Skip to content
Merged
Show file tree
Hide file tree
Changes from 11 commits
Commits
Show all changes
22 commits
Select commit Hold shift + click to select a range
ec3b3c8
feat: add purpose-limited PII event protection
seonghobae Aug 21, 2026
5e06114
fix: harden purpose-limited PII protection
seonghobae Aug 21, 2026
33f312c
fix: bound durable audit retention and CI runtime deps
seonghobae Aug 21, 2026
5c51c3a
fix: isolate authorization audit churn
seonghobae Aug 21, 2026
f13c224
fix: bound durable analytics retention
seonghobae Aug 21, 2026
606eb37
test: prove authorization retention is bounded
seonghobae Aug 21, 2026
9f8b094
test: restore security test lintability
seonghobae Aug 21, 2026
1f83652
docs: describe cryptography runtime dependency
seonghobae Aug 21, 2026
35692d5
fix: harden pii key derivation and retention
seonghobae Aug 21, 2026
26d1d30
fix: reject unprefixed PII encryption keys
seonghobae Aug 21, 2026
f646b04
fix: keep denied-request auditing off the durable-persistence hot path
seonghobae Aug 22, 2026
3e682c4
fix: preserve bounded PII audit streams
seonghobae Aug 23, 2026
7233f64
fix: bind PII field encryption context
seonghobae Aug 23, 2026
5c6670c
fix: persist governance audit events synchronously
seonghobae Aug 23, 2026
b399eee
Merge remote-tracking branch 'origin/main' into codex/pii-purpose-enc…
seonghobae Aug 24, 2026
ef375cc
fix(deps): keep cryptography as the sole runtime dependency
seonghobae Aug 24, 2026
641b2d1
Merge origin/main into codex/pii-purpose-encryption
seonghobae Aug 24, 2026
519dd9d
fix(deps): lock SQLAlchemy greenlet runtime
seonghobae Aug 25, 2026
6b6bd78
fix(pii): harden derivation and remove scratch artifacts
seonghobae Aug 25, 2026
030b434
fix(ci): preserve exact Atheris Python marker
seonghobae Aug 25, 2026
299e99a
Merge remote-tracking branch 'origin/main' into local-803
seonghobae Aug 25, 2026
68064ee
Merge remote-tracking branch 'origin/main' into local-803
seonghobae Aug 25, 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
3 changes: 3 additions & 0 deletions .github/workflows/fuzz.yml
Original file line number Diff line number Diff line change
Expand Up @@ -87,6 +87,9 @@ jobs:
- name: Fuzz model-judge response parser
run: python fuzz/fuzz_model_judge.py -max_total_time="${FUZZ_SECONDS}" -artifact_prefix=crash- fuzz/corpus/judge

- name: Fuzz PII encryption key boundary
run: python fuzz/fuzz_pii_key.py -max_total_time="${FUZZ_SECONDS}" -artifact_prefix=crash- fuzz/corpus/pii_key

- name: Upload crash artifacts
if: failure()
uses: actions/upload-artifact@330a01c490aca151604b8cf639adc76d48f6c5d4 # actions/upload-artifact@v5
Expand Down
4 changes: 2 additions & 2 deletions .github/workflows/tests.yml
Original file line number Diff line number Diff line change
Expand Up @@ -30,8 +30,8 @@ jobs:

- name: Install test dependencies
# Hash-pinned per OpenSSF Scorecard Pinned-Dependencies. Reuses the
# property-test lockfile (pytest + hypothesis), which covers the full
# suite's requirements: the package itself is stdlib-only.
# property-test lockfile (pytest, hypothesis, and the package runtime
# dependencies), which covers the full suite without unpinned installs.
run: python -m pip install --require-hashes -r fuzz/requirements-property.txt

- name: Run full test suite
Expand Down
6 changes: 3 additions & 3 deletions CLAUDE.md
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,7 @@ CI gates: `.github/workflows/security.yml` (CodeQL + pip-audit on `requirements.

