Oracle flagged a regression introduced in PR #3507 commit 21554be8:
event.ts routed session.error through tmux pane cleanup BEFORE the
existing session-recovery / model-fallback logic ran.
Problem: when session.error was recoverable (context window limit,
quota rate limit, provider fallback), the recovery/fallback code would
successfully continue the SAME session - but by then its tmux pane had
already been destroyed. User-visible symptom is exactly the original
complaint - 'screen appears but streaming stops working' after an
auto-retry.
Fix is the minimal revert: remove the onSessionError funnel from
event.ts and drop onSessionError from the manager. Fatal errors that
actually end a session still fire session.deleted, which continues to
trigger cleanup correctly. Non-fatal error streams stay attached to
the surviving pane.
Three defects observed with tmux.isolation="session" where the omo-agents
session was left with orphan fish panes after subagents finished:
1. cleanup() never ran 'tmux kill-session -t omo-agents'. If any pane
lingered (for example because opencode attach stayed blocked on SSE),
the isolated session survived process shutdown. Now we explicitly kill
the shared session through killTmuxSessionIfExists when isolation is
"session".
2. session.error events bypassed tmux cleanup entirely. Only session.deleted
closed panes, so any provider error that did not escalate into a delete
left the pane behind. Added onSessionError on TmuxSessionManager, wired
from plugin/event.ts, which funnels through the same onSessionDeleted
close path for tracked sessions only.
3. retryPendingCloses() only ran when a new session was created. If the
main process went idle after a failed close, the pending session stayed
pending forever. TmuxPollingManager now accepts the retry callback and
fires it on every tick, alongside the existing stability-based close
sweep.
Manager tests cover isolation=session kill invocation, inline/window
isolation skipping the kill, the onSessionError happy + untracked paths,
and an isolated-session kill failure that must not break cleanup.
Updates the canonical Anthropic Opus model in every fallback chain
(sisyphus, oracle, prometheus, metis, momus, visual-engineering,
ultrabrain, deep, artistry, unspecified-high), the unspecified-high
category default, the think-mode HIGH_VARIANT_MAP, the Claude Code
alias map, the claude-thinking legacy alias, the context-limit GA
regex, and event.ts fallback strings.
Widens supportsCachedAnthropicLimit to accept both claude-*-4-6 and
claude-*-4-7 so the 1M context cache still applies across the bump.
Regenerates the bundled model-capabilities snapshot from models.dev
and the model-fallback snapshot to match the new source output.
- #3354: Coerce data.name to String in loadSkillFromPath/loadSkillFromPathAsync
to prevent crash when YAML parses numeric skill names (e.g., name: 12306)
- #3416: Add required run_in_background parameter to all task() examples in
ultrawork prompts (default, gpt, gemini, planner) to match tool schema
- #3379/#3417/#3418/#3337/#3335: Strip ZWSP (U+200B) before agent name
comparisons in agent-tool-restrictions, sync-prompt-sender, tool-execute-after,
tool-execute-before, oracle-verification-detector, call-omo-agent,
recovery-prompt-config, and agent-variant to prevent ZWSP-prefixed display
names from breaking exact-match lookups
The dedup logic dropped real session.idle events when a synthetic
idle had fired within the dedup window, losing data that downstream
hooks needed. Now real idle events always reach dispatchToHooks while
synthetic duplicates are still dropped when a real event came first.
🤖 Generated with OhMyOpenCode assistance
https://github.com/code-yeongyu/oh-my-opencode
Extract hardcoded GPT apply_patch permission logic into a reusable module
to ensure consistent behavior across all agents. This prevents GPT models
from using the unreliable apply_patch tool while allowing other models.
- Add gpt-apply-patch-guard.ts with GPT_APPLY_PATCH_GUIDANCE and getGptApplyPatchPermission
- Update Hephaestus agent to use centralized permission logic
- Update Sisyphus-Junior agent to use centralized permission logic
- Update all GPT prompt builders to reference shared guidance constant
🤖 Generated with assistance of OhMyOpenCode
The /stop-continuation command was ineffective because the stop-
continuation-guard cleared its stopped state on the very next
chat.message event. Since any user message (including normal chat
after stopping) triggers chat.message, the continuation would
resume immediately.
Root cause: the chat.message handler called clear(sessionID) on
every user message, treating it as a 'user resumed work' signal.
But the user expects /stop-continuation to persist until they
explicitly start work again.
Changes:
- stop-continuation-guard chat.message: no longer clears stop state
- tool-execute-before: /start-work, /ralph-loop, /ulw-loop now
explicitly clear the stop state (so continuation resumes when
user intentionally restarts work)
- Updated and added tests: 12 pass (3 new), 125 related tests pass
Closes#3276
AGENT_LIST_SORT_PREFIXES prepended U+200B Zero Width Space characters
to the four core agent display names so they would sort ahead of user
agents in the Tab cycle. Two problems with that approach surfaced:
1. Some terminal emulators (Ghostty, certain Windows Terminal
builds) render ZWSP as a visible box or extra space, producing a
visible black gap in the status bar before "Sisyphus" and
misaligning the layout (#3259).
