Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 9fa46b923c
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 new memory entry capturing a “detect-changes workflow + fine-grained workflow split” CI parallelization pattern observed in a sibling repository, and updates the memory index to reference it (plus related redaction updates).
Changes:
- Added a new memory file documenting the external “detect-changes” CI pattern and how it composes with existing Zeta backlog/ruleset work.
- Updated
memory/MEMORY.mdto include the new entry and to replace several existing index entries with redacted “sibling org” placeholders.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 10 comments.
| File | Description |
|---|---|
| memory/feedback_detect_changes_pattern_sibling_repo_parallel_optimized_external_anchor_aaron_2026_05_01.md | New memory entry documenting CI parallelization patterns and rollout guidance. |
| memory/MEMORY.md | Adds an index entry for the new memory and updates multiple existing index links/labels. |
… new content only (Aaron 2026-05-01) Aaron 2026-05-01 calibration: *"../it's folder name was fine"* + *"that's why i named it that."* The `../no-copy-only-learning-agents-insight` directory name IS the abstraction discipline Aaron designed. New substrate uses generic placeholders (`<sibling-org>/<sibling-repo>`); existing filenames with company names are GRANDFATHERED per the established naming + scope-of-org-access rule (2026-04-28). My prior commit's `<sibling-org>` substitution was over- aggressive — it broke valid existing MEMORY.md link targets to real files (feedback_servicetitan_naming_*, project_account_setup_snapshot_codex_servicetitan_*, project_frontier_burn_rate_ui_*_servicetitan_*, feedback_servicetitan_demo_*, etc.). The link TEXT was already redacted to "the sibling-repo external anchor" / generic phrasing; the link TARGETS need to point at real files. Restored link targets to the real filenames. New content (memory body + MEMORY.md link text) keeps the redacted phrasing per the no-copy-only-learning principle for new substrate. Resolves PR #1130 review threads PRRT_kwDOSF9kNM5_BSNv + PRRT_kwDOSF9kNM5_BSNx.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6ab8ff9776
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".
…e the" duplicate Two threads from PR #1130 review: 1. Copilot + chatgpt-codex-connector (line 11): composes_with referenced `feedback_otto_352_external_anchor_lineage_for_substrate_credibility_2026_04_26.md` which doesn't exist on disk. The "external-anchor-lineage" concept is captured in the Beacon promotion memo (`feedback_beacon_promotion_load_bearing_rules_earn_external_anchors_aaron_amara_2026_04_28.md`) which IS shipped. Repointed. 2. Copilot (line 281): "the the sibling repo team" — duplicate "the". Removed. Note: the 8 stale `<sibling-org>` placeholder threads on MEMORY.md are outdated post-restoration commit 6ab8ff9 and will resolve as outdated.
… new content only (Aaron 2026-05-01) Aaron 2026-05-01 calibration: *"../it's folder name was fine"* + *"that's why i named it that."* The `../no-copy-only-learning-agents-insight` directory name IS the abstraction discipline Aaron designed. New substrate uses generic placeholders (`<sibling-org>/<sibling-repo>`); existing filenames with company names are GRANDFATHERED per the established naming + scope-of-org-access rule (2026-04-28). My prior commit's `<sibling-org>` substitution was over- aggressive — it broke valid existing MEMORY.md link targets to real files (feedback_servicetitan_naming_*, project_account_setup_snapshot_codex_servicetitan_*, project_frontier_burn_rate_ui_*_servicetitan_*, feedback_servicetitan_demo_*, etc.). The link TEXT was already redacted to "the sibling-repo external anchor" / generic phrasing; the link TARGETS need to point at real files. Restored link targets to the real filenames. New content (memory body + MEMORY.md link text) keeps the redacted phrasing per the no-copy-only-learning principle for new substrate. Resolves PR #1130 review threads PRRT_kwDOSF9kNM5_BSNv + PRRT_kwDOSF9kNM5_BSNx.
…e the" duplicate Two threads from PR #1130 review: 1. Copilot + chatgpt-codex-connector (line 11): composes_with referenced `feedback_otto_352_external_anchor_lineage_for_substrate_credibility_2026_04_26.md` which doesn't exist on disk. The "external-anchor-lineage" concept is captured in the Beacon promotion memo (`feedback_beacon_promotion_load_bearing_rules_earn_external_anchors_aaron_amara_2026_04_28.md`) which IS shipped. Repointed. 2. Copilot (line 281): "the the sibling repo team" — duplicate "the". Removed. Note: the 8 stale `<sibling-org>` placeholder threads on MEMORY.md are outdated post-restoration commit 6ab8ff9 and will resolve as outdated.
5c81e99 to
fddbc76
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 905119204b
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".
