OpenClaw to Hermes Migration (7/13) — A 5-Week Plan: Function Category Breakdown

Week 1 is the SQLite schema, Week 5 is the cutover

ํ•ต์‹ฌ ์š”์•ฝ

What this post covers: - Transplant unit design: collapsing 30+ scripts into a function category breakdown (A–K) to fix migration units at a manageable count. - 5-week cutover schedule template: Week 1 schema → Week 2 migration → Week 3 install → Week 4 asset transplant → Week 5 dual-run + cutover. - Three deliverable specs: detailed schedule, Profile definitions, memcore API spec — required fields and inter-document relationships for each.

Function Category Breakdown — Grouping by Functional Unit

Enumerating transplant specs at the script level expands the move/verification surface to 30+ units. Grouping by functional unit reduces that same scope to a tractable set. The grouping criterion is "shares the same schema or protocol," and the table below is the result of applying that criterion.

Category Core
A Memory System 4-tier, Retain tags, U-tag dialectic, Reflect, micro-cycle, Topic-Cued Recall, semantic search
B Heartbeat 3-tier + 11 escalation reasons + Quota check
C Self-Review 3-stage + Skill Evolution Loop
D Daily Audit Phase A–E (guide/update detection, diagnosis, aggregation, final review)
E Researcher Deep Mode 5-round AutoSearch + 3-tier web access
F Channel Routing Discord primary channel + Telegram Forum 6 topics
G Profile Mapping Workspaces → 5–6 profiles
H Hooks PostToolUse, PreToolUse, SessionStart/End/Compact
I Security Model loopback, token, dmPolicy, single-user scope
J Auto Update com.hermes.update-check (new)
K Deprecation Targets OpenClaw-specific skills and scripts

How It Works

Grouping by category produces two effects. First, the transplant checklist shrinks from N (scripts) to K (categories). Second, because items within a category share the same data structures, verification scenarios need to be designed only once and reused across all internal items. Take Heartbeat (B) as an example: the 3-tier logic, 11 escalation reasons, and Quota check all share the same message schema, so a single regression test covers the entire group.

Limitations

Cross-category dependencies (e.g., A Memory ↔ H Hooks) must be tracked separately. Grouping alone does not fix execution order; the actual sequence in the 5-week schedule is derived from the dependency graph.

5-Week Schedule — Dependency Order and Verification Checkpoints

Week Work
1 memcore SQLite schema design + library skeleton
2 bank/ → memcore conversion (rsync snapshot + migration)
3 Hermes install + 1 profile + MemoryProvider adapter
4 Transplant 8 preserved assets + define Hermes profiles (5–6)
5 Dual-run + verification + cutover

Each week is broken down to input / output / verification / rollback conditions at the day level. At any interruption point in the migration, it must be possible to reconstruct "which stage is currently active." The Week 5 dual-run — running both systems simultaneously before stopping one — carries the most densely defined rollback conditions.

Three Deliverable Specs

1. Hermes-Migration-Detailed-Plan.md

The 5-week schedule itemized as Day 1–35. Each day's input, output, verification criteria, and rollback conditions are defined in table form. The objective: at any arbitrary point mid-migration, "current state · next action · recovery path on failure" must be queryable from a single document.

2. Hermes-Profile-Definitions.md

Draft AGENTS.md for 5–6 Hermes profiles: - mir-coordinator — user persona, routing - research — web research + Deep Mode - writer — writing - coder — coding + CLI - google-workspace — Gmail / Calendar - memory-ops — Reflect, audit, housekeeping

Each profile explicitly declares its active skill categories. The compression from multiple workspaces down to 6 profiles is the core design point and is the tangible output of category G (Profile Mapping).

3. memcore-API-Spec.md

External API signatures for the memcore library. Includes a mapping table to Hermes's MemoryProvider interface — initialize / system_prompt_block / prefetch / sync_turn / get_tool_schemas / handle_tool_call / shutdown / on_session_end / on_pre_compress / on_memory_write / on_delegation. This mapping is what makes the Week 3 adapter work mechanical rather than exploratory.

