Skip to content

Codex/chat ux polish - #68

Closed
jubaoliang wants to merge 12 commits into
mainfrom
codex/chat-ux-polish
Closed

Codex/chat ux polish#68
jubaoliang wants to merge 12 commits into
mainfrom
codex/chat-ux-polish

Conversation

@jubaoliang

Copy link
Copy Markdown
Collaborator

Summary

Target branch

  • Base is develop (feature / fix — default)
  • Base is main (release/* or hotfix/* only)

Type of change

  • Bug fix
  • New feature
  • Breaking change
  • Documentation
  • Refactor / chore
  • Release / hotfix

Test plan

  • make all passes locally
  • Added/updated tests

Checklist

  • Updated CHANGELOG.md (if user-facing)
  • README / docs updated (if needed)

jubaoliang and others added 12 commits July 22, 2026 15:36
Document the branching model for agents and contributors, and align the
publish skill so tags are cut on main only after the release PR merges.

Co-authored-by: jubaoliang <jubaoliang@tencent.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(skills): hot-persist skills_disabled without agent rebuild (#47)

Avoid scheduling a full harness reload when enabling/disabling skills or
finishing SkillHub install, which raced list fetches with AGENT_NOT_FOUND.
Also move the experts template "installed" badge to the card footer.

Co-authored-by: jubaoliang <jubaoliang@tencent.com>

* fix: replace ChatGPT/Codex OAuth PKCE flow with device code flow

The PKCE browser-redirect flow used a dynamically built redirect_uri
(Octop's own callback URL), but OpenAI's shared Codex CLI OAuth client
only whitelists the exact http://localhost:1455/auth/callback used by
the local CLI, so every login attempt failed with
authorize_hydra_invalid_request regardless of deployment domain.

Switch to the device code flow instead: no redirect_uri is involved,
so it works no matter how/where Octop is hosted. The backend now
requests a user_code + verification URL and polls in the background;
the frontend shows the code and lets the user open the verification
page instead of doing a popup + postMessage dance.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

* style: format codex OAuth device flow

---------

Co-authored-by: jubaoliang <jubaoliang@gmail.com>
Co-authored-by: jubaoliang <jubaoliang@tencent.com>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
* fix(skills): hot-persist skills_disabled without agent rebuild (#47)

Avoid scheduling a full harness reload when enabling/disabling skills or
finishing SkillHub install, which raced list fetches with AGENT_NOT_FOUND.
Also move the experts template "installed" badge to the card footer.

Co-authored-by: jubaoliang <jubaoliang@tencent.com>

* fix(skills): fetch SkillHub rankings over HTTP

* fix(skills): preserve SkillHub display metadata

* feat(skills): use HTTP for SkillHub search and install

* refactor(skills): support source-neutral packages

---------

Co-authored-by: jubaoliang <jubaoliang@gmail.com>
Co-authored-by: jubaoliang <jubaoliang@tencent.com>
Co-authored-by: leoxyang <leoxyang@tencent.com>
* feat(browser): support real remote drag with shared canvas pointer

Forward CDP mousedown/mousemove/mouseup (with clickCount) so remote
browser and desktop can drag instead of pan-to-scroll, and share a
connecting indicator while the stream starts.



* fix(chat): restore loading earlier messages on scroll up

Keep the load-older gate armed across virtual list updates, release
the in-flight latch when history declines to start, and expose a
clickable fallback when the list does not overflow.

Co-authored-by: Cursor <cursoragent@cursor.com>

* feat(workspace): clarify path semantics with from_workspace

Add from_workspace so workspace UI keeps leading-slash relative paths
while chat/tool downloads can pass host-absolute send_file paths
through BackendWorkspace, with a host download allowlist.

Co-authored-by: Cursor <cursoragent@cursor.com>

* fix(workspace): preserve chat document path semantics

* fix(workspace): harden file URL and host path checks on Windows

Preserve Unix-style file:// paths instead of url2pathname backslashes,
decode percent-encoded segments, and deny POSIX system roots like /etc
before Path.resolve() on Windows.