…chor (Aaron 2026-05-01) Aaron 2026-05-01: *"../no-copy-only-learning-agents-insight is the best repo in github i've seen setup to be parallel"* + *"you should be able to use the gh cli i think to see it's setting in read only mode too"* + *"this has had a lot of intention put into by others not so much by me the gh config for that project"* + *"opposite of your Pages-with-Jekyll preference; not my preference, i've just used it before cause i didn't want to write ts at the time, bun is probably enough"*. Direct inspection of `../no-copy-only-learning-agents-insight` + `gh api repos/servicetitan/stcrm` (per the DST grade-A pull-to-sibling-repo + gh-api-on-host discipline) revealed seven load-bearing patterns: 1. detect-changes.yaml workflow emitting per-change-class outputs — PRs only run relevant workflows 2. 42 fine-grained workflows (vs our ~15 monolithic) 3. Test parallelism at script level (bash+PS1 pair) 4. **5 concern-aligned rulesets** — empirically validates B-0155 architecture (Copilot review / Flux config / Release Pending / Status Checks + Merge Queue / deployment-branch protection) 5. **Branch protection effectively empty** (zero contexts, strict:null) — empirical proof B-0155 Phase 3 cleanup endpoint works at production scale 6. STCRM uses Wiki not Pages — Aaron's prior Jekyll-on-Pages was a workaround, not preference; bun makes TS-native static gen viable (Astro / Vitepress as B-0154 candidates) 7. AGENTS.md learning-discipline patterns we already have analogues for (lessons-in-PR, joint-learning, source- investigation, search-first authority) Attribution: STCRM is deliberate-by-others (multi-engineer org-scale design); Zeta is Aaron-clicked-alone. High- credibility external anchor for design patterns we extract. Carved candidate: "The detect-changes pattern is the parallel-PR primitive. Without it, every PR contests for every check. With it, PRs that touch different concerns don't see each other."
…ect-changes memory
Per the no-copy-only-learning principle (the directory is named
`../no-copy-only-learning-agents-insight` for exactly this reason):
extract patterns/insights, not company-internal names.
Renamed file: feedback_detect_changes_pattern_stcrm_*
→ feedback_detect_changes_pattern_sibling_repo_*
Replaced all in-body references to STCRM / servicetitan / stcrm
with "the sibling repo" / "<sibling-org>/<sibling-repo>" placeholders.
Cleaned up double-mentions from automated sed pass.
The patterns + architectural insights are preserved. The company
name is gone.
Note: prior commit message + PR title still leak the name (history
already on origin); fixing forward only.
… new content only (Aaron 2026-05-01) Aaron 2026-05-01 calibration: *"../it's folder name was fine"* + *"that's why i named it that."* The `../no-copy-only-learning-agents-insight` directory name IS the abstraction discipline Aaron designed. New substrate uses generic placeholders (`<sibling-org>/<sibling-repo>`); existing filenames with company names are GRANDFATHERED per the established naming + scope-of-org-access rule (2026-04-28). My prior commit's `<sibling-org>` substitution was over- aggressive — it broke valid existing MEMORY.md link targets to real files (feedback_servicetitan_naming_*, project_account_setup_snapshot_codex_servicetitan_*, project_frontier_burn_rate_ui_*_servicetitan_*, feedback_servicetitan_demo_*, etc.). The link TEXT was already redacted to "the sibling-repo external anchor" / generic phrasing; the link TARGETS need to point at real files. Restored link targets to the real filenames. New content (memory body + MEMORY.md link text) keeps the redacted phrasing per the no-copy-only-learning principle for new substrate. Resolves PR #1130 review threads PRRT_kwDOSF9kNM5_BSNv + PRRT_kwDOSF9kNM5_BSNx.
…e the" duplicate Two threads from PR #1130 review: 1. Copilot + chatgpt-codex-connector (line 11): composes_with referenced `feedback_otto_352_external_anchor_lineage_for_substrate_credibility_2026_04_26.md` which doesn't exist on disk. The "external-anchor-lineage" concept is captured in the Beacon promotion memo (`feedback_beacon_promotion_load_bearing_rules_earn_external_anchors_aaron_amara_2026_04_28.md`) which IS shipped. Repointed. 2. Copilot (line 281): "the the sibling repo team" — duplicate "the". Removed. Note: the 8 stale `<sibling-org>` placeholder threads on MEMORY.md are outdated post-restoration commit 6ab8ff9 and will resolve as outdated.
…capitalization in non-verbatim text (Copilot 2 threads)
…place; preserve only the new detect-changes index entry (Copilot 7 threads)
5347ca9 to
2aa55d6
Compare
Summary
Aaron 2026-05-01: "
../no-copy-only-learning-agents-insightis the best repo in github i've seen setup to be parallel."Direct inspection (DST grade-A pull-to-sibling-repo + gh-api-on-host disciplines) revealed seven load-bearing parallel-optimization patterns. Memory file captures them as substrate; company name redacted per the no-copy-only-learning principle.
Patterns captured
detect-changes.yamlworkflow emitting per-change-class outputs — PRs only run relevant workflowsAttribution
The sibling repo's host config is deliberate-by-others (multi-engineer org-scale design); Zeta is Aaron-clicked-alone. High-credibility external anchor for design patterns we extract. Sharpens the click-vs-decision distinction across actors.
Carved candidate
composes_with
B-0153 (lint suite) + B-0154 (Pages + Wiki) + B-0155 (ruleset-split) + B-0156 (TS port) + Otto-235 (4-bash + PS) + task #305 (install.ps1) + task #341 (TS migration) + Otto-352 (external-anchor-lineage)
Test plan
🤖 Generated with Claude Code