Applicability and Open Questions

Applicability. The function grouping + 5-week cutover template is reusable for any migration satisfying these three conditions: (1) the source system is decomposable into script/skill units; (2) there are sufficient resources to run both systems in parallel; (3) an adapter layer sharing memory/state between the two can be designed. If any of the three conditions is absent, the Week 5 dual-run cannot be structured.

Open questions. - When the category count exceeds the current breakdown, the grouping benefit diminishes. Where is the upper bound? - Is day-level granularity sufficient for Week 5 cutover rollback conditions, or does it need to go down to the hour level? - The MemoryProvider interface's on_pre_compress hook does not map 1:1 to OpenClaw's existing compression logic. Should this gap be absorbed in the adapter, or should the interface itself be adjusted?

The next post covers the Week 1–2 execution results — SQLite schema structure, FTS5 index configuration, and measurements at the point where 7 of 8 preserved assets were transplanted.

Series overview: Series index

๋Œ“๊ธ€

๋Œ“๊ธ€ ์“ฐ๊ธฐ

์ด ๋ธ”๋กœ๊ทธ์˜ ์ธ๊ธฐ ๊ฒŒ์‹œ๋ฌผ

Agent Memory Engine (2/10) — Building an AI Agent Memory System with SQLite Alone

← 1/10 SQLite Memory Engine f… ๐Ÿ“š Series Index 3/10 memcore Completed → A memory engine that runs without daemons, without dependencies, anywhere ํ•ต์‹ฌ ์š”์•ฝ This post covers how to design an AI agent memory engine (memcore) on a single-file SQLite backend. - Structure: 25 modules, ~6,464 lines, 22 DB tables - Principles: no daemon, single-file portability, no required external dependencies, host-agnostic - Core techniques: 4-layer hybrid search (topic · vector · FTS5 · LIKE), U-tag dialectic-based confidence evolution, ingestion-layer bias rejection rules What This Post Covers One approach to building AI agent memory without a server or cloud vector DB. Specifically: (1) the limitations of text-file-based memory, (2) vector DB alternatives compared and selection criteria, (3) how to layer search tiers, (4) how to evolve confidence scores over time, and (5) how to block biased memory accumulation at the ingestion stage. Design Principles: Why Single-File SQLite Text-file-based ...
Read more »

"ML Foundations (9/9) — PyTorch vs TensorFlow, and the Road to Local LLMs"

์ด๋ฏธ์ง€
← 8/9 Deep Learning Architec… ๐Ÿ“š Series Index (series end) The final part. So far we've focused on models. Now we focus on the tools that actually run them . We'll lay out the philosophical difference between PyTorch and TensorFlow, and trace how Hugging Face transformers , llama.cpp , MLX , and Ollama built the bridge to running large language models on your own machine. By the end you should have the full mental model of "download a pretrained LLM and serve it locally." 0. Learning Objectives Explain the eager-vs-graph difference between PyTorch and TensorFlow. Explain, in graph terms, how autograd automates the backward pass. Use Hugging Face transformers ' AutoModel , AutoTokenizer , and pipeline abstractions. Describe llama.cpp's GGUF quantization, INT4 inference, and CPU-first flow. Describe Apple MLX's use of unified memory and how it differs from PyTorch. Run a local LLM with Ollama and call it through an OpenAI-compatible API. ...
Read more »

"RAG Core Study (14/26) — Evaluation Sets with RAGAS & DeepEval"

์ด๋ฏธ์ง€
Series overview: Series index ๐Ÿ“š Series Index Without an evaluation set, every RAG improvement is a story, not evidence. RAG systems fail in more than one place: retrieval, context selection, answer generation, and grounding. That means "looks better" is not a metric. Part 14 explains how to build a golden dataset , how RAGAS and DeepEval fit into the workflow, and why teams must separate retrieval quality from answer quality if they want tuning decisions to hold up. 0. Prerequisites Part 13 reranking — system quality now depends on multi-stage ranking. Part 1 grounding — a correct-looking answer is not enough. Part 12 Hybrid — multiple retrieval settings need comparison on fixed data. 1. Learning Objectives Build a small but useful golden dataset for RAG. Distinguish retrieval metrics from answer metrics . Use RAGAS and DeepEval in the right roles. Avoid the common traps in judge-model-based evaluation. 2. ํ•ต์‹ฌ ์š”์•ฝ An evaluation set for RAG is ...
Read more »

