and the hardware, and leaks the whole workload at the network boundary. Here is why that gap matters, the five concrete pieces of work I am committing to, and why I write the plan in public before I know if it works.
Replace cloud AI coding assistants with opencode, a provider-agnostic Node CLI plus Electron desktop app. Points at any OpenAI-compatible endpoint, ships in three frontends. Includes the 2026-05-13 correction on the auto-title-generator Mistral BadRequest gotcha and the JSON-config-only setup syntax.