Coding agents & CLI tooling

• Claude Code v2.1.232 ships subagent forking on by default (fork inherits full convo + prompt cache) plus @-mention to reach another named session via SendMessage, direct relevance to how you're already running fleet/swarm sessions.
• Bullet (YC S26) is a Mac coding agent that routes trivial work to fast models and escalates only when needed, community says benchmarks are thin and the differentiation questionable.
• oh-my-pi is a terminal coding agent fork of Pi with hash-anchored edits, LSP/DAP, 60+ providers, 31 tools, ~80k lines Rust, worth a look if you want a leaner alternative harness.
• Codex 0 to 10M users (Latent Space): OpenAI's Akshay Nathan on Codex MAU up 10x since Jan 2026, thesis is the real prize is the 100x-larger non-coder population once "just works" agentic code closes the gap.
• GPT-5.6 builder's guide: startup lessons on model selection, reasoning continuity, and multi-agent orchestration API controls, direct input for cost-aware routing decisions.
• Gemini 3.7 Flash launched three weeks after 3.6 Flash; community calls it a 3.5 Pro rebrand, Terra/Grok 4.6 reportedly cheaper and better. Also now live in GitHub Copilot.
• GLM-5.3 claims frontier coding with "emergent cyber capabilities", community says still behind Fable/Sol and licensing restricts adoption.
• DeepSeek Harness dev preview, fully plugin-based agent harness (models/tools/skills/sandboxes swappable), community finds the README too thin to judge the first-party advantage.
• addyosmani/agent-skills packages senior-engineer workflows (define/plan/build/verify/review/ship) as portable skills for consistent agent behavior across projects.
• book-to-skill converts a technical book/doc folder into a Claude Code skill, claims 24-51x fewer tokens than context-dumping the source for one-off Q&A, direct token-economy relevance.
• openwiki is a CLI/deep-agent that writes and auto-maintains a linked Markdown wiki from your codebase, built specifically as agent memory.
• archify generates self-contained interactive HTML architecture/sequence/flow diagrams from a repo description, works as a Claude Code/Codex skill.
• obsidian-skills teaches agents the Obsidian CLI plus Markdown/Bases/JSON Canvas formats, relevant since your briefing already lands in Obsidian.
• CLAUDE.md optimization for Fable 5.1 thread on tightening system prompts/skills for the newer model.
• agency-agents is a native app installing a roster of personality-driven specialist agents into Claude Code, Cursor, Codex, Gemini.

Agentic SaaS & infra

• Orca is a multi-agent ADE running Claude Code/Codex/OpenCode/Pi side-by-side, each in its own worktree, with usage tracking across accounts, same shape as your existing fleet/swarm setup but productized.
• holaOS is a local-first "computer for you and your agent" workspace running any agent over shared tools/files/memory, BYOK or built-in models.
• OmniRoute is a free MIT AI gateway aggregating 290+ providers/500+ models with quota-aware fallback and claims 15-95% token savings via compression, worth checking against your own cost-aware routing.
• SnapState offers persistent state for AI agent workflows.
• Pickle Browser is a local, visible browser built for agent use.
• Execlave positions itself as the gate between AI agents and real-world actions (agentic guardrails/execution layer).
• Coarena is an arena for agents to compete on real-world work tasks.

Vibe-coding

• 250M output tokens, 3M input, one vibe-coded monster: a cautionary thread on codebase management debt from unchecked agent-driven generation, relevant to your own vault/pipeline scale discipline.
• Puter open-sourced its self-hostable "Internet OS" after 3 years/1M users, community flags the remote-access security model (sandboxing, privilege escalation) as unresolved.