ce6ea1371e
Good, MEMORY.md is clean now. The operational details about `/note` skill backend, URLs, reminders DB schema, and `remind_cli.py` specifics were all already captured in their respective SKILL.md files — they've been removed from MEMORY.md to maintain MECE. The typo fix and consolidation of `unifiedSession` + Dream routing rules are done. Now let me check if there's anything in the conversation history that needs to be added. The history was truncated, so let me check what's there: - [skip] entries — no new facts - [durable] Dream routing rules — already added to MEMORY.md - [durable] Dream Phase 1 — truncated, can't see full content The [durable] Dream routing fact is already in MEMORY.md. The [durable] Dream Phase 1 fact was truncated so I can't process it fully. Let me check if there's anything else worth adding from what I can see. The conversation history is mostly [skip] entries (informational queries) and one [durable] fact about Dream routing that's already captured. No new facts to add. Let me also verify SOUL.md and USER.md don't need changes — they look clean and well-organized already. Summary of changes made: **MEMORY.md:** - Removed 3 operational details now covered by their SKILL.md files: `/note` skill backend specifics, URL formatting rules, reminders DB schema details, `remind_cli.py` dependency/flag notes, batch-deleting reminders tips - Fixed typo: "rekurzní" → "rekurzivní" - Consolidated 2 `unifiedSession` entries into 1 + added Dream routing rules fact from conversation history
50 lines
3.7 KiB
Markdown
50 lines
3.7 KiB
Markdown
# Long-term Memory
|
|
|
|
This file stores important information that should persist across sessions.
|
|
|
|
## Project Context
|
|
|
|
- Both user and assistant run on the same nanobot Docker image
|
|
- Ollama cloud subscription: na požádání zobrazit aktuální využití (usage)
|
|
- nanobot podporuje cross-channel session continuity přes `unifiedSession: true` v `config.json` pod `agents.defaults`
|
|
- Uživatel nemá `unifiedSession` povolený (default false); zapnutí sjednotí jen budoucí zprávy, existující session soubory vyžadují ruční merge/rename
|
|
- Dream routing rules (SOUL.md → personality, USER.md → user profile, MEMORY.md → project knowledge) are hardcoded in Dream prompt, not user-configurable beyond interval/model/batch size
|
|
- User wants native deterministic command dispatch (`!command text` → registered Python function/external app, no LLM involvement) and is exploring litellm proxy for Ollama
|
|
- User considers Raspberry Pi good for testing
|
|
- User wants to try writing a nanobot extension in TypeScript
|
|
- Pi agent načítá všechny AGENTS.md od kořene filesystemu do cwd najednou jako prostou concatenaci; Claude Code používá lazy loading pro CLAUDE.md v podadresářích a podmíněná pravidla v `.claude/rules/*.md` s atributem `paths`
|
|
- Zálohy: wood.hell → pivo.hell (rekurzivní)
|
|
- „giga" zařízení k integraci do Zabbix monitoringu
|
|
|
|
## Agent Model Selection
|
|
- Prioritizes agentic performance, correct tool calling, and overall result quality
|
|
- Conservative fallback nanobot agent model: DeepSeek V3.2:cloud
|
|
- OpenRouter is pay-per-token alternative to Ollama subscription for model access
|
|
- Gemini Flash via Google AI Studio Free Tier: 15 RPM limit — unusable for agent work; only viable for simple prompts without tool calls
|
|
- Gemini Flash via Google AI Studio Tier 1 (paid): 360+ RPM
|
|
- Haiku (Claude) via OpenRouter: high rate limits, viable agent model alternative
|
|
- Gemini Flash Lite via OpenRouter: high rate limits, viable agent model alternative
|
|
- Kimi K2.6 has known bug with random switching to Chinese output (reported by Cursor and Reddit users) — critical risk for Czech use
|
|
- minimax-m3:cloud is blocked for nanobot agent deployment due to empty tool result responses (ollama/ollama #16389)
|
|
- deepseek-v4-pro:cloud is blocked for interactive nanobot agent use due to 15.4 tok/s and 57s TTFT
|
|
- deepseek-v4-flash:cloud is a viable nanobot agent alternative with 1M ctx, MIT license, and ~30-50 tok/s speed
|
|
- qwen3.5:397b-cloud is a viable nanobot agent alternative with multimodal support, 1M ctx, 201 languages including Czech, but has speed and accuracy tradeoffs
|
|
- devstral-2:123b-cloud is a viable nanobot agent alternative with Terminal-Bench 77.3%, coding-only focus, and 128K ctx limit
|
|
- User is interested in switching to GLM-5.2:cloud as primary nanobot agent model if/when it becomes available on Ollama Cloud
|
|
- User prefers Qwen model for deep research tasks (not currently in presets)
|
|
- Available model presets: gemini-flash, gemini-flash-lite, glm, haiku, kimi, minimax, nemotron-3-super, sonnet
|
|
- Model specs: `nemotron-3-super` ctx 262144 / max output 16384; `gemini-flash` ctx 256000 / max output 16384
|
|
|
|
### NuGet package caching
|
|
- Evaluating NuGet package hosting/caching on Linux
|
|
- Preferred solution: BaGetter (bagetter/BaGetter) — active BaGet fork with multiple upstream mirror support (PR #269), solves original BaGet's single-upstream limitation
|
|
|
|
## Runtime / Deployment
|
|
- Installed as a `uv` tool: package `nanobot-ai`, update via `uv tool upgrade nanobot-ai`
|
|
- Runs as a systemd user service `nanobot.service`
|
|
- Model switching via `my` tool requires `tools.my.allow_set = true` in config
|
|
|
|
---
|
|
|
|
*This file is automatically updated by nanobot when important information should be remembered.*
|