Co-authored-by: Cursor <cursoragent@cursor.com>

* fix(workspace): decode Windows file URLs to native paths

Unquote percent-encoded segments in file:///C:/… URLs and normalize
via Path so send_file enrichment matches str(Path.resolve()) on Windows.

Co-authored-by: Cursor <cursoragent@cursor.com>

---------

Co-authored-by: jubaoliang <jubaoliang@tencent.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
* chore(yuanbao): persist default channel endpoints

* chore(deps): require harness-gateway 0.8.7

* docs: note Yuanbao channel update

---------

Co-authored-by: leoxyang <leoxyang@tencent.com>
Keep deepagents conversation offloads in the agent workspace via
harness-agent≥0.9.12, map catalog context_window into model caps for
UI/summarization, and make /compact force summarization on the current
thread instead of resetting it.
Resolve CHANGELOG.md conflicts by keeping the released 0.9.13 notes
and combining duplicate issue references.

Co-authored-by: Cursor <cursoragent@cursor.com>
* feat: PostgreSQL dual-backend for control plane

Add SqlitePool|PostgresPool behind DatabasePool, parallel PG migrations,
setup wizard defer/bind, pg_dump backups with cross-engine refusal, and
memory defaulting to the control-plane DSN when using PostgreSQL.

* fix(chat): delete checkpoint data when a conversation thread is deleted

Deleting a thread only removed Octop's own metadata row; the actual
message history stayed in the agent's LangGraph checkpointer forever,
making "delete conversation" cosmetic. Also confirm before deleting a
session in the UI, since the action is now genuinely destructive.

Checkpoint deletion runs before the metadata row is removed: if it
fails outright (not just "nothing to delete"), the thread stays
visible/retryable instead of orphaning data with no remaining handle.

* fix(memory): stop suggesting pg_dump for portable export of postgres memory

The 501 hint told users to "use pg_dump on the memory schema instead".
That advice predates the shared-table layout: memory for ALL agents now
lives in one harness_memory schema isolated by a namespace column, so a
schema-level pg_dump would export every agent's memory, not just the
requester's — a cross-agent data exposure if followed on a multi-user
install.

Document the two migration models where the refusal is implemented:
sqlite memory is per-agent files moved via pack/adopt; postgres memory
is shared, and the supported way to use it from another host (e.g.
OpenClaw's harness-memory bridge with --backend postgres) is pointing
that host at the same DSN and namespace — shared, not migrated.

* fix(test): update test_slash_compact to renamed SqlitePool

The /compact test module still imported the pre-rename DBPool, which
broke collection (ImportError) on this branch after develop's compact
feature was merged in — DBPool was renamed to SqlitePool (the concrete
class) with DatabasePool now the Protocol.

DBPool -> SqlitePool at all three sites: the import, the _agent_manager
type hint, and the fixture's DBPool(path) instantiation. SqlitePool is
the byte-for-byte successor of DBPool and matches how every other test
constructs a pool; DatabasePool cannot be used at the instantiation
site (a Protocol is not instantiable).

pytest -m live now collects cleanly (31 selected, 0 errors);
test_slash_compact 3 tests pass.

* fix(test): make memory_backend db_path assertion separator-agnostic

test_explicit_sqlite_overrides_postgresql_control_plane compared the
resolved db_path against the literal "/tmp/ws/memory.sqlite", which the
code never produces on Windows: it builds the path with pathlib, so the
Windows CI job got "\tmp\ws\memory.sqlite" and the string equality failed.

Compare as Path objects so the assertion holds on every OS separator,
matching the .as_posix() discipline already used in test_config.py.
…er auth/PWA load.

Co-authored-by: Cursor <cursoragent@cursor.com>
Queue outbound chat sends while the agent is busy, merge terminal and remote browser into a keep-alive Workbench, force headless when DISPLAY is stale, and reuse in-workspace media without copying into outbound/.

Co-authored-by: Cursor <cursoragent@cursor.com>
@jubaoliang jubaoliang closed this Jul 25, 2026
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.

4 participants