## What this is

A stdlib-Python lab implementing a single OpenAI-compatible API that routes, delegates, verifies, and synthesizes work across a configurable pool of model agents — plus the org's cost-review and sync-vs-batch routing hub. Runtime dependencies are the Python standard library only (Hypothesis is the sole listed dependency, for the property tests); FastAPI/SQLAlchemy/psycopg exist as *optional* extras for the hardened production target, not the current runtime.
A stdlib-Python lab implementing a single OpenAI-compatible API that routes, delegates, verifies, and synthesizes work across a configurable pool of model agents — plus the org's cost-review and sync-vs-batch routing hub. The core runtime uses the Python standard library plus the selected `cryptography` dependency for field-level PII protection; Hypothesis is used for property tests, while FastAPI/SQLAlchemy/psycopg exist as *optional* extras for the hardened production target.

## Architecture

Expand All @@ -88,7 +88,7 @@ A stdlib-Python lab implementing a single OpenAI-compatible API that routes, del

- `orchestrator.py` — the domain heart: `ModelAgent`, `WorkflowStep`, `OrchestrationPolicy`, `ModelClient`, `TaskOrchestrator`, secret/PII redaction, budget enforcement, spend analytics, and the commercial-readiness report generators behind `/api/v1/*`. Domain code stays here until a second implementation forces extraction (see `docs/code_conventions.md`).
- `server.py` — HTTP delivery adapter and `SecurityConfig`; all request validation lives here.
- `admin.py` — static HTML/CSS/JS for the `/admin` operator console (stays inline while the product is dependency-free).
- `admin.py` — static HTML/CSS/JS for the `/admin` operator console (stays inline while the stdlib HTTP/admin surface remains sufficient).
- `credentials.py` / `kv_config.py` — the KV seam: `get_credential`/`register_credential` over pluggable backends (`InMemoryCredentialBackend` default; pgcrypto-encrypted `PostgresCredentialBackend`, selected via `CONTEXTUAL_ORCHESTRATOR_KV_BACKEND`).
- `cost_ledger.py` / `cost_router.py` / `batch_routing.py` / `token_counting.py` — the cost-review + routing hub: prompt-safe usage ledger with seven attribution dimensions, `RoutingPolicy` (sync vs batch from request hints + KV thresholds), and the [pg-llm-batch](https://github.com/ContextualWisdomLab/pg-llm-batch) batch/embeddings backends (a local in-process backend keeps the standalone path working with no external service).
- `api_contract.py` / `conventions.py` — API-shape and naming-rule enforcement helpers.
Expand All @@ -98,7 +98,7 @@ Agent pools are **data, not code**: `examples/agents.mock.json` and `examples/ag

### `conductor/` — context, not code

`conductor/` is the CDD (context-driven development) directory, not a Python package: `product.md` (intent and non-goals), `tech-stack.md` (stdlib-only rationale), `workflow.md` (the TDD/DDD/CDD method and the Ponytail design gate), `tracks.md` (active tracks). Update it when scope, dependencies, workflow, or domain terms change.
`conductor/` is the CDD (context-driven development) directory, not a Python package: `product.md` (intent and non-goals), `tech-stack.md` (stdlib HTTP/core rationale plus selected runtime dependencies), `workflow.md` (the TDD/DDD/CDD method and the Ponytail design gate), `tracks.md` (active tracks). Update it when scope, dependencies, workflow, or domain terms change.
Comment thread
coderabbitai[bot] marked this conversation as resolved.
Resolved

## Key conventions

Expand Down
5 changes: 4 additions & 1 deletion conductor/tech-stack.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,10 @@ Python 3.11+.

## Dependencies

Runtime dependencies: none beyond the Python standard library.
Runtime dependencies:

- `cryptography` for AES-256-GCM protection of explicitly marked PII fields.
- The HTTP server, persistence, and orchestration core remain Python standard-library based.

