Skip to content

Refactor agent learning process and enhance tests#2

Open
marcomc wants to merge 8 commits into
mainfrom
feat/composable-agent-learning-architecture
Open

Refactor agent learning process and enhance tests#2
marcomc wants to merge 8 commits into
mainfrom
feat/composable-agent-learning-architecture

Conversation

@marcomc

@marcomc marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner

Refactor the agent learning process to improve efficiency and clarity. Introduce new commands for creating template drafts and summarizing runs. Enhance test coverage for routing metadata and error handling. Update documentation to reflect these changes and ensure robust operation.

- Updated SKILL.md to streamline the agent learning process, including changes to the promotion and finalization steps.
- Added new commands for creating template drafts and summarizing runs, with detailed explanations.
- Enhanced test coverage in test_agent_learning.py, including tests for routing metadata, enforcement of prevention targets, and handling of state locks.
- Improved error handling and validation in the finalize note process to ensure robust operation.
- Introduced checks for privacy verdicts and candidate rules in template draft creation.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: aedcbf78d2

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".

Comment thread scripts/agent_learning.py
Broaden the private-material detector to catch Linux home and workspace-local paths before draft handoffs are written. Add a create-template-draft regression proving clean-verdict candidate rules with those paths are rejected without creating .work output.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3b2eb869f9

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".

Comment thread skills/consolidate-agent-learnings/SKILL.md
Add the required routing rationale to create-template-draft examples that use --enforce-routing. This keeps the skill prompt and architecture doc aligned with the CLI routing contract.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7e9d78ab6e

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".

Comment thread scripts/agent_learning.py Outdated
Write an owner file for state update locks and reclaim only locks whose recorded process is no longer running. Add a finalize-note regression proving old lock directories are not evicted by age alone and queued notes remain untouched on timeout.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4b6af76aa8

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".

Comment thread scripts/agent_learning.py
Switch state lock acquisition to an atomic hard-link lock file containing the owner PID. This removes the exposed ownerless directory window while preserving dead-owner reclaim and age-only lock timeout behavior.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 122dc2a752

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".

Comment thread scripts/agent_learning.py
Add a safe fallback for filesystems that cannot create hard links and recover old ownerless lock artifacts after a grace period. Extend lock regressions for owned lock timeout, ownerless recovery, and the fallback path.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6bc1865683

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".

Comment thread scripts/agent_learning.py Outdated
Distinguish parsed empty JSON lists from non-JSON strings when coercing routing targets. Add a finalize-note regression proving [] does not satisfy required prevention targets under --enforce-routing.
@marcomc

marcomc commented Jun 6, 2026

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. What shall we delve into next?

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".

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