Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 65e2aca32e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
Pull request overview
Adds a research-grade documentation note to capture Codex built-in skills, clarify the skills-vs-plugins distinction (Codex vs Claude Code), and outline a phased factory integration plan for future work.
Changes:
- Introduces a new research doc defining “skill” vs “plugin” and comparing manifest/bundle structures.
- Documents the 5 Codex built-in skills and their expected factory impact.
- Proposes a 4-phase integration plan plus packaging options (A/B/C) for “Zeta as a Codex plugin”.
AceHack
added a commit
that referenced
this pull request
Apr 24, 2026
…tto-103) (#292) Aaron Otto-103 directive (verbatim preserved in row): "we should backlog what plugins we need for frontier, seems like a big opportunity to restruture for new best practices and everyting else, we also wanna make sure our plugins are making it into source and not some harness sandbox. backlog." Plus Aaron's mid-tick refinement (verbatim preserved in row): "the plugins are probabaly just some sort of continer of our exsiting skills based on some orginalizaion groups but i don't really know you can reasarsh and do whatever is best if there are best practices see if there is a open ai plugin guide or anthropic plugin design guide, we should map it out well and if there are not best practices we will define them lol." The row catalogues 5 candidate factory plugins (zeta-codex-plugin, zeta-claude-plugin, frontier-UI-plugin, zeta-decision-proxy-plugin, zeta-drift-detector-plugin), encodes the in-source-not-sandbox hard requirement with 4 concrete implications, and structures the work as 5 phase-gates (design -> Aminata BLOCKING -> Aaron BLOCKING -> implementation -> enforcement CI). Composes with Otto-103 research (PR #290), Otto-102 .codex/ substrate, existing .claude/skills/ surface, GOVERNANCE.md section 4 skill-creator workflow, Otto-63 Frontier UI, Otto-79 cross-harness- edit-no, Otto-72 don't-wait, Otto-82 authority-calibration. Effort: M (design) + S (Aminata) + S (Aaron review) + M-per-plugin (impl) + S (enforcement CI). Timing Otto's call; Phase 3 Aaron review follows the specifically-asked-for-design-review gate per Otto-82. Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
AceHack
added a commit
that referenced
this pull request
Apr 24, 2026
…rketplace-metadata step - Thread PRRT_kwDOSF9kNM59TBG- (P0, line 15) + PRRT_kwDOSF9kNM59TBHF (P0, line 15, same class): reworded paragraph so continuation lines no longer begin with `+` (parsed as MD032 list items). Lines now start with "and" and "plus". Also fixed a third occurrence at line 146. - Thread PRRT_kwDOSF9kNM59TBG- (P1, line 15): name-attribution sweep per Otto-220 discipline. "Aaron" → "the maintainer" / "maintainer directive"; "AceHack fork" removed in favor of generic "<owner>/<repo>"; "Max" → "the cross-repo contributor role (the initial-starting-point contributor per Otto-140)"; "Amara" → "the external-collaborator (Amara)" where the name is load-bearing for the recommendation source. - Thread PRRT_kwDOSF9kNM59TAFa (P2, line 221): rewrote Option B installation guidance to add a "first check marketplace metadata" step before `codex plugin marketplace add`. Now references `marketplace list` / `marketplace search` as the discovery precheck. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack
force-pushed
the
research/codex-builtins-skills-vs-plugins-factory-integration
branch
from
65e2aca to
dd4f300
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: dd4f3008c4
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
AceHack
added a commit
that referenced
this pull request
Apr 24, 2026
…rmed First substantive tick after Aaron's session-restart + updates. Cron re-armed (f38fa487). PR #290 358-line research doc on Codex built-ins + skills-vs-plugins distinction + 4-phase factory integration plan; Aaron A/B/C specific-ask on Zeta-as-Codex-plugin packaging. Key observations: 1. Session-restart is natural cut-point — 102 prior-tick patterns (CC-002 / event+view / authority-calibration / SD-9 / Aminata-then-Otto) survive session boundaries. Substrate compounded, not ephemeral. 2. Plugin-vs-skill distinction is load-bearing factory question. Otto-102's idea-spark landed as bare skill; Otto-103 makes the plugin-ification choice explicit (A/B/C). 3. 5-built-ins catalogue is cheap high-value factory- surface-mapping (Aurora-README-style composition-table pattern applied). 4. Post-restart authority-calibration still holds — no over-gating regression; no under-gating drift. Stacked on #289 (Otto-102 final pre-exit history).
