apnea

Do agent loops push up token use? A balanced brief

Scope: coding agents (SWE-bench harnesses + production CLIs) and general agent frameworks (LangGraph, Instructor, PydanticAI). Compiled July 2026 from practitioner reports and research papers. Items labelled reported are vendor/blogger claims, not independently verified. The one academic token-economics paper cited is under peer review. ~3,000 words incl. tables and references; ~15 min read at a technical pace (~200 wpm).

Provenance: I used OpenCode with GLM-5.2 for source gathering and drafting. I provided scope, direction, curated sources, some factual corrections and editing. This session was 12 turns with token count 682K (8.0M) in 26K out.

MiMo-Code Feature Architecture Report

This report describes six core features of the MiMo-Code platform (the packages/opencode package): Persistent Memory, Intelligent Context Management, Task Tracking, the Subagent System, Goal / Stop Conditions, and Dream & Distill. Each section traces the feature from user-facing surface down through its implementation layers, database schema, and integration points. A final section analyzes how the features interact.

1. Persistent Memory

Overview

The AI Coding Session Observability Landscape (mid-2026)

What this field is

A new tool category has emerged around observing, recording, and analyzing the interactions between developers and AI coding agents (Claude Code, Cursor, Copilot, Codex, OpenCode, etc.). These tools sit alongside the coding agent -- intercepting traffic, reading session files, hooking into agent events, or integrating into the IDE -- and provide visibility into what was asked, what was done, how much it cost, and what patterns emerge over time.

This is distinct from production LLM observability (Langfuse, LangSmith, Helicone, etc.), which monitors AI applications running in production. The tools here are focused on developer workstations and coding sessions.

The field is roughly six months old as a recognizable category and is evolving rapidly.

Z.ai Coding Plan Lite vs NeuralWatt Cloud -- Pricing Comparison

Date: May 30, 2026
Models compared: GLM-5.1 (zai-org/GLM-5.1-FP8)

What you currently pay: Z.ai Coding Plan Lite

Detail	Value
Price	$18/month (billed monthly on z.ai)

Moving Off a Workspace Account to a Personal Google Account

Best tool: Takeout Transfer

Go to takeout.google.com/transfer while signed into your Workspace account. This copies Gmail and Drive files you own directly to your personal account. Takes hours to a week. The verification code expires in 24 hours.

Known limitations:

Only covers Gmail and My Drive files you own
Does not include contacts, calendar, tasks, or chats
Files in "Shared with me" or Shared Drives are not included -- move them to My Drive first

The Governance Spectrum Scaffold

A Framework for Comparing AI Governance Across Model Risk and Agent Risk


Version	2.0
Date	20 May 2026
Critique cycles	2 (3 independent reviewers per cycle)
Source documents	19 (5 regulations, 6 practitioner/academic papers, 5 macro-prudential and cross-sector frameworks, 2 US fair lending guidance, 1 government agentic AI framework, 1 EU implementing guidance)

Beyond Text: The Spectrum of Code Representation for LLM Coding Agents

An analysis of how code can be represented to LLMs — from raw text to architectural patterns — and where the research frontier currently sits.

The Spectrum

Code can be represented to LLMs at progressively richer levels of abstraction. Each level captures more structural and semantic information, but also requires more sophisticated tooling and domain knowledge to construct.

Vera vs opencode-codebase-index vs AFT: Feature & Metrics Comparison

Vera (v0.7.0) — Local-first semantic code search CLI in pure Rust. 65 languages, ONNX embeddings + cross-encoder reranker, fully offline.

opencode-codebase-index (v0.8.0) — Semantic codebase indexing plugin for OpenCode, also runs as standalone MCP server. Hybrid TypeScript + Rust, API-first embeddings, 17+ languages.

AFT (v0.26.4) — Agent File Toolkit. Tree-sitter powered code manipulation and analysis for AI agents. Rust binary + thin TS plugins. 17 languages. Semantic search + trigram grep + structural editing + call-graph navigation + LSP diagnostics.

GLM-5/5.1 System Prompt Research & Design

Date: 2026-05-12 Purpose: Design an optimal system prompt for GLM-5.1 in opencode (a coding agent CLI), informed by the GLM-5 paper, Z.AI docs, and community findings.

Sources

| Source | Key Takeaway |

	# Lattice Plugin Release Workflow

	## Overview

	Publishing a new version of `@apnea/opencode-lattice` involves three steps:
	bump the version in package.json, push a git tag, and let CI handle the rest.

	The GitHub Action (`.github/workflows/publish-plugin.yml`) automatically:
	1. Publishes to npm when a `v*` tag is pushed
	2. Creates a GitHub release with auto-generated notes