{"entries":[{"tag":"Building","emoji":"🤖","title":"Rasputin_Loop: a Slack-native agent, one brain, no router","bodyMd":"A single Strands agent loop (MiniMax M3) that thinks, calls a tool, and replies in-thread -> no intent parser, no command router. Provenance-stamped episodic memory in SQLite, hybrid RAG over libSQL/Turso fusing FTS5 + local embeddings via Reciprocal Rank Fusion, runtime MCP tools, Jira auto-triage, guardrails, and a `loop-eval` golden-set CI gate. Runs several distinct Slack apps from one process. Repo: [github.com/Allmight-456/Rasputin_Loop](https://github.com/Allmight-456/Rasputin_Loop).","links":[{"label":"github.com/Allmight-456/Rasputin_Loop","href":"https://github.com/Allmight-456/Rasputin_Loop"}]},{"tag":"Building","emoji":"🛠️","title":"SkillForge: agents that auto-detect their own skills","bodyMd":"AST-based retrieval over a Claude Skill library, human-gated approval loop. Cuts repeat-class error context from **~3000 → ~80 tokens (~50×)**. Repo: [github.com/Allmight-456/SkillForge](https://github.com/Allmight-456).","links":[{"label":"github.com/Allmight-456/SkillForge","href":"https://github.com/Allmight-456"}]},{"tag":"Running","emoji":"🔁","title":"Hermes & OpenClaw on Hostinger cloud","bodyMd":"Self-hosted long-running agent boxes, kept alive specifically to watch where production-grade agent loops degrade. Tool-loop convergence, memory bleed across runs, where they crack under sustained load. Operational notes feed back into SkillForge.","links":[]},{"tag":"Harness engineering","emoji":"🧪","title":"guardrails, evals, agent loops","bodyMd":"Wiring them *into* the LLM workflow itself, not bolted on after. The open question: where do evals belong -> pre-tool, post-tool, or as a separate critic agent?","links":[]},{"tag":"Agent memory","emoji":"🧠","title":"mem0 vs SkillForge","bodyMd":"[mem0](https://github.com/mem0ai/mem0) for long-horizon memory. Mapping the gap between mem0 (episodic, vector-backed) and SkillForge (procedural, AST-indexed) -> different halves of the same problem.","links":[{"label":"mem0","href":"https://github.com/mem0ai/mem0"}]},{"tag":"Retrieval","emoji":"🔍","title":"PageIndex & Agentic RAG","bodyMd":"[PageIndex](https://github.com/VectifyAI/PageIndex) and tree-structured retrieval -> moving past vector-only. Reads more like how you'd actually search a codebase or a textbook.","links":[{"label":"PageIndex","href":"https://github.com/VectifyAI/PageIndex"}]},{"tag":"Tooling","emoji":"🧰","title":"Conductor.build & Superset side-by-side","bodyMd":"Running them against real PRs to see which agentic loop converges. Cursor stays the daily driver; these are the challengers.","links":[]},{"tag":"Studying","emoji":"📡","title":"Karpathy's nanochat & Autoresearch","bodyMd":"Single-GPU nanochat lineage. Reference for what \"tiny but real\" looks like at the model layer.","links":[]}],"fetchedAt":"2026-09-19T10:26:19.080Z","source":"github.com/Allmight-456"}