- Strengthen Prometheus plan template with FORMAT constraint
- Add task label format check to Oracle phase-2 (N/6 → N/7)
- Add format checks to self-review checklist
- New plan-format-validator hook: compares raw checkbox count
against getPlanProgress() after plan writes, warns agent when
labels are malformed (0/0 or partial skip scenarios)
Adds a new `notepad-write-guard` hook that intercepts Write tool calls
whose target path matches `**/.sisyphus/notepads/**` and throws an
actionable error instead of allowing the write to proceed.
Without this guard, an agent that hits an Edit hash-mismatch failure
could silently fall back to Write, destroying the entire history of an
append-only notepad file (decisions.md, issues.md, etc.). The file
carries an explicit "NEVER overwrite" warning that the agent ignores
under context pressure.
The guard is path-based so it works regardless of plan name or nesting
depth. Non-notepad `.sisyphus/**` paths (e.g. plan files) are
unaffected. The hook is wired into `create-tool-guard-hooks` under the
hook name `notepad-write-guard` and follows the same safeCreateHook +
HookName schema pattern as every other tool-guard hook.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Source code change:
- src/shared/model-requirements.ts: prepend claude-sonnet-4-6 to metis fallback
chain so Sonnet becomes the default. Opus 4.7 max remains as the immediate
fallback for callers who want extra reasoning.
- src/shared/model-requirements.test.ts: update assertion to expect Sonnet
primary + Opus secondary.
AGENTS.md accuracy fixes (verified against source):
- Agent modes: Sisyphus/Hephaestus are 'primary' (not 'all'); Sisyphus-Junior
is 'subagent' (not 'all'). Confirmed via 'const MODE: AgentMode = ...' in
each agent file. Also clarified Prometheus has no agentSources factory and
is built via buildPrometheusAgentConfig.
- Sisyphus fallback chain: corrected order to kimi-k2.6 → k2p5 → kimi-k2.5
→ gpt-5.5 medium → glm-5 → big-pickle (was missing kimi-k2.5).
- Librarian/Explore: added missing minimax-m2.7 step between -highspeed and
claude-haiku-4-5.
- Metis chain: removed fictitious gemini-3.1-pro entry.
- Sisyphus-Junior chain: spelled out the actual fallback (was 'user-configurable').
- Temperatures: Sisyphus/Hephaestus do not set explicit temperature (model
default); Sisyphus-Junior is 0.1 via SISYPHUS_JUNIOR_DEFAULTS.
- Quick category default: gpt-5.4-mini (not gpt-5.4-mini-fast).
Team-mode corrections:
- Eligibility registry has 3 verdicts: eligible (sisyphus, atlas, sisyphus-junior),
conditional (hephaestus — needs D-36 teammate permission), hard-reject
(oracle, librarian, explore, multimodal-looker, metis, momus, prometheus).
- Schema has 11 fields, not 4: added max_messages_per_run, max_wall_clock_minutes,
max_member_turns, base_dir, message_payload_max_bytes, recipient_unread_max_bytes,
mailbox_poll_interval_ms.
- Hooks: 'team-session-events' is 4 sub-handlers in src/plugin/event.ts
(team-idle-wake-hint, team-lead-orphan-handler, team-member-error-handler,
team-member-status-handler), not a single Continuation-tier hook.
- Tier counts now show base + team-mode: ToolGuard 14/15, Transform 5/7.
- Total: 52 base hooks, 59 with team-mode.
Doc cascade for the Metis change:
- docs/guide/orchestration.md, agent-model-matching.md, installation.md
- docs/reference/configuration.md, features.md
Adjacent "hpp ulw" or "ulw hpp" (8 form combinations: short/long, both
orders) triggers a fused mode that suppresses the standalone ultrawork
and hyperplan banners and toasts in favor of one combo banner. The
combo banner explicitly preserves hyperplan's mandatory adversarial
workflow contract (do NOT improvise, do NOT skip rounds) instead of
silently downgrading it.
Suppression runs as a named pipeline step (suppressComboStandalones)
immediately after detection and before all consumers (planner filter,
session filters, toasts, message injection), so standalone toast checks
naturally see the already-suppressed list. Combo is allowed in non-main
sessions like ultrawork, filtered for planner agents like both
standalones, and blocked in subagent sessions via the existing gate.
disabled_keywords uses the intersection rule: disabling either
"ultrawork" or "hyperplan" also disables the combo, so no flavored
content leaks via the combo embedding when either base keyword is
disabled.
Includes a same-PR refactor of KEYWORD_DETECTORS from {pattern, message}
to {type, pattern, message} tuple shape, dropping the parallel hardcoded
types array in detector.ts that previously coupled type assignment to
registry index. Future detector additions can no longer silently corrupt
DetectedKeyword.type via reorder or insertion.
10 behavioral contract tests in hyperplan-ultrawork.test.ts cover both
trigger orders, non-adjacent rejection, suppression of injection and
toast, intersection-rule disable behavior, session/agent policy, and
ultrawork variant routing through the combo. The pre-existing combined
"ultrawork hyperplan" assertion in hyperplan.test.ts is removed in
favor of the new file.
Plan distilled from a hyperplan adversarial review (5 members,
3 rounds: skeptic, validator, researcher, architect, creative).