"ML Foundations (8/9) — Deep Learning Architectures: CNN, RNN, Attention"

์ด๋ฏธ์ง€
← 7/9 Deep Learning Training ๐Ÿ“š Series Index 9/9 PyTorch vs TensorFlow → What did we keep adding on top of an MLP? Vision went CNN, sequences went RNN/LSTM, and eventually both converged on Attention and the Transformer. Each architecture is easier to remember if you read it as what it gave up to gain something else . This part is the one-line summary of the decisive inflection points: LeNet → AlexNet → ResNet → LSTM → Attention → Transformer. 0. Learning Objectives Explain why CNN's convolution, pooling, and weight sharing match images so well. Trace what got unblocked from LeNet → AlexNet → VGG → ResNet as depth grew. State BPTT for RNNs and the vanishing/exploding gradient problem. Write the LSTM gate equations (forget, input, output) with intuition. Write the attention formula in query/key/value form and the scaled dot-product variant. State the precise reasons Transformers replaced RNNs (parallelism + long-range dependencies). 1. ํ•ต์‹ฌ ์š”์•ฝ CNN : weight sharin...
Read more »

"ML Foundations (7/9) — Deep Learning Training: Optimizers, Regularization, Initialization"

← 6/9 Neural Networks ๐Ÿ“š Series Index 8/9 Deep Learning Architec… → Building an MLP, as we did in Part 6, is not the same as getting it to train well . The right optimizer, regularization, initialization, and learning rate — when these line up, deep networks converge. When they don't, the network refuses to learn at all. This part is the catalog of those crafts, with formulas, paper citations, and code in one place. 0. Learning Objectives Compare and write the update rules for SGD, Momentum, Nesterov, and Adam. Explain how Dropout, BatchNorm, and LayerNorm work and where they belong in a model. Derive the variance formulas for Xavier and He initialization and match them to activations. Implement step, cosine, and warmup learning-rate schedules in PyTorch. Explain why gradient clipping is effectively required for RNNs and Transformers. Diagnose the most common training failures (NaN, plateau, overfitting) and apply first-line fixes. 1. ํ•ต์‹ฌ ์š”์•ฝ SGD : \(w \leftarro...
Read more »

OpenClaw to Hermes Migration (2/13) — What to Preserve, Partially Port, or Discard

← 1/13 Current Structure Inve… ๐Ÿ“š Series Index 3/13 What Moves to Hermes → A code review record classifying operational scripts into three tiers: preserve, partial-preserve, and discard. What This Post Covers An asset classification framework (preserve / partial-preserve / discard) for migrating a legacy agent system to a successor A concrete case applying the Strangler Fig + Subset Migration strategy to a memory pipeline Classification results evaluated across LOC, external dependencies, and replaceability axes A method for slimming down a bloated single file (1,310 LOC) using the migration as a natural entry point Problem Definition: Port Everything or Cut Deep? The OpenClaw inventory comprises multiple agents, multiple scheduler/daemon processes, and multiple operational scripts. When moving to the successor system, Hermes, the choice is not binary. A full port carries accumulated technical debt along with it; a minimal port discards hard-won domain logic. Code-revie...
Read more »

AI Agents I Built (5/7) — Building an Automated Blogger API Publishing System

← 4/7 My Life Organizer Agent ๐Ÿ“š Series Index 6/7 Dual → Designing a Markdown-Based Auto-Publish Pipeline with Blogger API v3 + OAuth 2.0 Key Summary Blogger API v3 + OAuth 2.0 enables a pipeline that converts Markdown files into publishable HTML posts blogger_publish.py handles the full flow — frontmatter parsing → body cleanup → HTML conversion → CSS injection → API post — in a single script For bulk publishing, quota management (delay, retry, TOC suppression) is the critical design concern Platform Selection Criteria API access is mandatory for AI agent-driven content publishing. A comparison of platforms: Platform API Cost AdSense Decision WordPress.com REST API Paid plan required Paid only Cost overhead Ghost REST API Self-hosted or paid Manual setup Ops overhead Blogger v3 API Free Native integration Adopted Blogger is free, exposes a REST API, and integrates natively with AdSense. For personal blog automation, this combination i...
Read more »