feat(kernel): enforce stage allowedTools at tool-call dispatch
A stage's allowedTools only shaped which tool definitions were offered to the model; the dispatch path resolved returned tool calls straight from the full registry and executed them without checking stage membership. A hallucinated, jailbroken, or edited-transcript tool call for any registered tool would run unchecked (violating invariant #5 / policy-is-absolute). dispatchToolCalls now rejects any tool call whose name is not in the stage's allowedTools (STAGE_COMPLETE_TOOL exempt), emitting ToolExecutionRejectedEvent and feeding an error ContextEntry back to the model instead of executing. Empty allowedTools means deny-all domain tools, consistent with offering only STAGE_COMPLETE_TOOL at inference time.
This commit is contained in:
+36
@@ -406,6 +406,42 @@ abstract class SessionOrchestrator(
|
||||
request = request,
|
||||
),
|
||||
)
|
||||
if (toolCall.function.name != STAGE_COMPLETE_TOOL &&
|
||||
toolCall.function.name !in stageConfig.allowedTools
|
||||
) {
|
||||
val denyReason = "tool '${toolCall.function.name}' is not permitted in stage '${stageId.value}'"
|
||||
emit(
|
||||
sessionId,
|
||||
ToolExecutionRejectedEvent(
|
||||
invocationId = invocationId,
|
||||
sessionId = sessionId,
|
||||
toolName = toolCall.function.name,
|
||||
tier = tier,
|
||||
reason = denyReason,
|
||||
),
|
||||
)
|
||||
val sourceId = toolCall.id ?: invocationId.value
|
||||
return@flatMap listOf(
|
||||
ContextEntry(
|
||||
id = ContextEntryId(UUID.randomUUID().toString()),
|
||||
layer = ContextLayer.L2,
|
||||
sourceType = "assistantToolCall",
|
||||
sourceId = sourceId,
|
||||
content = Json.encodeToString(ToolCallRequest.serializer(), toolCall),
|
||||
tokenEstimate = estimateTokens(toolCall.function.arguments),
|
||||
role = EntryRole.ASSISTANT,
|
||||
),
|
||||
ContextEntry(
|
||||
id = ContextEntryId(UUID.randomUUID().toString()),
|
||||
layer = ContextLayer.L2,
|
||||
sourceType = "toolResult",
|
||||
sourceId = sourceId,
|
||||
content = "ERROR: $denyReason",
|
||||
tokenEstimate = estimateTokens(denyReason),
|
||||
role = EntryRole.TOOL,
|
||||
),
|
||||
)
|
||||
}
|
||||
val plane2Risk: RiskSummary? = runPlane2Assessment(
|
||||
sessionId, stageId, invocationId, toolCall.function.name, request, tool, effectives,
|
||||
)?.let { assessment ->
|
||||
|
||||
@@ -15,6 +15,7 @@ import com.correx.core.artifacts.DefaultArtifactReducer
|
||||
import com.correx.core.events.events.ApprovalRequestedEvent
|
||||
import com.correx.core.events.events.ToolCallAssessedEvent
|
||||
import com.correx.core.events.events.ToolExecutionRejectedEvent
|
||||
import com.correx.core.events.events.ToolInvocationRequestedEvent
|
||||
import com.correx.core.events.events.ToolRequest
|
||||
import com.correx.core.events.execution.RetryPolicy
|
||||
import com.correx.core.events.risk.RiskAction
|
||||
@@ -371,4 +372,57 @@ class ToolCallGateTest {
|
||||
val toolEntry = secondPack.layers.values.flatten().first { it.sourceType == "toolResult" }
|
||||
assertEquals("line1\nline2\nline3", toolEntry.content)
|
||||
}
|
||||
|
||||
@Test
|
||||
fun `tool not in allowedTools is not executed and emits ToolExecutionRejectedEvent`(): Unit = runBlocking {
|
||||
val executor = RecordingExecutor()
|
||||
// Stage only allows "allowed_tool"; the LLM returns a call to "file_write" (not in the set).
|
||||
val (orchestrator, eventStore, _) = buildOrchestrator(executor, FakeFileWriteTool(), assessorRule = null)
|
||||
val sessionId = SessionId("gate-stage-deny")
|
||||
val config = OrchestrationConfig(retryPolicy = RetryPolicy(maxAttempts = 1, backoffMs = 0))
|
||||
val graph = singleStageGraph(allowedTools = setOf("allowed_tool"))
|
||||
|
||||
orchestrator.run(sessionId, graph, config)
|
||||
|
||||
val events = eventStore.read(sessionId)
|
||||
assertTrue(!executor.executeCalled.get(), "executor must NOT be called for a tool outside allowedTools")
|
||||
|
||||
// Denial is recorded in the event log (invariant #5).
|
||||
val rejected = events.firstNotNullOfOrNull { it.payload as? ToolExecutionRejectedEvent }
|
||||
assertNotNull(rejected, "ToolExecutionRejectedEvent must be emitted when tool is not in allowedTools")
|
||||
assertEquals("file_write", rejected!!.toolName)
|
||||
assertTrue(
|
||||
rejected.reason.contains("not permitted"),
|
||||
"rejection reason should mention 'not permitted', got: ${rejected.reason}",
|
||||
)
|
||||
|
||||
// ToolInvocationRequestedEvent is still emitted so the request itself is in the log.
|
||||
assertTrue(
|
||||
events.any { (it.payload as? ToolInvocationRequestedEvent)?.toolName == "file_write" },
|
||||
"ToolInvocationRequestedEvent must be emitted to record the denied request",
|
||||
)
|
||||
}
|
||||
|
||||
@Test
|
||||
fun `empty allowedTools denies all domain tools and emits ToolExecutionRejectedEvent`(): Unit = runBlocking {
|
||||
val executor = RecordingExecutor()
|
||||
// Stage has allowedTools = emptySet() — only stage_complete is permitted.
|
||||
val (orchestrator, eventStore, _) = buildOrchestrator(executor, FakeFileWriteTool(), assessorRule = null)
|
||||
val sessionId = SessionId("gate-empty-allowed")
|
||||
val config = OrchestrationConfig(retryPolicy = RetryPolicy(maxAttempts = 1, backoffMs = 0))
|
||||
val graph = singleStageGraph(allowedTools = emptySet())
|
||||
|
||||
orchestrator.run(sessionId, graph, config)
|
||||
|
||||
val events = eventStore.read(sessionId)
|
||||
assertTrue(!executor.executeCalled.get(), "executor must NOT be called when allowedTools is empty")
|
||||
|
||||
val rejected = events.firstNotNullOfOrNull { it.payload as? ToolExecutionRejectedEvent }
|
||||
assertNotNull(rejected, "ToolExecutionRejectedEvent must be emitted when allowedTools is empty")
|
||||
assertEquals("file_write", rejected!!.toolName)
|
||||
assertTrue(
|
||||
rejected.reason.contains("not permitted"),
|
||||
"rejection reason should mention 'not permitted', got: ${rejected.reason}",
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user