Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
25 commits
Select commit Hold shift + click to select a range
247a5f5
fix(database): align schema migrations through upstream v22 (#528)
SaladDay Oct 22, 2026
a285cd7
chore(release): prepare v5.11.3 and thank contributors (#529)
SaladDay Oct 22, 2026
11bf812
fix(tui): clarify managed Codex OAuth quota refresh controls
zydo Oct 24, 2026
db41f84
fix(tui): align managed OAuth quota controls with upstream
SaladDay Oct 25, 2026
a9eb0ce
Merge pull request #534 from zydo/fix/533-codex-oauth-usage-query
SaladDay Oct 25, 2026
14d4b98
fix(codex): preserve provider authentication in forms
SaladDay Oct 25, 2026
73c6b34
test(tui): load Codex preview through provider form
SaladDay Oct 25, 2026
6c2f7f0
Merge pull request #535 from SaladDay/fix/532-codex-form-auth
SaladDay Oct 25, 2026
b54a105
fix(claude): align configuration editing with upstream (#536)
SaladDay Oct 25, 2026
c42a20f
docs: add IPWO as sixth sponsor
SaladDay Nov 4, 2026
d5c8df0
fix(presets): refresh provider model templates to current lineups (#548)
zydo Nov 4, 2026
15f5563
fix(proxy): keep mid-conversation system messages in place for prefix…
Xiao-Maker-droid Nov 4, 2026
12e4613
fix(codex): omit unsupported MCP type field from generated TOML (#512)
RGB-loop Nov 4, 2026
8a03888
fix(codex): bind managed ChatGPT accounts to official providers (#550)
SaladDay Nov 5, 2026
0d95f37
fix(proxy): emit response.incomplete for incomplete Codex responses (…
RGB-loop Nov 5, 2026
4dfe82f
fix(sessions): allow Windows snapshot publication and cleanup (#552)
SaladDay Nov 5, 2026
dca6620
fix(claude): align proxy placeholder authentication with upstream (#556)
SaladDay Nov 12, 2026
3fc7167
fix(database): support upstream schemas 23–26 for sync restore (#549)
stan-lei Nov 12, 2026
c6ae6fa
fix(tui): exit when the input terminal hangs up (#558)
SaladDay Nov 12, 2026
65c2c83
fix(webdav-sync): keep file symlink names in skills archives (#559)
SaladDay Nov 12, 2026
3b52a3c
docs: unify the blind review gate (#560)
SaladDay Nov 12, 2026
b1fa06e
docs: refine the blind review gate (#561)
SaladDay Nov 12, 2026
bfdb1aa
chore(release): prepare v5.11.4
SaladDay Nov 17, 2026
9a2bdf5
fix(ci): use supported macOS runners for releases
SaladDay Nov 17, 2026
0eb96e7
fix(deeplink): keep a Codex link's provider config instead of rebuild…
RGB-loop Nov 17, 2026
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
6 changes: 3 additions & 3 deletions .github/workflows/release.yml
Original file line number Diff line number Diff line change
Expand Up @@ -84,12 +84,12 @@ jobs:
matrix:
include:
# macOS ARM64
- os: macos-14
- os: macos-15
platform: darwin-arm64
target: aarch64-apple-darwin
binary: cc-switch
# macOS x86_64
- os: macos-14
- os: macos-15
platform: darwin-x64
target: x86_64-apple-darwin
binary: cc-switch
Expand Down Expand Up @@ -178,7 +178,7 @@ jobs:
# Create macOS universal binary
universal-macos:
name: Universal macOS Binary
runs-on: macos-14
runs-on: macos-15
needs: build
steps:
- name: Download ARM64 binary
Expand Down
20 changes: 18 additions & 2 deletions .github/workflows/rust-ci.yml
Original file line number Diff line number Diff line change
Expand Up @@ -82,11 +82,12 @@ jobs:
cargo check --lib --no-default-features
cargo test --lib --no-default-features --no-run

