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

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-review-based asset classification resolves that gap.

Decision Context — The Five Structural Constraints That Drive Classification

Before classifying assets, the target architecture must be fixed. Classification criteria derive from the question: "Can this code fit into the destination structure?"

  • Strategy: Strangler Fig (incremental replacement) + Subset Migration (partial port)
  • Memory: SQLite + Python library (memcore). No daemon. Memory is accessed by importing the library, not via a separate process.
  • Directory: Isolated under ~/Hermes/. OpenClaw under ~/Project/ is preserved as a read-only backup.
  • Agents: A new "Hermes management" agent is created. The "OpenClaw management" agent is demoted to backup-only duty.
  • Cutover: Dual execution (running the same functionality on both sides for comparison) → cutover after verification passes.

These five constraints automatically derive the classification criteria: "Can it be absorbed into the memcore library? Can it be folded into a Hermes profile? Does a replacement already exist?"

Preserve (Must Port)

File LOC Port Form Value
user-pattern-stage.py 260 memcore.dialectic module Highest. U-tag 3-phase (observe → hypothesize → verify), Honcho-inspired, anti-bias guard
confidence-decay.py 82 memcore cron tick Decays opinion confidence by 0.02; removes entries below 0.30
topics-validate.py 131 memcore consistency check Ensures bank ↔ topics coherence
topics-expand.py 186 memcore auto-registration Auto-registers newly created files
bank-lint.py 176 memcore validator Validates Retain tag format and data integrity
entity-audit.py 152 memcore stale detection Detects 30-day stale entries via SQLite timestamp queries
docs-snapshot.sh 141 Hermes cron payload Detects docs changes; only the URL needs updating
release-check.sh 66 Hermes cron payload Detects version changes

Classification Principle

The decisive axis for a preserve verdict is absence of external dependencies. Every preserved asset uses only the standard library; file I/O routes through lib_common.py. The migration reduces to a single operation: replacing lib_common.py with memcore.io. Average size is around 175 LOC — small enough to verify equivalence with unit tests.

Top-Priority Asset: user-pattern-stage.py

This file implements a three-phase dialectic (observe → hypothesize → verify) for tracking user utterance patterns and blocking confirmation bias. No equivalent exists in Hermes. Whether or not this single file gets ported accounts for a significant share of the migration's justification.

Partial-Preserve

File LOC Handling
bank-size-watch.py 176 Weakened after SQLite migration. Extract only the MemTree split logic.
proactive-exception-alerts.py 1,310 Severely bloated. Extract only the escalation-reason scoring logic (~300 LOC); delegate notification routing and context collection to the Hermes gateway/hook infrastructure.

Bloated File Slim-Down Pattern

proactive-exception-alerts.py at 1,310 LOC carries the highest accumulated debt. Decomposing its structure reveals three layers:

  1. Scoring logic (~300 LOC): the core evaluation of escalation reasons — migration target
  2. Notification routing: delegatable to the Hermes gateway
  3. Context collection: delegatable to the Hermes hook infrastructure

Migration provides a natural entry point for slimming bloated files like this. Reducing from 1,310 to 300 LOC is debt elimination without functional loss. Layers 2 and 3 have equivalent capabilities in the destination infrastructure, so the rewrite cost approaches zero.

Discard

File LOC Replacement
heartbeat-tick.py 262 100% replaced by Hermes cron + on_turn_start hook

heartbeat-tick.py provides functionality that the destination infrastructure (cron + on_turn_start hook) covers by default. No migration value. Discard here means isolation into the read-only backup (~/Project/), not physical deletion — the file remains accessible for reference at any time.

Code Review Checklist (Reusable)

Applying these questions in sequence to each file converges on a verdict.

  1. Does the destination have an equivalent feature? — If yes, discard candidate.
  2. Are there external dependencies? — If none, porting cost is low → preserve candidate.
  3. Is file size proportional to its responsibility? — If bloated, partial-preserve (slim down).
  4. Is there unique domain logic? — If present and expensive to reimplement, top-priority preserve.
  5. Can equivalence be verified by tests? — Must be true for dual-execution cutover to hold.

Four Distilled Principles

  1. Porting difficulty is determined by the dependency graph, not LOC. Preserved assets port quickly because they carry no external dependencies beyond lib_common.py.
  2. The work lives in verification, not in the port itself. Moving code takes a day or two; dual-execution verification takes longer.
  3. Unique domain logic has nonlinear value relative to LOC. user-pattern-stage.py is 260 LOC, but its lack of any replacement makes it the highest-priority asset.
  4. Migration is a natural opportunity for debt reduction. Slimming a bloated file (1,310 LOC) during migration costs less than a standalone refactoring project.

Scope of Applicability and Open Questions

This classification technique applies readily to the memory and pipeline layer of a "legacy system → successor system" migration. The same criteria do not transfer unchanged to the UI layer or the data schema layer, where schema compatibility, data volume, and migration downtime become the dominant variables.

Open questions:

  • Do the assets classified as preserve actually demonstrate equivalence under dual-execution? This remains an assumption not yet confirmed by data.
  • Is the slim-down ratio seen in the partial-preserve case (1,310 → 300 LOC) a reproducible pattern, or a structural peculiarity of proactive-exception-alerts.py? Needs isolation.

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 »

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 »