Ecosystem Update - 2026-06-03

TL;DR

• Today's Tier 1 crawl mostly reinforces patterns this harness already has: lifecycle hooks, read-only specialist agents, skills, MCP, plugin marketplaces, route classification, and verification-first closure.
• Implemented one safe Quick Win: completed nickname_candidates metadata for all local Codex agents so multi-agent views stay scannable without touching policy, hooks, scripts, or binaries.
• No automatic runtime upgrade, native-memory toggle, plugin-hook enablement, or wholesale third-party skill import passed the anti-overengineering and trust gates.

Quick Wins

Auto-Implemented

• Backed up ~/.codex/config.toml, ~/.codex/hooks.json, and all ~/.codex/agents/*.toml files under ~/.codex/backups/2026-06-03/.
• Added only TOML metadata to ~/.codex/agents/chad-twin.toml, ~/.codex/agents/planner.toml, ~/.codex/agents/python-reviewer.toml, ~/.codex/agents/typescript-reviewer.toml, and ~/.codex/agents/validator.toml.
• Verification passed: ~/.codex/config.toml, ~/.codex/hooks.json, and all ~/.codex/agents/*.toml parse successfully.
• Verification passed: python3 ~/.codex/bin/codex-runtime-doctor exits with errors=0 warnings=4; warnings are unchanged runtime posture warnings around CLI split/pre-release app CLI and TERM=dumb.

Build Queue

• Dynamic workflow trigger intake (agent-pattern) - https://howborisusesclaudecode.com/ - Boris's May 28 notes describe dynamic workflows triggered by asking for a workflow on large migrations/refactors. Codex already has /auto, planning-gate, max_threads = 3, and governed slices; the useful next step is a lightweight intake note or eval for when to ask Codex to decompose into parallel agents, not a new orchestration engine.
• Granular approval profile review (Codex-md) - https://developers.openai.com/codex/config-reference - Official config now documents granular approval categories and approvals_reviewer. Current power-user default is approval_policy = "never" with conservative profiles; review whether a profile-specific auto_review setting belongs in a future constrained profile, not the global runtime.
• Status line and title defaults evaluation (Codex-md) - https://developers.openai.com/codex/cli/slash-commands#built-in-slash-commands - Official slash commands expose /statusline and /title for scannability. Evaluate with the current notification and agent nickname setup before persisting TUI defaults.

Research

• No fresh Tier 2 paper from today's crawl cleared the "directly applicable and not previously seen" bar. Recent relevant themes remain persistent-agent re-entry defense, out-of-scope action measurement, operational safety taxonomies, and structured multi-agent role baselines; those are already represented in prior state or Build Queue items.

Already Have

OpenAI developer docs MCP, live web search posture, GPT-5.5 default model, goals enabled, direct/lightweight/autonomous route contract, UserPromptSubmit route classifier, PreToolUse Bash guard, protected main/master push blocking, PostToolUse verification and failure-context hooks, SessionStart startup/resume/clear/compact coverage, Stop and PreCompact omni-mem hooks, read-only explorer/planner/reviewer/python-reviewer/typescript-reviewer/validator agents, scoped worker and chad-twin agents, complete agent nickname metadata, bounded agent thread/depth/runtime caps, conservative and review profiles, plugin and app connectors, destructive app tools disabled by default, prompt telemetry off, local runtime doctor, config posture checker, skill creator/installer/audit flows, skills-janitor, rlm-scan, planning-gate, /auto, orchestrate-local, codex-security, security-audit, what-would-chad-do, and policy-sized AGENTS.md with runtime-reference handoff.

Rejected

• Auto-upgrade Codex CLI during ecosystem-update - changes active runtime binaries and still needs explicit rollback notes plus smoke coverage; current PATH remains codex-cli 0.133.0 and packaged app CLI remains 0.135.0-alpha.1.
• Enable native Codex memories as a Quick Win - conflicts with the current omni-mem-first policy and prior rejection history; requires an explicit pilot.
• Enable plugin hooks globally - current config intentionally has plugin_hooks = false; enabling third-party hook execution needs a trust review and targeted need.
• Install community skill/plugin catalogs wholesale - duplicates existing local skills and would bypass the outside-skill audit rule.
• Add new hooks from community catalogs - the skill hard limit blocks hook wiring unless the target script already exists and is explicitly identified; today's candidates would require new scripts or trust review.
• Edit ~/.codex/AGENTS.md or ~/AGENTS.md as a Quick Win - constitutional policy docs are explicitly out of scope for automatic daily mutations.
• Add a dynamic-workflow daemon or external control plane - current governed /auto plus subagents covers the recurring need; a new orchestrator is not justified by today's crawl.

Sources checked: https://github.com/hesreallyhim/awesome-claude-code, https://raw.githubusercontent.com/hesreallyhim/awesome-claude-code/main/THE_RESOURCES_TABLE.csv, https://howborisusesclaudecode.com/, https://github.com/shanraisshan/codex-cli-best-practice, https://raw.githubusercontent.com/shanraisshan/codex-cli-best-practice/main/README.md, supplemental web search for Codex hooks/agents/skills, https://arxiv.org/search/?searchtype=all&query=LLM+agent+coding&order=-announced_date_first, arXiv web search supplement, https://developers.openai.com/codex/config-reference, https://developers.openai.com/codex/cli/slash-commands Tier 2 fetched: yes Tier 3 fetched: no - last tier3 run was 2026-05-30T10:34:13Z, still within 7 days Run at: 2026-06-03T10:34:09Z