- name: Test Windows session cache file synchronization
- name: Test Windows session manifest publication and cleanup
if: runner.os == 'Windows'
working-directory: src-tauri
# These tests use only their own temporary directories.
run: cargo test --lib --no-default-features session_manager::paged_manifest::tests::sync_private_file -- --nocapture
run: |
cargo test --lib --no-default-features session_manager::paged_manifest::tests:: -- --test-threads=1 --nocapture

- name: Check default Windows CLI build
if: runner.os == 'Windows'
Expand Down Expand Up @@ -247,3 +248,18 @@ jobs:
export XDG_STATE_HOME="$sandbox_home/.state"
export RUST_TEST_THREADS=1
cargo test --test proxy_database -- --nocapture

- name: Terminal hangup
working-directory: src-tauri
run: |
sandbox_home="$(mktemp -d)"
export HOME="$sandbox_home"
export USERPROFILE="$sandbox_home"
export CC_SWITCH_CONFIG_DIR="$sandbox_home/.cc-switch"
export CLAUDE_CONFIG_DIR="$sandbox_home/.claude"
export CODEX_HOME="$sandbox_home/.codex"
export XDG_CONFIG_HOME="$sandbox_home/.config"
export XDG_RUNTIME_DIR="$sandbox_home/.runtime"
export XDG_STATE_HOME="$sandbox_home/.state"
export RUST_TEST_THREADS=1
cargo test --test tui_terminal_hangup -- --nocapture
29 changes: 20 additions & 9 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -92,15 +92,26 @@ The proxy command surface is in `src/cli/commands/proxy.rs`, orchestration lives

Request handling flows through `HandlerContext`, `ProviderRouter`, `RequestForwarder`, provider adapters in `src/proxy/providers/`, and response builders/handlers in `src/proxy/response*.rs`. Claude `/v1/messages` traffic may be transformed between Anthropic and OpenAI-compatible formats; Codex/OpenAI, Gemini, Copilot, and streaming-response routes are handled by provider-specific adapters. Proxy tests are split across focused integration targets such as `proxy_claude_streaming`, `proxy_claude_openai_chat`, `proxy_claude_response_parity`, `proxy_claude_forwarder_alignment`, `proxy_multi_app_passthrough`, `proxy_takeover`, `proxy_service`, and `proxy_daemon`.

## Blind review protocol

- Before development, define the requirements, acceptance criteria, and task boundaries. Keep changes minimal and avoid unrelated refactoring or behavior changes.
- After implementation and local validation are complete, send the full change set to one new subagent for blind review.
- Start every reviewer with fresh context, without inheriting the development conversation. Provide only the user's goal, intended behavior, acceptance criteria, task boundaries, relevant constraints, repository location, and comparison baseline. Before review begins, tell the user the requirements, criteria, and boundaries given to the reviewer.
- Do not disclose implementation details, change summaries, self-assessments, or findings from prior review rounds. Ask the reviewer to independently inspect all current modifications and report correctness, regression, security, performance, UX, and test-coverage issues.
- Validate every finding against the code. Make minimal fixes for confirmed issues within scope, run the relevant checks, then start another blind review with a new subagent under the same context rules.
- If a confirmed finding also exists upstream, fixing it would expand the task, and compatibility is the intended behavior, it may remain unchanged. Record that decision and prefer upstream parity over an unrelated local fix.
- If review and repair keep cycling without convergence, stop patching and reconsider the overall design and task boundaries. If that still cannot produce a sound resolution, stop expanding the change set and report the unresolved issues, tradeoffs, and evidence to the user.
## Independent blind review gate

Before development, define the requirements, acceptance criteria, and task boundaries. Keep changes minimal and avoid unrelated refactoring or behavior changes. When a confirmed issue also exists upstream and fixing it would expand the task, prefer upstream parity over an unrelated local fix.

Apply this gate after any code change, before declaring the work complete. Changes limited to documentation, comments, formatting or translated text skip it.

