Skip to content
PidocsCLI & tooling

Review bootstrap/doctor flow changes in pi-setup-bootstrap repo

Session from heypocket-reader-sessions on Hugging Face by thomasmustier · MIT · condensed by Coders Talk

3msession
0interventions
GOAL

Review the working tree diff against origin/main for a Mac setup/bootstrap tool, checking correctness, setup completeness, maintainability, and safety of the bootstrap/doctor flow.

  1. Prompt · +0
    Review the working tree changes in pi-setup-bootstrap against origin/main for correctness, setup completeness on new/partial Macs, maintainability, and bootstrap/doctor safety.
  2. Agent did · +0m
    Pulled the diff stat and full diff, then inspected scripts/merge-bootstrap-settings.py, scripts/ensure-skill-vendors.py, and the Brewfile to understand the scope of changes.
  3. Agent did · +0m
    Read through bootstrap.sh, install.sh, and merge-bootstrap-settings.py line by line to trace the settings-merge and dependency-install logic.
  4. Agent did · +1m
    Checked settings.json packages list and configure-package-filters.py to see how package/skill filtering is configured.
  5. Agent did · +1m
    Searched for theme/vendor directories, listed skills, and inspected the pptx skill's vendored upstreams (anthropic-pptx, openai-slides) to check vendoring correctness.
  6. Outcome · +3m
    Delivered a findings report, flagging issues including bootstrap.sh aborting on externally managed settings.json symlinks and other P2-level correctness/safety concerns.
VERDICT

I'd open next time by pointing straight at the merge/settings logic and vendor dirs instead of letting the agent crawl the whole repo top to bottom, since that's where the real risk (breaking existing setups) turned out to live. The methodical line-by-line read of bootstrap.sh and merge-bootstrap-settings.py is what actually surfaced the symlink-handling bug worth fixing.

I tried this · 0

Discussion · 0

How Builds work