Skip to content

[codex] document extract-knowhow redesign need #24

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
ktwu01 wants to merge 1 commit into from
Closed

[codex] document extract-knowhow redesign need #24

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions extract-knowhow/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,19 @@ The command will:
5. **Present** a report for your review and confirmation
6. **Generate** OpenScientist-format skill files ready for contribution

## Current Limitation

`extract-knowhow` is installed for both Claude Code and Codex CLI, but the workflow is still effectively **Claude-first** in how it is framed and validated. It does not yet have a first-class cross-agent review step, and it does not yet treat peer review between Claude Code, Codex CLI, and Gemini as part of the extraction contract.

That means the current package is useful, but it likely needs a broader redesign if OpenScientist wants extraction to be genuinely multi-agent rather than "generated once by a single assistant and then trusted."

The likely redesign direction is:

- make agent-to-agent review an explicit step rather than an optional manual habit
- define a single-review handoff contract across Claude Code, Codex CLI, and Gemini
- make the generated report preserve reviewer identity, review outcome, and review notes
- distinguish "agent-generated draft knowledge" from "peer-reviewed reusable skill"

## Output

Generated skill files follow the [OpenScientist skill format](https://github.com/OpenScientists/OpenScientist/blob/main/utils/SKILL_SCHEMA.md) with all required sections:
Expand Down
2 changes: 2 additions & 0 deletions readme.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,6 +62,8 @@ $extract-knowhow

The command automatically scans your conversation history, extracts research know-how, and opens an interactive report in your browser — where you can review, edit, and submit skills directly to OpenScientist via GitHub.

Note: the current `extract-knowhow` workflow is still stronger in Claude Code than in a true multi-agent setup. If OpenScientist wants extracted skills to be reviewed across Claude Code, Codex CLI, and Gemini, the package likely needs a redesign rather than just incremental prompt edits.

### Method B: One-Click Prompt for Web Users (ChatGPT / Claude / Gemini)

Enable memory so the AI can access your history:
Expand Down