Intended way of storing / clean up of all markdown artifacts after a sucessful workflow?

[juliuszaesar]

These skills / workflow generate lots of documents per feature. Let's take a Brownfield project for example; I run the egm-bootstrap and get all these readme files in project root and subfolders. Then i want to build a feature (using git worktrees) and this process generates even more markdown files. At the end when everything is done, I might not want to keep the generated ones (from egm-brainstorm etc).

If i wanted to remove or archive all the generated markdown files after complete feature, how would you suggest i do that in an efficient way? How are the Google Developers handling this case, or do all files just get pushed to git repo in all cases?

[i6moons]

You're not meant to commit all of it. The approved design doc goes in docs/designs/ and through a PR. Briefs, drafts and review reports can go to a folder you keep out of git, using the EGM artifact directory setting.

What's documented

• Skills write briefs, drafts and reports to the working directory, or to the harness's artifact directory if it has one. In Claude Code, Cowork, Codex and ChatGPT you can redirect them by adding EGM artifact directory: .dev/docs to your prompt or to CLAUDE.md/AGENTS.md (project or user level). Design docs stay where they are, and the docs say to add the folder to .gitignore (harnesses.md). This setting was added in #118 and ships in v0.1.0, so update the plugin if yours is older.
• When you tell /egm-draft to save, it writes <DESIGN_DIR>/<FEATURE>.md (default docs/designs/) and adds it to SITEMAP.md (egm_draft). The guide then says to send it to a human as a PR (getting_started).
• The bootstrap READMEs are meant to stay as long-lived context (FAQ). Bootstrap never commits them, though: tracking is up to you (egm_bootstrap). To keep them apart from your own READMEs, run it with target_filename set to EGM.md (L44-L47). That also makes them easy to find or delete later.

There's no documented archive or cleanup step. #110, still open, asks for a default output folder. In that issue, the author of #118 describes their own setup: intermediate docs in an untracked ./.dev/docs/, design docs in a committed ./docs/. I can't say how teams inside Google handle it.

With worktrees (my suggestion, not documented)

Put EGM artifact directory: .dev/docs in a user-level CLAUDE.md/AGENTS.md (or the repo's). The path is relative to the working directory, so each worktree gets its own .dev/docs. Then ignore it once for every worktree:

echo '.dev/' >> "$(git rev-parse --git-common-dir)/info/exclude"   # or commit it to .gitignore

When the feature is done, commit the design doc, then archive or drop the rest:

# optional archive, from inside the feature worktree
mkdir -p ~/egm-archive && mv .dev/docs ~/egm-archive/"$(git branch --show-current | tr / -)"
# from the main checkout; ignored files are removed with the worktree, no --force needed
git worktree remove ../my-feature

For files left in the repo root from earlier runs, run this from the root. -n only previews, so run it again with -f to delete:

git clean -n -- concepts_brief.md discussion_summary.md design_draft.md \
  zoom_out_report.md comprehension_report.md critic_report.md \
  intermediate_zoom_out_report.md implementation_readiness_report.md \
  consolidated_review_report.md goldfish_review_report.md ':(glob)*_DRIFT_REPORT.md'

/egm-draft looks for the brief in ./, scratch/ and the harness artifact directory. If it doesn't find it in the new folder, pass the path as an argument: /egm-draft .dev/docs/concepts_brief.md (/egm:egm-draft in Claude Code).