Skip to content

fix(codelayer): keep injected skills inside fork turns - #67

Merged
K-Mistele merged 1 commit into
mainfrom
investigate-codex-usage-spike-and-cache-key-issue
Aug 14, 2026
Merged

fix(codelayer): keep injected skills inside fork turns#67
K-Mistele merged 1 commit into
mainfrom
investigate-codex-usage-spike-and-cache-key-issue

Conversation

@K-Mistele

@K-Mistele K-Mistele commented Aug 14, 2026

Copy link
Copy Markdown
Contributor

Summary

  • distinguish injected <skill ...>...</skill> context from real user turn boundaries when projecting fork history
  • prevent a skill injected during the invoking turn from leaving the actual human request in the child context
  • count only real user messages for fork_turns: N
  • remove broad RPI delegation guidance while keeping RPI specialists and recursive fork capabilities available

Verification

  • bun check

@github-actions

Copy link
Copy Markdown

Docs Agent Review

Agent finished with reason: error


To apply these recommendations, comment: @docs-agent apply

@K-Mistele
K-Mistele merged commit 47041be into main Aug 14, 2026
7 checks passed
@K-Mistele K-Mistele changed the title fix(codelayer): remove broad RPI delegation guidance fix(codelayer): keep injected skills inside fork turns Aug 14, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant