mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-13 23:39:21 +02:00
DottaandPaperclip
2ad36f4ff2
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - GitHub sources import skills and keep their updates together.
> - Disconnect only disabled a source and kept its read-only imports.
> - A disabled source could not be removed or imported again.
> - Removing a source needs an explicit decision about its imported
skills.
> - This pull request removes the source after the user keeps editable
copies or deletes the imports.
> - Existing disabled sources use the same removal flow.
## Linked Issues or Issue Description
Refs #15731, #14713.
**What happened?**
Disconnect left the repository on Sources. Its Disconnect action was
then disabled. The saved repository still blocked a new import.
**Expected behavior**
Remove the repository from Sources. Require a choice to keep editable
copies or delete its imports. Allow removal of sources that are already
disconnected.
**Steps to reproduce**
1. Import a repository in Skills → Sources.
2. Choose Disconnect source.
3. Observe the disconnected row and disabled removal action.
4. Try to import the repository again.
**Paperclip version or commit**
d67e3507b2.
## What Changed
- Add one confirmation with a required keep-or-delete choice.
- Remove the source and entries, including legacy disabled sources.
- Keep complete editable copies in My Skills, including shared
repository files and executable modes. Preserve skill IDs, keys,
versions, and agent assignments.
- Delete imported skills and source in one transaction with its activity
record. Check policy for every linked skill, including deselected
entries. Reject deletion of skills still used by agents.
- Fence removal with the source revision. Reload stale confirmations and
require a new choice. Prevent an in-flight refresh from restoring
removed imports.
- Reject incomplete legacy copies with a recovery message. Lock copied
skill rows against concurrent deletion and check publication succeeded.
- Add regression tests and production-page Storybook examples. Make an
initially unselected radio-card group keyboard accessible.
## Verification
- Workspace typecheck and build passed. Final affected server/UI
typechecks passed.
- Focused server, UI, and API suites passed: 114 tests initially. After
review fixes, all 21 source persistence and 58 route authorization tests
passed, including the complete repository copy regression; affected
server/UI typechecks and token gates passed again.
- Design token gates passed.
- Real isolated app: imported a public repository, assigned bro to the
CEO, verified that deletion reports the assignment, kept an editable
copy, observed the source disappear, opened the copy in My Skills, saved
an edit, and refreshed. The assignment remained. Re-imported the same
repository, deleted its new unused import, and verified the original
editable copy and assignment remained. No model tasks ran.
- Full local Vitest was attempted but terminated with exit 137 before
completion; it is not counted as a pass. Current-head sharded CI is
pending. Greptile reviewed the current head at 5/5 with no new
actionable comments. The final workspace build passed again.
- Storybook catalog aligned. The legacy removal example was inspected in
the browser; the Delete choice enables removal.
- Storybook: Skills / GitHub-synced skills / Sources · Disconnect source
and Sources · Remove disconnected source.
## Risks
- DELETE of a skill source now requires the saved revision and an
explicit copy/delete choice. Old callers must update their request body.
- Deletion removes imported skills and their records. Assigned skills
must be detached first.
- Incomplete legacy versions must be refreshed before keeping copies;
removal does not silently discard supporting files.
- Copies intentionally stop receiving upstream updates. Their IDs and
keys remain stable for assigned agents.
- No database migration. Copy writes use new managed directories and
clean up on failure. Delete restores quarantined files before releasing
publication locks on failure.
## Model Used
OpenAI GPT-6 through Codex, configured model `gpt-6-astra`. The runtime
does not expose the context-window size. Used shell tools, code editing,
tests, and browser interaction.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip <noreply@paperclip.ing>