Adds keyword_detector.disabled_keywords config so users can opt out of
specific keyword detectors individually without disabling the entire
keyword-detector hook. Allowed values: 'ultrawork', 'search', 'analyze',
'team'. Default empty/missing -> all four detectors active (no behavior
change for existing configs).
Motivation: an audit revealed search and analyze patterns trigger on
~30-60% of normal conversational user messages (e.g. 'how to', 'why is',
'show me', '왜', '어떻게'). The disable list is the immediate kill switch
while the patterns themselves are tightened in a separate PR.
Schema follows the existing per-feature config block convention shared
by team_mode, ralph_loop, runtime_fallback, and comment_checker. The
KeywordType enum (z.enum) lives next to the config schema and is
re-imported by the detector to keep the union type in lockstep with the
schema.
Threading:
pluginConfig.keyword_detector
-> create-transform-hooks.ts (factory wiring)
-> createKeywordDetectorHook(config)
-> detectKeywordsWithType(text, agent, model, disabledKeywords)
-> Set-based filter at the source-of-truth detector
Adds 8 regression tests covering per-keyword disable, multi-keyword
disable, partial disable (one keyword off, another still firing),
ultrawork toast suppression, undefined config, and empty array.
Blocking fixes:
- B1: Return empty restrictions for unknown/custom agents instead of
EXPLORATION_AGENT_DENYLIST, allowing custom agents full tool access
- B2: Use Object.create(null) consistently across all 5 agent-loading
result objects to prevent prototype pollution
- B3: Add code comment documenting custom agent bash access trust model
- B4: Mock getOpenCodeConfigDir in opencode-config-agents-reader tests
to prevent global config dir leakage
Non-blocking fixes:
- N1: Use resolveAgentDefinitionPaths with project boundary enforcement
in opencode-config-agents-reader for path containment
- N2: Add session-scoped 30s TTL cache to resolveCallableAgents to
avoid redundant SDK IPC calls per tool invocation
- N3: Extract shared parseToolsConfig into src/shared/parse-tools-config.ts
replacing 4 duplicated local implementations
- N4: Add .min(1) to AgentDefinitionPathSchema rejecting empty paths
- N5: Add resolve-agent-definition-paths.test.ts covering tilde expansion,
relative paths, boundary enforcement, and null containmentDir
- N6: Validate agent mode against allowed values instead of bare type
assertion in opencode-config-agents-reader
Wave 1 of agent definitions enhancement (PR #2299):
Schema & Configuration:
- Add agent_definitions field to oh-my-opencode config schema
- Support list of file paths to .md or .json agent definition files
- Add to PARTIAL_STRING_ARRAY_KEYS for Set-union merge semantics
- Implement eager path resolution in loadPluginConfig() before merging
Path Resolution:
- Create resolve-agent-definition-paths.ts helper
- User-level paths resolve from ~/.config/opencode/ (no containment)
- Project-level paths resolve from project root (with containment check)
- Homedir expansion, absolute/relative path handling
JSON Agent Loader:
- Create parseJsonAgentFile() for .json/.jsonc agent definitions
- Validate required fields (name, prompt)
- Support tools as string (comma-separated) or array
- Map model via mapClaudeModelToOpenCode()
- Comprehensive test suite (7 test cases, all passing)
Type Extensions:
- Extend AgentScope: add 'definition-file' and 'opencode-config'
- Add AgentJsonDefinition interface for JSON agent schema
All automated checks passing:
- lsp_diagnostics clean on all changed files
- json-agent-loader.test.ts: 7/7 passing
- Full typecheck: zero new errors
- QA evidence saved to .sisyphus/evidence/
- Fix zod/v4 imports in background-task schema tests
- Remove ZWSP prefix from agent-key-remapper test (fixed in #3136)
- Use toMatchObject for openai-only catalog tests (fallback_models added by #3144)
- Replace z.toJSONSchema (zod v4) with zodToJsonSchema (zod v3 compat)
- Fix task-list.ts type narrowing for zod v3 inferred types
Adds a defensive tool-pair-validator hook that runs as the final step in
the messages transform pipeline. When compaction or context-window recovery
removes user messages containing tool_result blocks without removing the
preceding assistant message with tool_use blocks, this validator detects
the mismatch and either:
1. Injects missing tool_result parts into the next user message, or
2. Creates a synthetic user message with placeholder tool_results
This prevents Anthropic API errors like 'tool_use ids found without
tool_result blocks immediately after'.
Fixes#3014
Embed user-level skills into the plugin's built-in system so they ship
with the product rather than requiring per-user configuration.
- review-work: 5-agent parallel post-implementation review orchestrator
- ai-slop-remover: per-file AI-generated code smell detector and remover
- /remove-ai-slops: command that orchestrates parallel ai-slop-remover runs
TTL (pruneStaleTasksAndNotifications) now resets on last activity:
- Uses task.progress.lastUpdate as TTL anchor for running tasks
(was always using startedAt, causing 30-min hard deadline)
- Added taskTtlMs config option for user-adjustable TTL
- Error message shows actual TTL duration, not hardcoded '30 minutes'
- 3 new tests for the new behavior
heunghingwan