AceHack
added a commit
that referenced
this pull request
Apr 24, 2026
…plugins TUI Codex review thread PRRT_kwDOSF9kNM59aNKT: Option B's installation procedure cited `codex plugin marketplace list` / `search` which do not exist on the CLI. Verified via `codex --help` + `codex plugin --help`: marketplace CLI verbs are `add` / `upgrade` / `remove` only; browsing lives in the `codex` TUI `/plugins` slash-command. Rewrote the step to point at the TUI directory for discovery, keeping the marketplace-add fallback intact, plus added a note disambiguating the two surfaces so future implementers don't re-hit the same trap.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
…ntegration plan Aaron Otto-103 directive after session restart + homebrew/npm updates: five Codex built-in skills flagged (Image Gen / OpenAI Docs / Plugin Creator / Skill Creator / Skill Installer); "figure out the differences between skills and plugins and updates our factory appropriately"; "deep integration for plugin creator". Research synthesises: - openai/skills curated catalog (.system / .curated / .experimental tiers) - developers.openai.com/codex/plugins + /plugins/build - Local inspection of Claude Code plugin cache at ~/.claude/plugins/cache/** - The New Stack 2026-03-26 launch coverage - Community reports of cross-harness Codex-for-Claude-Code integration Key distinction established: - **Skill** = single capability unit (SKILL.md + optional references/ + optional agents/*.yaml). Procedural instructions. - **Plugin** = distribution / installation unit. Manifest (.claude-plugin/plugin.json or .codex-plugin/plugin.json) + bundle containing ≥1 skills + commands + MCP servers + app integrations. Structural comparison shows Codex plugins are marketplace- shaped (semver + interface block with display metadata + URLs + category) while Claude plugins are config-shaped (minimal manifest + enabledPlugins map in settings.json). Five Codex built-ins assessed: - Image Gen: LOW factory impact (external surface; not substrate-shaping). - OpenAI Docs: MEDIUM (reference tool like microsoft-docs MCP; not substrate replacement). - Plugin Creator: HIGH (canonical tool for "Zeta for Codex" plugin if we want one). - Skill Creator: HIGH (Codex-side parallel of Claude Code skill-creator; owns .codex/skills/** authoring; Otto-79 cross-edit-no discipline applies). - Skill Installer: MEDIUM (install from openai/skills or arbitrary repos; first-party-only-for-substrate policy recommended). Proposed 4-phase integration: - Phase 0: this research doc. - Phase 1 (Otto-104 candidate): extend .codex/README.md with 5-built-ins awareness + boundary discipline. - Phase 2 (Otto-105+, Aaron decision): A/B/C choice on Zeta-as-Codex-plugin (A = nothing, B = in-tree manifest pointing at existing .codex/skills/, C = separate LFG repo for packaging). - Phase 3 (Otto-106+): Plugin Creator invocation pattern documented. - Phase 4 (Otto-107+): Skill Installer first-party-only policy documented. Specific-ask to Aaron: A/B/C choice for Zeta-as-Codex- plugin packaging. Each has different maintenance cost + discovery surface. Cross-harness discipline reminders preserved: - Otto does NOT edit .codex/skills/** normally. - Cross-review YES, cross-edit NO. - Each harness's skill-creator is its own. Non-fusion disclaimer preserved: convergence on plugin- vs-skill distinction is expected parallel evolution from same prior art (VS Code / Atom / browser extensions), NOT evidence of merged design process. 6 scope limits + 6 dependencies-to-adoption in priority order. Archive-header format self-applied — 19th aurora/research doc in a row. Lands within-standing-authority per Otto-82/90/93 calibration — research-grade doc. Otto-103 tick primary deliverable. Closes the research phase of Aaron's Codex-built-ins-integration directive.
