Bitter Frontier
Founding member access recorded.
Checkout cancelled.

Research Version

Candidate rerun for initial frontier roll-up

2026-05-06-candidate-2026-04-22_2026-05-06-frontier-v1

Status
candidate
Window
2026-04-22 to 2026-05-06
Signals
5

Mode: editorial_rerun

Revision Reason

First rerun preserved the main signal shape but still treated Codex /goal as part of generic worker-native state rather than as its own long-horizon autonomy primitive.

Sources harvested

[object Object] [object Object] [object Object] [object Object] [object Object]

Accepted signals from this run

  1. Codex Provider-native long-horizon state is now table stakes.
  2. Codex Authority semantics are explicit but fragmented.
  3. Claude Code Verification is becoming a worker capability.
  4. Codex Plugin, extension, and skill ecosystems are becoming the integration surface.
  5. Pi Coding Agent Worker integrations are not durable doctrine.

Artifact contents

Every file the loop produced for this run, anchored in the repo. Internal links go to the rendered page; the repo path opens the raw artifact on GitHub.

Run digest

The last two weeks were not about one winning coding agent. They were about worker tools becoming fuller environments.

Codex added persisted goals, richer permission profiles, plugin workflows, external session import, and multi-agent controls. Claude Code pushed cloud multi-agent review, session recaps, plugin distribution, hook behavior, MCP governance, and telemetry attribution. Gemini CLI hardened workspace trust and environment loading while experimenting with reviewable memory patches. Hermes introduced a background Curator for skill-library maintenance. Pi kept its minimal-harness posture while rapidly changing providers, transports, extensions, and terminal rendering.

Main Signals

Provider-Native State Is Becoming Real

Codex /goal, Claude session recaps, Gemini Auto Memory, and Hermes Curator all point in the same direction: workers are learning how to carry context forward.

This is useful, but the operator still needs a durable record of which worker state governed the run and what crossed into operator-owned memory.

Signal: 2026-05-06-worker-native-state

Authority Semantics Are Explicit, But Fragmented

Workers increasingly expose permission, trust, environment, hook, MCP, extension, and transport semantics, but each worker does it differently.

Signal: 2026-05-06-fragmented-authority-semantics

Verification Is Moving Into the Workers

Provider-native review, multi-agent controls, subagent evals, and curator reports can produce useful evidence, but they should not become final truth without settlement.

Signal: 2026-05-06-worker-verification

Plugins, Extensions, and Skills Are the New Surface Area

The practical power of worker CLIs is moving into plugins, hooks, extensions, skills, and transport modules.

Signal: 2026-05-06-plugin-extension-skill-surface

Worker Integrations Are Not Doctrine

Provider lists, release channels, app-server surfaces, and built-in integrations change quickly enough that wrappers must be thin and versioned.

Signal: 2026-05-06-worker-integrations-not-doctrine

Candidate Verdict

This rerun is readable and structurally coherent, but it underweights Codex /goal. Persistent objectives are not generic memory. They are a long-horizon autonomy primitive and should be split out in the next run.