1. Freeze a review range the reviewer can reproduce, and do not commit or edit while a reviewer is running: normally `main...HEAD` for committed work, or, for uncommitted work, the working tree against its merge base with `main`, with new files marked by `git add -N`. Make sure every in-scope modification is visible in that range.
2. Choose reviewers for an implementation's first round by risk. By default, start one fresh, independent subagent with no inherited conversation context. Start two such reviewers when the change touches database schema or migrations, writes to live app configuration files, proxy routing or failover, concurrency, credentials or other security-sensitive paths, or import, export and sync of user data. Every reviewer inspects that round's complete range.
3. Give each reviewer detailed but neutral context: the user's goal, intended behavior, acceptance criteria, task boundaries, relevant constraints, the repository path, repository layout, architectural boundaries, authoritative contracts or documentation, the exact review range, relevant read-only validation commands, and the dimensions to inspect (correctness, regressions, data integrity, concurrency, security, performance, compatibility, UI state, i18n, and tests). Before the review starts, tell the user the requirements, criteria and boundaries given to the reviewer.
4. Do not tell reviewers the implementation approach, what changes or fixes were made, the chosen design rationale, a self-assessment, known concerns, prior review findings, or another reviewer's conclusions. Reviewers must independently infer how the code attempts to meet the stated requirements.
5. Reviewers are read-only. Require prioritized findings with severity (blocker, high, medium or low; data integrity is a category at any severity), confidence, an exact file and line, a concrete failure scenario, reasoning, and a recommended direction. Require an explicit "no findings" result when applicable. Style-only comments are not findings unless they create a maintenance or correctness risk.
6. The primary agent must independently validate every finding against the source and tests. Do not apply reviewer suggestions mechanically. A problem the change causes or worsens is in scope wherever it appears. A validated pre-existing or unrelated problem that cannot be fixed within the task boundaries is recorded and reported to the user instead of fixed, and does not block completion. When an in-scope finding is deliberately left unfixed, such as to keep upstream-compatible behavior, record the decision and its reason and list it in the final report. Later reviewers receive deliberate decisions as intended behavior, and known out-of-boundary problems as task boundaries, among the constraints in step 3. Leaving an in-scope finding unfixed never satisfies step 10: an unfixed validated blocker, high-severity or data-integrity finding still blocks completion.
7. Fix confirmed in-scope issues, confirm each fixed failure scenario with a test or a reproduction, and run proportionate regression tests. After fixing a blocker, high-severity or data-integrity finding, or after any other material fix, review the changes made since the reviewed range with one new blind reviewer, started and briefed as in steps 2 to 5; when uncommitted new changes cannot be separated, review the whole working-tree range again.
8. Track findings by underlying invariant or failure scenario, not wording. Run at most three review rounds for one implementation, and end the gate in one of two ways:
- If the third round's validated in-scope findings are all medium or low and none concerns data integrity, finish without another round: fix the small, localized findings, confirm each with a test or a reproduction, and list the rest in the final report.
- If the third round reports a validated in-scope blocker, high-severity or data-integrity finding, or earlier if review and repair begin cycling, patches contradict earlier patches, or the same invariant repeatedly fails, stop patching and reassess the design: ownership, component boundaries, state model, data flow, contracts, and test strategy. A corrected design that fits within the task boundaries, which come from the user, goes through this gate once as a new implementation. If the design holds, or the task boundaries themselves look wrong, stop and report to the user.
9. If the design-level correction reaches either condition of the second case, stop and report the unresolved condition, review history, conflicting constraints, and viable options. Never start a second correction or broaden the task.
10. Completion requires no validated in-scope blocker, high-severity or data-integrity finding, all relevant checks passing, and a final primary-agent audit of the complete diff.

The reviewer count is a review-quality gate, not a requirement to split the implementation into multiple pull requests.

## Testing requirements

Expand Down
Loading
Loading