Skip to content

feat(ops): P0-4 Tracing - OPS_CHAT_TRACER + trace_span for deep/react #231

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Cyning12 merged 5 commits into from
Jul 9, 2026
Merged

feat(ops): P0-4 Tracing - OPS_CHAT_TRACER + trace_span for deep/react #231

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
99a8463
feat(ops): P0-4 Tracing - OPS_CHAT_TRACER + trace_span for deep/react
CyningMM Jul 9, 2026
d664dda
docs(harness): 50 reinspect Prompt for P0-4 Tracing
CyningMM Jul 9, 2026
9b2b054
docs(harness): 50 reinspect R1 P0-4 Tracing audit report
CyningMM Jul 9, 2026
76cac53
build(vercel): pin Python 3.13 + pglast 7.15 to fix Vercel build
CyningMM Jul 9, 2026
c3f8df4
style(ruff): fix I001/F401/F821/UP035 in P0-3/P0-4 tests
CyningMM Jul 9, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .python-version
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
3.13
81 changes: 52 additions & 29 deletions api/ops/orchestrator/core.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,7 @@
from api.ops.queries import OpsQueries
from api.ops.review.rules import review_result
from api.ops.store.runs import OpsRunStore, append_event
from api.ops.tracing import traceable, update_current_span_metadata
from api.ops.tracing import trace_span, traceable, update_current_span_metadata


class Intent:
Expand Down Expand Up @@ -206,6 +206,8 @@ def run_deep(
update_current_span_metadata(
{
"ops_run_id": run_id,
"run_id": run_id,
"agent_role": "deep",
"route": "deep",
"intent": intent or "issue_contribution",
"agent": agent_name,
Expand All @@ -222,18 +224,29 @@ def run_deep(
payload={"route": "deep", "intent": intent or "issue_contribution", "slots": slots, "agent": agent_name},
node_id="classify",
)
append_event(
run_id,
with trace_span(
"handoff",
handoff_payload(
from_route="classify",
to_route="deep",
intent=intent or "issue_contribution",
slots=slots,
agent=agent_name,
),
store=store,
)
run_type="tool",
run_id=run_id,
session_id=session_id,
agent_role="deep",
from_route="classify",
to_route="deep",
intent=intent or "issue_contribution",
agent=agent_name,
):
append_event(
run_id,
"handoff",
handoff_payload(
from_route="classify",
to_route="deep",
intent=intent or "issue_contribution",
slots=slots,
agent=agent_name,
),
store=store,
)

store.append_event(
run_id,
Expand Down Expand Up @@ -282,24 +295,34 @@ def run_deep(
)

verdict, detail = review_result(analyst_result, queries)
store.append_event(
run_id,
"review",
f"review.{verdict}",
payload={"rule": detail.get("rule"), "message": detail.get("message"), "attempt": attempt},
node_id="review",
)
append_event(
run_id,
with trace_span(
"review",
review_payload(
verdict=verdict,
rule=detail.get("rule"),
message=detail.get("message"),
attempt=attempt,
),
store=store,
)
run_type="tool",
run_id=run_id,
session_id=session_id,
agent_role="deep",
verdict=verdict,
rule=detail.get("rule"),
attempt=attempt,
):
store.append_event(
run_id,
"review",
f"review.{verdict}",
payload={"rule": detail.get("rule"), "message": detail.get("message"), "attempt": attempt},
node_id="review",
)
append_event(
run_id,
"review",
review_payload(
verdict=verdict,
rule=detail.get("rule"),
message=detail.get("message"),
attempt=attempt,
),
store=store,
)

if verdict == "pass":
final_verdict = "done"
Expand Down
80 changes: 51 additions & 29 deletions api/ops/react_loop.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@
from api.ops.react_tools import _build_v0_registry, _truncate_summary
from api.ops.review.rules import review_result
from api.ops.store.runs import OpsRunStore, append_event
from api.ops.tracing import traceable, update_current_span_metadata
from api.ops.tracing import trace_span, traceable, update_current_span_metadata

MAX_STEPS_DEFAULT = int(os.getenv("OPS_REACT_MAX_STEPS", "6"))
MAX_RETRIES_DEFAULT = 2
Expand All @@ -41,6 +41,8 @@ def run_react_fallback(
update_current_span_metadata(
{
"ops_run_id": run_id,
"run_id": run_id,
"agent_role": "react",
"route": "react",
"intent": "fallback",
"max_steps": max_steps,
Expand All @@ -59,18 +61,28 @@ def run_react_fallback(
payload={"route": "react", "intent": "fallback", "max_steps": max_steps},
node_id="classify",
)
append_event(
run_id,
with trace_span(
"handoff",
handoff_payload(
from_route="classify",
to_route="react",
intent="fallback",
slots={},
agent=None,
),
store=store,
)
run_type="tool",
run_id=run_id,
session_id=session_id,
agent_role="react",
from_route="classify",
to_route="react",
intent="fallback",
):
append_event(
run_id,
"handoff",
handoff_payload(
from_route="classify",
to_route="react",
intent="fallback",
slots={},
agent=None,
),
store=store,
)

# System prompt for ReAct
system_prompt = _build_react_system_prompt(tools_json)
Expand Down Expand Up @@ -225,24 +237,34 @@ def run_react_fallback(
review_feedback: dict[str, Any] | None = None
while attempt <= max_retries:
verdict, detail = review_result(react_result, queries)
store.append_event(
run_id,
with trace_span(
"review",
f"review.{verdict}",
payload={"rule": detail.get("rule"), "message": detail.get("message"), "attempt": attempt},
node_id="review",
)
append_event(
run_id,
"review",
review_payload(
verdict=verdict,
rule=detail.get("rule"),
message=detail.get("message"),
attempt=attempt,
),
store=store,
)
run_type="tool",
run_id=run_id,
session_id=session_id,
agent_role="react",
verdict=verdict,
rule=detail.get("rule"),
attempt=attempt,
):
store.append_event(
run_id,
"review",
f"review.{verdict}",
payload={"rule": detail.get("rule"), "message": detail.get("message"), "attempt": attempt},
node_id="review",
)
append_event(
run_id,
"review",
review_payload(
verdict=verdict,
rule=detail.get("rule"),
message=detail.get("message"),
attempt=attempt,
),
store=store,
)

if verdict == "pass":
final_verdict = "done"
Expand Down
Loading