ποΈ Target architecture recommendation
The right architecture is not the purest on day one. It is the one that lets you validate the workflow without lying about sovereignty.
Step 1 β Practical MVP
To learn fast:
- Cursor CLI or Aider;
- manual run;
- Markdown report;
- dedicated Git branch;
- human validation.
Cursor CLI (owned by SpaceX since August 20261) is very productive for testing the idea. Aider is closer to the sovereign target because it can call Ollama directly, but its development is frozen (see its fiche).
Step 2 β Controlled runner
To automate:
- scheduled task (cron, systemd timer, self-hosted GitHub Actions, or Agent Canvas automation on an internal backend2);
- dated branch;
- run logs via
vault-log-rununder.agents/vault-maintenance/runs/; - source report;
- notification without automatic merge.
Step 3 β Sovereign target
Recommended stack:
| Layer | Recommended choice | Role |
|---|---|---|
| Code agent | OpenHands (CLI/SDK) β or Aider (frozen, see fiche) for experiments | Modifies files and works with Git |
| Local model | Ollama or vLLM + 27β30B agentic model (Qwen3.8-27B, Muse Glimmer 30B) or specialized coder | On-prem inference with sufficient reasoning |
| Gateway | LiteLLM (lexicon) β monthly update mandatory (vulnerabilities exploited in 2026, one month of support per minor line)34 | OpenAI-compatible API, routing, logs |
| Search | SearXNG | Self-hosted web search |
| Runner and sandbox | OpenHands / Agent Canvas25 | Docker agent, scheduled automations, orchestration of third-party agents via ACP |
| Fast MVP | Cursor CLI | Initial productivity |
Minimum sizing for local Aider
For a realistic sovereign target:
| Use | Recommended local model | Hardware reading |
|---|---|---|
| Simple suggestions, small files | Specialized coder 7B/8B | useful to learn, not reliable as an autonomous agent |
| Controlled fixes on Markdown vault | Dense 14B coder (Qwen2.5-Coder-14B, from 2024; the Qwen3-Coder range has no 14B)6 | practical floor, with strict human validation |
| Regular maintenance, multi-file audit | Coder 24β32B or higher: Devstral Small 2 (24B, December 2025), the Qwen3-Coder-30B-A3B MoE, or Qwen2.5-Coder-32B (2024)6 | recommended target if the agent must produce usable diffs |
| Agentic work with long context, 24 GB VRAM | Agentic generalist 27β30B such as Qwen3.8-27B (~18 GB in Q4, 262k context) or Muse Glimmer 30B in Q4 2026; the Qwen3.6-35B-A3B MoE (3B active) remains the one recommended by the OpenHands docs (May 2026) | context β₯ 32k tokens7; see Choose your model |
| Large refactor or long reasoning | 32B+ with large context, or frontier non-sovereign model in MVP | sovereignty vs quality trade-off |
Key point: the agent that acts on files needs more reasoning than the assistant that retrieves information. VRAM budget must be sized for the editing model, not only the chat model.
Concrete recommendation for this vault
- Short term: continue with Cursor/Aider under human validation.
- Medium term: OpenHands (CLI, or Agent Canvas with scheduled automations and a local LLM profile) or Aider as long as it works, + Ollama + recent agentic model (in Q4 2026, a 27β30B generalist such as Qwen3.8-27B or Muse Glimmer 30B; or the Qwen3.6-35B-A3B MoE recommended by OpenHands in Q2 2026) + SearXNG + maintenance scripts27.
- Long term: LiteLLM as gateway (maintained minor line, patched within the month)4, vLLM if throughput is needed, OpenHands/Agent Canvas for automations and complex sandboxed tasks25.
See also
π Sources and References
Footnotes
-
Cursor, Cursor is now a part of SpaceX, 2026-08-14. https://cursor.com/blog/joining-spacex β©
-
OpenHands, Introducing Agent Canvas (self-hosted control center, scheduled or event-driven automations, LLM profiles, local/Docker/VM/Kubernetes backends, MIT license), 2026-06-16. https://www.openhands.dev/blog/introducing-agent-canvas β© β©2 β©3 β©4
-
BerriAI, GHSA-7hp6-4w63-5g45 (
internal_userβproxy_adminβ host execution escalation, CVSS 9.9, fixed in 1.100.4 / 1.101.3 / 1.102.2 / 1.103.1), 2026-09-30; CISA, Known Exploited Vulnerabilities Catalog (CVE-2026-42208, CVE-2026-42271, CVE-2026-59822), catalog dated 2026-10-08. https://github.com/BerriAI/litellm/security/advisories/GHSA-7hp6-4w63-5g45 Β· https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json β© -
LiteLLM, Version Support Policy (since 2026-06-29, only the four most recent stable minor lines receive fixes), 2026-06-20. https://docs.litellm.ai/blog/version-support β© β©2
-
OpenHands, Use any coding agent in OpenHands with ACP (Agent Client Protocol: Claude Code, Codex, Gemini CLI;
ACPAgentin the SDK), 2026-06-18. https://www.openhands.dev/blog/use-any-coding-agent-in-openhands-with-acp β© β©2 -
Qwen, Qwen2.5-Coder-32B-Instruct (family from 0.5 to 32B, 14B and 32B under Apache 2.0, technical report arXiv:2409.12186 of September 2024); QwenLM, Qwen3-Coder (README: 30B-A3B, Next and 480B-A35B, no dense 14B or 32B variant), accessed 2026-11-19; Mistral AI, Introducing Devstral 2 and Mistral Vibe CLI (Devstral Small 2: 24B, Apache 2.0; Devstral 2: 123B, modified MIT), 2025-12-09. https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct Β· https://github.com/QwenLM/Qwen3-Coder Β· https://mistral.ai/news/devstral-2-vibe-cli β© β©2
-
OpenHands Docs, Local LLMs (Ollama, vLLM, SGLang, LM Studio; Qwen3.6-35B-A3B recommended, β₯ 24 GB of VRAM when quantized, 32k context), updated 2026-05-21. https://docs.openhands.dev/openhands/usage/llms/local-llms β© β©2