Production target dependencies after this lab hardens:

Expand Down
176 changes: 163 additions & 13 deletions contextual_orchestrator/orchestrator.py
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@
import math
import os
from pathlib import Path
import queue
import random
import re
import socket
Expand All @@ -33,6 +34,14 @@

from .conventions import require_object_name
from .credentials import NotConfigured, get_credential
from .pii_protection import (
DEFAULT_PII_KEY_NAME,
ENCRYPTED_FIELDS_KEY,
PiiFieldEncryptor,
PiiProtectionError,
is_encrypted_detail,
load_pii_encryptor,
)


# content is usually str; multimodal vision messages use OpenAI content-parts lists.
Expand Down Expand Up @@ -1542,20 +1551,25 @@ class _StateStore:
"""Minimal write-through sqlite persistence for orchestrator runtime state.

ponytail: one generic table, no ORM. Keyed kinds (workflow_run, evaluation_run)
upsert by key; stream kinds (analytics, audit) append. Stream rows grow unbounded
on disk while the in-memory deques stay capped — add pruning if db size matters.
upsert by key; stream kinds append. Durable audit streams use the same bounded
retention as their in-memory deques so request traffic cannot grow the DB forever.
Runtime values (kind, key, payload, limit) are always bound through SQLite
placeholders so persisted prompts and identifiers cannot become SQL syntax.
"""

_KEYED = {"workflow_run", "evaluation_run"}
_STREAM_LIMITS = {"audit": 256, "authorization": 256, "analytics": 256}
_CREATE_RECORDS_SQL = (
"CREATE TABLE IF NOT EXISTS records ("
"seq INTEGER PRIMARY KEY AUTOINCREMENT, kind TEXT NOT NULL, key TEXT, payload TEXT NOT NULL)"
)
_CREATE_RECORDS_KIND_SEQ_INDEX_SQL = "CREATE INDEX IF NOT EXISTS records_kind_seq ON records(kind, seq)"
_DELETE_KEYED_SQL = "DELETE FROM records WHERE kind = ? AND key = ?"
_INSERT_SQL = "INSERT INTO records (kind, key, payload) VALUES (?, ?, ?)"
_PRUNE_STREAM_SQL = (
"DELETE FROM records WHERE kind = ? AND seq NOT IN ("
"SELECT seq FROM records WHERE kind = ? ORDER BY seq DESC LIMIT ?)"
)
_SELECT_ALL_SQL = "SELECT payload FROM records WHERE kind = ? ORDER BY seq"
_SELECT_LIMIT_SQL = "SELECT payload FROM records WHERE kind = ? ORDER BY seq DESC LIMIT ?"

Expand All @@ -1565,16 +1579,50 @@ def __init__(self, path: str) -> None:
self._conn.execute(self._CREATE_RECORDS_SQL)
self._conn.execute(self._CREATE_RECORDS_KIND_SEQ_INDEX_SQL)
self._conn.commit()
# Stream kinds (audit/authorization/analytics) are already bounded and
# best-effort by design (see class docstring). Route them through a
# background queue so an unauthenticated request flood — e.g. denied
# authorizations — cannot force synchronous, lock-serialized disk
# commits on the request-handling thread. Keyed kinds (workflow_run,
# evaluation_run) need durability and stay synchronous.
self._stream_queue: queue.Queue[tuple[str, str | None, dict[str, Any]]] = queue.Queue(maxsize=2048)
self._stream_worker = threading.Thread(
target=self._drain_stream_queue,
name="contextual-orchestrator-state-store",
daemon=True,
)
self._stream_worker.start()

def save(self, kind: str, key: str | None, payload: dict[str, Any]) -> None:
if kind in self._STREAM_LIMITS:
try:
self._stream_queue.put_nowait((kind, key, payload))
except queue.Full:
pass
Comment thread
coderabbitai[bot] marked this conversation as resolved.
Resolved
return
self._save_sync(kind, key, payload)
Comment thread
seonghobae marked this conversation as resolved.
Resolved
Comment on lines +1719 to +1727

@devin-ai-integration devin-ai-integration Bot Aug 25, 2026 •

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

📝 Info: Analytics and denials become best-effort async

_append_audit_event defaults durable=True, so audit writes stay synchronous (now DB-capped at 256). Only record_analytics_event and authorization denials go through the async best-effort queue, which can silently drop on queue overflow or a failed worker write. Intended per the ADR, but analytics/denials are no longer guaranteed durable.

