setSessionFallbackChain stored the fallback chain but the sync path
never consumed it when sendSyncPrompt returned an error. Added a
retry loop that iterates getNextFallback() until the prompt succeeds
or the chain is exhausted, preserving the finally-block cleanup.
🤖 Generated with OhMyOpenCode assistance
https://github.com/code-yeongyu/oh-my-opencode
The sync poller returned 'Task aborted' immediately when the abort
signal fired, even when a terminal assistant message had already
arrived during the previous wait. Now attempts one final fetch and
returns the completion result if available before emitting the
abort message.
🤖 Generated with OhMyOpenCode assistance
https://github.com/code-yeongyu/oh-my-opencode
- context-info-builder referenced a non-existent /plan command;
now directs users to the Prometheus agent for planning
- skill tool description example referenced 'code-review' which
does not exist; changed to 'review-work' (actual built-in skill)
- skill tool execute path now surfaces the missing host permission
gap so callers understand OpenCode plugin context limits
🤖 Generated with OhMyOpenCode assistance
https://github.com/code-yeongyu/oh-my-opencode
Anthropic API accepts claude-opus-4.6 (dot format) but not
claude-opus-4-6 (dash format). Added anthropic case to
transformModelForProvider to normalize dash-format model IDs before
they reach the provider. Updated model config snapshots and test
expectations to match the new dot-format output.
🤖 Generated with OhMyOpenCode assistance
https://github.com/code-yeongyu/oh-my-opencode
- Replace getAgentRuntimeName with getAgentDisplayName for consistency
- Add stripAgentListSortPrefix helper to normalize agent names
- Strip sort prefixes in subagent-resolver and sync-executor
- Backfill canonical names for core agents when builtin configs omit name
- Update tests to match new behavior
Update various modules to use centralized constants from plugin-identity:
- get-local-version/formatter: Use PUBLISHED_PACKAGE_NAME
- run/session-resolver: Use PUBLISHED_PACKAGE_NAME
- background-agent/task-poller: Use PUBLISHED_PACKAGE_NAME
- mcp-oauth/provider: Use PUBLISHED_PACKAGE_NAME
- auto-update-checker/constants: Use ACCEPTED_PACKAGE_NAMES
- comment-checker/downloader: Use PUBLISHED_PACKAGE_NAME
- legacy-plugin-toast/hook: Use PLUGIN_NAME
- shared/data-path: Use CACHE_DIR_NAME
- shared/external-plugin-detector: Use ACCEPTED_PACKAGE_NAMES
- shared/logger: Use LOG_FILENAME
- tools/ast-grep/downloader: Use PUBLISHED_PACKAGE_NAME
- tools/call-omo-agent/tools: Use PUBLISHED_PACKAGE_NAME
- tools/delegate-task/category-resolver: Use PUBLISHED_PACKAGE_NAME
- tools/grep/constants: Use PUBLISHED_PACKAGE_NAME
- tools/grep/downloader: Use PUBLISHED_PACKAGE_NAME
- tools/lsp/lsp-client-wrapper: Use PUBLISHED_PACKAGE_NAME
🤖 Generated with assistance of OhMyOpenCode
The depth limit (default maxDepth=3) was being silently bypassed when
sync-task.ts could not reach the manager's spawn enforcement methods --
the fallback hardcoded childDepth: 1, allowing infinite recursion of
delegate_task calls in degraded environments.
This was hard to catch because:
1. The fallback path took the dangerous default silently (no log).
2. There were no end-to-end smoke tests asserting that the depth value
coming back from reserveSubagentSpawn is actually used.
3. The unit tests for resolveSubagentSpawnContext only covered error
cases, not the actual depth calculation.
Changes:
- sync-task.ts: split the spawnContext fallback into an explicit if/else
with a WARNING log when the manager is missing enforcement methods.
This makes the dangerous path observable in logs.
- subagent-spawn-limits.test.ts: add depth calculation regression tests
(root, depth-1, depth-2, depth at max, parent cycle detection).
- sync-task.test.ts: add two regression smoke tests:
1. depth limit error from reserveSubagentSpawn must be propagated and
must NOT create the session.
2. spawnDepth recorded in metadata must equal what reserveSubagentSpawn
returns -- guards against silent fallback to childDepth: 1.
15 new spawn-limits tests + 2 new sync-task tests pass.
Full suite: 5105 pass, 0 fail.
prompt-builder.ts now includes project-level skills from
.opencode/skills/ when building delegated session prompts.
5007 tests pass, tsc clean.
Closes#2687
model-selection.ts now separates model selection from variant/reasoning
tier, so agent model overrides don't lock the variant.
34 tests pass, 5009 total, tsc clean.
Closes#3163
Added await for session ready state before sending prompt in
background-agent/manager.ts. Also improved image resizer error handling.
132 tests pass, tsc clean.
Closes#2932
When Read tool is disabled (attachment mode), the agent prompt now
instructs to analyze the attached content directly instead of
trying to use Read. 3 new tests, 55 total pass.
Closes#3033
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
When weaker models (GLM-5, MiniMax) omit the description parameter on
delegate_task, the tool now auto-generates it from the first 4 words of
the prompt instead of throwing an error.
Changes:
- Schema: description is now optional (tool.schema.string().optional())
- Runtime: auto-generates from prompt when missing/empty/whitespace
- DelegateTaskArgs.description type stays as string (guaranteed by auto-gen)
- Tests: 3 new cases - missing/empty/explicit description handling
- Metadata title set after description resolution (correct ordering)
Tool return values contained CTAs ("Use background_output to check") that
caused models to immediately poll instead of waiting for <system-reminder>
notifications. Unified all 5 tool return messages with explicit "Do NOT call
background_output now" guard, added pre-notification warning to BackgroundOutput
tool description, and strengthened Background Result Collection sections across
all 3 Sisyphus prompt variants (default, gpt-5-4, main) with BLOCKING
anti-pattern enforcement.
🤖 Generated with assistance of [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode)
zod v3 infers .default([]) fields as string[] | undefined in z.infer
output type (unlike v4 which marks them as string[]). Add nullish
coalescing guards and use any[] for the readJsonSafe result array
to avoid the type mismatch in task-list and task-update.
OpenCode's fromPlugin wrapper skips Zod validation for plugin tools, so
LLMs can omit required args like description without getting an error.
When Atlas orchestrates and the model omits description, it flows through
as undefined to manager.launch() and background task notifications show
'undefined' for all completed tasks.
Two fixes:
- Add runtime validation for description in delegate-task tool (matches
existing run_in_background and load_skills validation pattern)
- Defensive fallback in notification template: use task ID when
description is missing instead of rendering 'undefined'
- Remove kimi from auto-detected unstable agents in category-resolver (only gemini/minimax remain)
- Add resolveCallID() to safely resolve callID/callId/call_id variants from tool context
- Use resolveCallID across all 5 delegate task execution paths (sync, background, unstable, continuations)
- Update writing category test to verify kimi runs sync when kimi provider is available
- Add atlas metadata preservation tests confirming tool-execute-after does not clobber metadata
Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
YeonGyu-Kim