| #379 |
MEDIUM: ToolRegistry singleton not thread-safe
registry.py:305-320 - get_tool_registry() singleton creation has race condition. Multiple threads co...
|
closed |
medium |
2025-12-11 21:54 |
- |
|
| #378 |
MEDIUM: SchedulerService breaks atomic transaction boundary
scheduler_service.py uses autocommit=True but needs atomic scan+submit+update operation. Uses differ...
|
closed |
medium |
2025-12-11 21:54 |
- |
|
| #377 |
HIGH: WaitOperator uses non-deterministic datetime.now()
operators.py:989 uses datetime.now(timezone.utc) to calculate wake_at. On replay, different wake tim...
|
closed |
high |
2025-12-11 21:54 |
- |
|
| #376 |
HIGH: ForEach results lost on resume
operators.py:601-624 - Resumed ForEach loops reinitialize results=[]. Previously completed iteration...
|
closed |
high |
2025-12-11 21:54 |
- |
|
| #375 |
HIGH: SQL injection pattern in platform.py tenant deletion
platform.py:1083-1121 uses f-strings for table names in DELETE queries. Should use psycopg.sql.Ident...
|
closed |
high |
2025-12-11 21:54 |
- |
|
| #374 |
HIGH: Shell command injection via unescaped variables
shell_command.py:278-281 resolves variables and passes directly to shell=True subprocess. If {{user_...
|
closed |
high |
2025-12-11 21:54 |
- |
|
| #373 |
CRITICAL: Replace unsafe eval() with safe expression parser
variable_resolver.py:177-186 uses eval() with empty __builtins__ for condition evaluation. This is i...
|
closed |
critical |
2025-12-11 21:54 |
- |
|
| #372 |
LOW: Magic numbers should be configurable
## Issue
Hardcoded values throughout codebase should be in configuration:
- activity_threshold_sec...
|
closed |
low |
2025-12-11 21:40 |
- |
|
| #371 |
LOW: Logger message missing f-string prefix
## Issue
In engine/orchestrator.py:577, logger message uses wrong format string:
```python
logger....
|
closed |
low |
2025-12-11 21:40 |
- |
|
| #370 |
MEDIUM: _save_state not called automatically before sleep/commit
## Issue
The _save_state() method in engine/durable_context.py:1462-1496 persists executed_tasks, f...
|
closed |
medium |
2025-12-11 21:40 |
- |
|
| #369 |
MEDIUM: Circuit breaker cache uses file system flag for reset
## Issue
Circuit breaker cache reset in engine/tools/shell_command.py:63-84 relies on flag file /tm...
|
closed |
medium |
2025-12-11 21:39 |
- |
|
| #368 |
MEDIUM: Silent encryption failure (fail-open security)
## Issue
In engine/db.py:131-147, encryption configuration failures are silently swallowed:
```pyt...
|
closed |
medium |
2025-12-11 21:39 |
- |
|
| #367 |
MEDIUM: Connection leak in get_raw_db_connection docstring example
## Issue
The docstring for get_raw_db_connection() in engine/db.py:515-547 shows improper usage:
`...
|
closed |
medium |
2025-12-11 21:39 |
- |
|
| #366 |
HIGH: TenantCache TTL creates stale tenant resolution
## Issue
TenantCache in engine/orchestrator.py:55-110 has 5-minute TTL. If a workflow run is delete...
|
closed |
high |
2025-12-11 21:39 |
- |
|
| #365 |
HIGH: Global determinism state shared across workflows
## Issue
Global flags in engine/durable_context.py:39-43 are NOT thread-safe and shared across all ...
|
closed |
high |
2025-12-11 21:39 |
- |
|
| #364 |
HIGH: ForEach/While loop checkpoints not persisted to database
## Issue
ForEach and While loop operators save checkpoints to ctx._checkpoints_cache but do NOT cal...
|
closed |
high |
2025-12-11 21:39 |
- |
|
| #363 |
CRITICAL: sleeping_tasks.discard() called on dict - AttributeError
## Issue
In engine/durable_context.py:1389, the code calls discard() on sleeping_tasks which is a d...
|
closed |
critical |
2025-12-11 21:39 |
- |
|
| #362 |
CRITICAL: Heartbeat updates not committed to database
## Issue
The heartbeat service in engine/services/heartbeat_service.py:179-207 updates heartbeats w...
|
closed |
critical |
2025-12-11 21:39 |
- |
|
| #361 |
CRITICAL: ctx.now is non-deterministic - breaks replay guarantees
## Issue
The ctx.now property in engine/durable_context.py:486-503 returns datetime.now(timezone.ut...
|
closed |
critical |
2025-12-11 21:39 |
- |
|
| #360 |
CRITICAL: DataShard script storage uses workflow_name+version instead of definition_hash - causes stale data collisions
## Problem
When storing Python DSL scripts in DataShard, the key is based on workflow_name + version...
|
closed |
critical |
2025-12-11 10:26 |
- |
|