Open in Devin Review

Was this helpful? React with 👍 or 👎 to provide feedback.


def _save_sync(self, kind: str, key: str | None, payload: dict[str, Any]) -> None:
blob = json.dumps(payload, ensure_ascii=False)
with self._lock:
if kind in self._KEYED:
self._conn.execute(self._DELETE_KEYED_SQL, (kind, key))
self._conn.execute(self._INSERT_SQL, (kind, key, blob))
if kind in self._STREAM_LIMITS:
limit = self._STREAM_LIMITS[kind]
self._conn.execute(self._PRUNE_STREAM_SQL, (kind, kind, limit))
Comment thread
seonghobae marked this conversation as resolved.
Resolved
self._conn.commit()

def _drain_stream_queue(self) -> None:
while True:
kind, key, payload = self._stream_queue.get()
try:
self._save_sync(kind, key, payload)
finally:
self._stream_queue.task_done()
Comment thread
github-actions[bot] marked this conversation as resolved.
Resolved
Comment thread
github-actions[bot] marked this conversation as resolved.
Resolved

def load(self, kind: str, limit: int | None = None) -> list[dict[str, Any]]:
self._stream_queue.join()
Comment thread
github-actions[bot] marked this conversation as resolved.
Resolved
with self._lock:
if limit is None:
rows = self._conn.execute(self._SELECT_ALL_SQL, (kind,)).fetchall()
Expand All @@ -1585,6 +1633,7 @@ def load(self, kind: str, limit: int | None = None) -> list[dict[str, Any]]:

def close(self) -> None:
"""Close the sqlite handle so Windows can release the database file."""
self._stream_queue.join()
with self._lock:
self._conn.close()
Comment thread
devin-ai-integration[bot] marked this conversation as resolved.
Resolved

Expand Down Expand Up @@ -1667,6 +1716,7 @@ def __init__(
agents_db: str | None = None,
cache_ttl: float = 0.0,
cache_max_entries: int = 256,
pii_key_name: str = DEFAULT_PII_KEY_NAME,
) -> None:
# Optional durable model-group management: stored operator changes overlay the
# seed agents file at startup (stored rows win by id; stored-new rows append).
Expand All @@ -1687,8 +1737,9 @@ def __init__(
self.budget_max_cost_usd = budget_max_cost_usd
self._workflow_runs: dict[str, dict[str, Any]] = {}
self._evaluation_runs: dict[str, dict[str, Any]] = {}
self._analytics_events: deque[dict[str, Any]] = deque(maxlen=512)
self._analytics_events: deque[dict[str, Any]] = deque(maxlen=256)
Comment thread
seonghobae marked this conversation as resolved.
Resolved
self._audit_events: deque[dict[str, Any]] = deque(maxlen=256)
self._authorization_events: deque[dict[str, Any]] = deque(maxlen=256)
Comment thread
seonghobae marked this conversation as resolved.
Resolved
self._run_order: deque[str] = deque(maxlen=128)
# Per-agent circuit breaker: consecutive failures trip an agent "open"
# so a persistently failing provider is skipped until it cools down.
Expand All @@ -1702,6 +1753,10 @@ def __init__(
# Optional durable persistence: default None keeps all state purely in-memory
# (zero behavior change). When set, runs/audit/analytics survive restart.
self._store = _StateStore(state_db) if state_db else None
if not isinstance(pii_key_name, str) or not pii_key_name:
raise ValueError("pii_key_name must be a non-empty string")
Comment on lines +1946 to +1947

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 State store thread leaks when PII key name is rejected

The durable state store starts its background thread and opens its sqlite connection (_StateStore(state_db)) before pii_key_name is validated. An invalid name then raises ValueError, orphaning that thread and connection for the life of the process because close() never runs.

