Add guardrails and progress tracking to Jr agent

Implement runtime guardrails in Engine.hs: - Cost budget limit (default
200 cents) - Token budget limit (default 1M tokens) - Duplicate tool
call detection (same tool called N times) - Test failure counting
(bild --test failures)

Add database-backed progress tracking: - Checkpoint events stored in
agent_events table - Progress summary retrieved on retry attempts -
Improved prompts emphasizing efficiency and autonomous operation

Worker.hs improvements: - Uses guardrails configuration - Reports
guardrail violations via callbacks - Better prompt structure for
autonomous operation

Task-Id: t-203

1 files changed, 23 insertions, 0 deletions

diff --git a/Omni/Task/Core.hs b/Omni/Task/Core.hs
index f54cf81..1212a56 100644
--- a/Omni/Task/Core.hs
+++ b/Omni/Task/Core.hs
@@ -1702,3 +1702,26 @@ getEventsSince sessionId lastId =
       "SELECT id, task_id, session_id, timestamp, event_type, content \
       \FROM agent_events WHERE session_id = ? AND id > ? ORDER BY id ASC"
       (sessionId, lastId)
+
+-- | Insert a checkpoint event (for progress tracking)
+insertCheckpoint :: Text -> Text -> Text -> IO ()
+insertCheckpoint taskId sessionId =
+  insertAgentEvent taskId sessionId "Checkpoint"
+
+-- | Get all checkpoints for a task (across all sessions)
+getCheckpointsForTask :: Text -> IO [StoredEvent]
+getCheckpointsForTask taskId =
+  withDb <| \conn ->
+    SQL.query
+      conn
+      "SELECT id, task_id, session_id, timestamp, event_type, content \
+      \FROM agent_events WHERE task_id = ? AND event_type = 'Checkpoint' ORDER BY id ASC"
+      (SQL.Only taskId)
+
+-- | Get progress summary for a task (concatenated checkpoint contents)
+getProgressSummary :: Text -> IO (Maybe Text)
+getProgressSummary taskId = do
+  checkpoints <- getCheckpointsForTask taskId
+  if null checkpoints
+    then pure Nothing
+    else pure <| Just <| T.intercalate "\n\n---\n\n" [storedEventContent e | e <- checkpoints]