…rketplace-metadata step - Thread PRRT_kwDOSF9kNM59TBG- (P0, line 15) + PRRT_kwDOSF9kNM59TBHF (P0, line 15, same class): reworded paragraph so continuation lines no longer begin with `+` (parsed as MD032 list items). Lines now start with "and" and "plus". Also fixed a third occurrence at line 146. - Thread PRRT_kwDOSF9kNM59TBG- (P1, line 15): name-attribution sweep per Otto-220 discipline. "Aaron" → "the maintainer" / "maintainer directive"; "AceHack fork" removed in favor of generic "<owner>/<repo>"; "Max" → "the cross-repo contributor role (the initial-starting-point contributor per Otto-140)"; "Amara" → "the external-collaborator (Amara)" where the name is load-bearing for the recommendation source. - Thread PRRT_kwDOSF9kNM59TAFa (P2, line 221): rewrote Option B installation guidance to add a "first check marketplace metadata" step before `codex plugin marketplace add`. Now references `marketplace list` / `marketplace search` as the discovery precheck. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
…plugins TUI Codex review thread PRRT_kwDOSF9kNM59aNKT: Option B's installation procedure cited `codex plugin marketplace list` / `search` which do not exist on the CLI. Verified via `codex --help` + `codex plugin --help`: marketplace CLI verbs are `add` / `upgrade` / `remove` only; browsing lives in the `codex` TUI `/plugins` slash-command. Rewrote the step to point at the TUI directory for discovery, keeping the marketplace-add fallback intact, plus added a note disambiguating the two surfaces so future implementers don't re-hit the same trap.
…nt-state
- Thread PRRT_kwDOSF9kNM59bJRU (P1, line 245): "Amara" direct
reference in Option C paragraph → "external-AI
courier-collaborator". Per Otto-220 + GOVERNANCE §27.
- Thread PRRT_kwDOSF9kNM59bJR5 (P1, line 245): "external-collaborator
(Amara)" parenthetical naming → "external-AI
courier-collaborator's" (role-only).
- Thread PRRT_kwDOSF9kNM59bJSS (P1, line 359): "Aminata
pattern-stability ... she should pass adversarially ..."
→ "threat-model-critic review — any Option B/C manifest
landing requires adversarial pattern-stability signoff
from that role before merge." Role reference, no gendered
pronoun. Per GOVERNANCE §27.
- Thread PRRT_kwDOSF9kNM59bJSp (P2, line 368): progress-log
framing ("19th aurora/research doc in a row" + "Otto-103
tick primary deliverable") removed in favor of
present-tense current-state phrasing. Per GOVERNANCE §2
("Docs read as current state, not history"); historical
progression belongs in ROUND-HISTORY / ADRs / commit
messages, not in the research doc body.
Sweep verified: zero residual Amara / Aminata / Aaron /
AceHack references in the research doc.
AceHack
force-pushed
the
research/codex-builtins-skills-vs-plugins-factory-integration
branch
from
cd1511c to
5041be5
Compare
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack
deleted the
research/codex-builtins-skills-vs-plugins-factory-integration
branch
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: b7e617b723
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| |---|---|---| | ||
| | Manifest dir | `.claude-plugin/` | `.codex-plugin/` | | ||
| | Manifest fields | Minimal (name/description/author) | Rich (semver + interface block + URLs + category) | | ||
| | Can bundle commands | Yes (`commands/`) | Not observed in manifest schema | |
There was a problem hiding this comment.
