OpenClaw Build and Operations (5/5) — OpenClaw vs Hermes: 18-Category Scoring

OC 63/90 vs Hermes 75/90: Decision Criteria Beyond the Score

ํ•ต์‹ฌ ์š”์•ฝ

  • A framework for comparing two agent platforms across 18 categories on a 0–5 scale. Result: Hermes 75/90 (83%), OpenClaw 63/90 (70%).
  • Distribution: OpenClaw leads in 7 categories (operations-focused), Hermes leads in 7 (autonomy and extensibility), 2 ties.
  • Key insight: Platform scores are not a ranking tool — they are a tool for asking "is the winning capability transferable?" Portable strengths and structural strengths must be treated differently.

What This Post Covers

A single aggregate score is a weak basis for platform selection decisions. This post covers: (1) the 18-category scorecard structure, (2) a two-axis interpretation method that separates "portability" from "architectural lock-in," and (3) criteria for determining which strengths are preserved and which weaknesses disappear during a platform migration.

Purpose of the Scoring Framework

"A is better" is not sufficient justification for a migration decision. Decomposing into 18 categories and assigning 0–5 per category makes win/loss positions explicit in numeric form. The role of the score is not to provide an answer — it is to generate questions: "Is losing in this category acceptable?"

18-Category Scorecard

OpenClaw Leads (7 categories)

Category OC Hermes Delta Notes
Channel / Gateway 5 4 +1 Discord + Telegram Forum with 6-topic granular routing
Agent Management 5 4 +1 8-layer binding (tool / model / permission / channel / isolation)
Memory 5 3 +2 4-tier + U-tag + Reflect 6-phase + semantic search
Scheduling 4 3 +1 9 cron jobs + 7 LaunchAgents + micro-cycle
Plugins 4 3 +1 oMLX integration, Scrapling, custom hooks
Config Management 4 3 +1 config-drift detection, agent-linter
Context Management 4 3 +1 safeguard compaction, softThreshold 6000

Hermes Leads (7 categories)

Category OC Hermes Delta Notes
Skill Auto-Generation 1 5 -4 Automatic skill generation from usage patterns
Learning Loop 2 5 -3 DSPy + GEPA self-evolution
Execution Backend 2 5 -3 6 backends: Docker / E2B / Modal / Fly.io / Lambda / local
Voice 0 4 -4 Native voice I/O
Sub-agents 2 4 -2 Native profile spawning + delegation
ML / Research 1 4 -3 Trajectory analysis, experiment framework
IDE Integration 1 4 -3 Native VS Code / JetBrains

Ties (2 categories)

Category OC Hermes
Model Routing 4 4
Tool System 4 4

Totals

  • OpenClaw: 63/90 (70%)
  • Hermes: 75/90 (83%)

Score Interpretation: Portability vs. Architectural Lock-in

1. Separating Platform-Native Strengths from Custom Layers

OC memory at 5 vs. Hermes memory at 3 does not reflect a design superiority of the platform itself. The score is a product of the custom layer built on top of OpenClaw — memcore, U-tag, Reflect 6-phase. These are not fixed attributes of the platform; they are layers on top of it, and therefore portable to another platform.

Applied pattern: label each category score as either "platform baseline" or "custom accumulated during operation." This separation makes it explicit what is lost vs. what is carried over during migration.

2. Architecturally Fixed Strengths Are Not Portable

Among Hermes' leading categories, the learning loop (DSPy + GEPA), 6-backend execution, voice, and trajectory analysis cannot be addressed by investing more time into OpenClaw. These are determined at the framework architecture level. Replicating them externally incurs a cost equivalent to rebuilding the platform.

Decision rule: classify whether a capability is "attachable as a layer" or "embedded in the architecture" before interpreting the score.

3. Weaknesses Offset by Portable Layers

Hermes' memory score of 3 stems from a 2,200-character MEMORY.md limit. If memcore — built on OpenClaw to overcome the same constraint — can fill that gap as a layer, then the Hermes + memcore combination converges on a memory score of 5 after migration.

Principle: a weakness that was resolved via a custom layer can be resolved again by applying the same layer to a new platform. Therefore, the weakness column in the scorecard should also flag "whether a portable solution already exists."

Limitations and Scope

  • 0–5 scoring introduces subjectivity onto qualitative indicators. Applying the same framework to a different team can shift scores by ±1.
  • The 18 categories are domain-specific to agent platforms. A different domain (e.g., general PaaS comparison) requires a redesigned category set.
  • The boundary between "portable" and "architecturally fixed" is not always clear. Intermediate cases — features built on top of a plugin system — require individual assessment.

Open Questions

  • When platform scores are near-equal, what is the next axis that breaks the tie — maintenance cost, community, upstream change velocity?
  • When portable custom layers grow thick enough, at what point should the layer itself be extracted as a "platform-independent component"?

Further entries continue in the same measurement format in separate records.

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 »