2. The prefixes leaked into the plugin API surface via config.agent
keys, breaking prompt_async consumers that received
ZWSP-contaminated agent names (#3238).
#3242 already removed every call site of getAgentListDisplayName() in
production code. That made the sort prefixes dead code: the constant
table was still defined but nothing read it. This PR finishes the
cleanup by:
- Deleting the AGENT_LIST_SORT_PREFIXES constant entirely
- Turning getAgentListDisplayName() into a thin alias over
getAgentDisplayName() for BC with external importers
- Keeping stripAgentListSortPrefix() as a legacy data migration for
users upgrading from v3.14.0-v3.16.0 whose config.agent keys may
still have ZWSP baked in from the old code path
- Documenting the history on stripAgentListSortPrefix() so future
maintainers understand why the stripper has to stay even after
the injector is gone
Sort ordering is preserved via JS object insertion order in
reorderAgentsByPriority() plus the `order` field it injects on the
four core agents. Both mechanisms are already in place and both
pre-date this PR; the ZWSP prefix was an older third layer that was
only meant to work around alphabetical sorting in legacy OpenCode
before the `order` field landed upstream.
Tests: 4445 pass, 0 fail. Added 3 new assertions to
agent-display-names.test.ts verifying that getAgentListDisplayName
returns plain names containing no zero-width characters. Updated
chat-message.test.ts to use a literal ZWSP string instead of the
helper so the defensive-strip path still has coverage.
Closes#3259
- Update plugin-interface.test.ts expectations to use getAgentListDisplayName
- Update chat-message.test.ts expectations for agent display names
- Add smoke test for quoted plan name resolution in chat-message.test.ts
- Aligns test expectations with proper agent name formatting
🤖 Generated with OhMyOpenCode assistance
When users configure agents.sisyphus-junior.model, category sessions
should use that model without falling back to hardcoded
AGENT_MODEL_REQUIREMENTS. Two leaks existed:
1. Launch-time: category-resolver.ts did not check overrideModel when
deciding whether to use the hardcoded fallbackChain. Fixed by adding
overrideModel to the suppression condition.
2. Runtime retry: model-fallback hook.ts treated undefined fallbackChain
(from setSessionFallbackChain(id, undefined)) as 'no override' and
fell through to AGENT_MODEL_REQUIREMENTS on session.error/status
retry. Fixed by storing explicit empty array and checking .has()
instead of truthiness.
Tests added:
- hook.test.ts: explicit empty session fallback suppresses agent chain
- event.model-fallback-2941.test.ts: session.error does not arm retry
- category-resolver.test.ts: overrideModel suppresses fallbackChain
- tools.test.ts: sisyphusJuniorModel override assertion
Display names with parentheses like 'Atlas (Plan Executor)' cause HTTP
header validation errors in x-opencode-agent-name. This was blocking
Atlas/Prometheus from working via /start-work and auto-retry.
Changes:
- Display names: parens -> dashes ('Atlas - Plan Executor')
- Hooks (start-work, no-hephaestus-non-gpt, no-sisyphus-gpt): use
config keys ('atlas', 'sisyphus', 'hephaestus') for agent API fields
- auto-retry: use config key instead of display name for promptAsync
- agent-override-protection: handle dash-suffix normalization
- Updated all test expectations to match new format
Closes#3138
Prevent ZWSP sort prefixes from leaking into stored agent names, config
key lookups, and override-protection normalization. Ensures prefixed
list-display names resolve correctly throughout the pipeline.
🤖 Generated with [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode) assistance
- Add afterAll(() => { mock.restore() }) to 52 test files missing cleanup
- Rewrite create-tool-guard-hooks.test.ts to use spyOn instead of barrel mock
- Fix skill-mcp-manager OAuth tests with missing mockTokens/mockLogin definitions
- Fix start-work hook: show worktree active block on resume with existing worktree_path
- Extract createWorktreeActiveBlock to worktree-block.ts to avoid circular import
- Replace 80-line isolated test runner CI config with single `bun test` command
Fix parseRawLoopSlashCommand to correctly extract and parse loop commands
that appear after injected background task messages (e.g., "[BACKGROUND
TASK COMPLETED]"). Previously, the function only checked if the entire
message started with a slash command, failing when injected content
preceded the command.
🤖 Generated with assistance of OhMyOpenCode (https://github.com/code-yeongyu/oh-my-opencode)
YeonGyu-Kim