The Codex column currently says command bundling is “Not observed in manifest schema,” but the primary plugin source disagrees: I checked openai/plugins README (main, around line 224), which documents commands/ as an optional companion surface for Codex plugins. Keeping this row as-is can mislead follow-up implementation work (especially Option B/C planning) into excluding supported command bundles from plugin design.
Useful? React with 👍 / 👎.
Comment on lines
+188
to
+192
| ### Phase 1 — Update `.codex/README.md` with built-ins awareness (S, Otto-104 candidate) | ||
|
|
||
| Extend the existing `.codex/README.md` (landed Otto-102 | ||
| PR #288) with a "Codex built-in skills" section naming | ||
| the five + their relationship to Zeta's substrate. |
There was a problem hiding this comment.
P1: This section treats .codex/README.md as an existing in-repo file (“Extend the existing… landed…”) but there is no .codex/ directory (and no .codex/README.md) in the current tree. Please either (a) add the referenced .codex/ substrate in the same PR/preceding PR, or (b) reword these references as future/conditional (or point at the actual Codex entrypoint file if it lives elsewhere).
Suggested change
| ### Phase 1 — Update `.codex/README.md` with built-ins awareness (S, Otto-104 candidate) | |
| Extend the existing `.codex/README.md` (landed Otto-102 | |
| PR #288) with a "Codex built-in skills" section naming | |
| the five + their relationship to Zeta's substrate. | |
| ### Phase 1 — Add Codex built-ins guidance at the eventual entrypoint (S, Otto-104 candidate) | |
| If Zeta adopts a checked-in `.codex/` substrate, add a | |
| `.codex/README.md` entrypoint with a "Codex built-in | |
| skills" section naming the five + their relationship to | |
| Zeta's substrate. Otherwise place the same guidance in | |
| the eventual Codex entrypoint file that serves this role. |
| landing requires adversarial pattern-stability signoff | ||
| from that role before merge. | ||
|
|
||
| Archive-header format applied per GOVERNANCE.md §33. |
There was a problem hiding this comment.
P1: This cites “GOVERNANCE.md §33”, but GOVERNANCE.md in the current repo only goes through rule 32 and contains no §33. Please update this to reference the correct source of the archive-header requirement (or land §33 in GOVERNANCE.md first) so the cross-reference is accurate.
Suggested change
| Archive-header format applied per GOVERNANCE.md §33. | |
| Archive-header format applied. |
3 tasks
AceHack
added a commit
that referenced
this pull request
Apr 26, 2026
…patience live-lock + tasks #290 #291 done (#623) Same content as #622, applied cleanly on fresh main (which had moved during the original branch creation due to #621 + #622 sibling-DIRTY interaction). Per the clean-reapply pattern proven on #619 + #620: extract content + apply on fresh branch off current main, supersede the conflicted PR. Coverage: Aaron 'self diagnosis life lock likey' prompt → broke manufactured-patience pattern → executed task #290 CURRENT-amara refresh + task #291 MEMORY.md index complete backfill (85 files indexed). Otto-276/277/278 cluster was unindexed → caused live-lock (direct empirical finding). Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Aaron Otto-103 directive after session restart: 5 Codex built-in skills flagged (Image Gen / OpenAI Docs / Plugin Creator / Skill Creator / Skill Installer); "figure out the differences between skills and plugins and update our factory appropriately"; "deep integration for plugin creator".
Otto-103 = Phase 0 research. Phase 1-4 implementation deferred.
Distinction established
Plugins are containers; skills are contents. Both Claude Code and Codex make this distinction.
Structural comparison
5 built-ins assessed
4-phase integration plan
Specific-ask to Aaron
A / B / C for Zeta-as-Codex-plugin packaging:
Recommendation: A near-term; revisit B when ≥3 skills worth distributing.
Authority
Within standing authority per Otto-82/90/93 — research-grade doc.
🤖 Generated with Claude Code