Prompt for agents
In TaskOrchestrator.__init__ the pii_key_name validation (the isinstance/non-empty check that raises ValueError) is placed after self._store = _StateStore(state_db) is created. _StateStore.__init__ starts a daemon worker thread and opens a sqlite connection, and only close() tears those down. When pii_key_name is invalid the ValueError propagates out of __init__ before the caller ever gets a reference to close(), leaking the thread and connection. Move the pii_key_name validation (and assignment of self._pii_key_name / self._pii_encryptors) so it runs before the _StateStore is constructed, so an invalid argument fails before any resource is allocated.
Open in Devin Review

Was this helpful? React with 👍 or 👎 to provide feedback.

self._pii_key_name = pii_key_name
self._pii_encryptors: dict[str, PiiFieldEncryptor] = {}
self._commercial_report_cache_local = threading.local()
if self._store is not None:
self._reload_state()
Expand Down Expand Up @@ -1770,6 +1825,8 @@ def _reload_state(self) -> None:
self._analytics_events.append(event)
for event in self._store.load("audit", self._audit_events.maxlen):
self._audit_events.append(event)
for event in self._store.load("authorization", self._authorization_events.maxlen):
self._authorization_events.append(event)

# Orchestration-only body keys that must not be forwarded to the provider.
_ORCHESTRATION_ONLY_KEYS = frozenset(
Expand Down Expand Up @@ -2708,15 +2765,55 @@ def _judge_verifier_output(self, verifier_output: str, thinker_output: str, work
"verifier_output": verifier_output,
}

def _append_audit_event(self, event_type: str, detail: dict[str, Any]) -> None:
def _protected_event_detail(self, detail: dict[str, Any], pii_fields: Iterable[str]) -> dict[str, Any]:
"""Encrypt explicitly declared PII fields before an event enters memory or storage."""
Comment thread
seonghobae marked this conversation as resolved.
Resolved
fields = tuple(pii_fields)
if not fields:
return detail
encryptor = self._pii_encryptors.get(self._pii_key_name)
if encryptor is None:
encryptor = load_pii_encryptor(self._pii_key_name)
self._pii_encryptors[self._pii_key_name] = encryptor
Comment thread
seonghobae marked this conversation as resolved.
Resolved
return encryptor.encrypt_fields(detail, fields)

def _append_audit_event(
self,
event_type: str,
detail: dict[str, Any],
*,
pii_fields: Iterable[str] = (),
stream: str = "audit",
) -> None:
"""Append an event to a bounded audit stream."""
event = {
"created_at": int(time.time()),
"event_type": event_type,
"event_detail": detail,
"event_detail": self._protected_event_detail(detail, pii_fields),
}
self._audit_events.append(event)
events = self._authorization_events if stream == "authorization" else self._audit_events
events.append(event)
if self._store is not None:
self._store.save("audit", None, event)
self._store.save(stream, None, event)

def record_authorization_decision(
self,
*,
scope: str,
purpose: str,
allowed: bool,
reason: str,
) -> None:
"""Record a secret-free role/purpose authorization decision."""
self._append_audit_event(
"authorization_decision",
{
"scope": scope,
"purpose": purpose,
"allowed": bool(allowed),
"reason": reason,
},
stream="authorization",
)

def _infer_provider_name(self, base_url: str) -> str:
if base_url.startswith("mock://"):
Expand Down Expand Up @@ -2805,8 +2902,15 @@ def list_recent_runs(self, page_number: int = 1, page_size: int = 10) -> list[di
run_ids = list(self._run_order)[start:end]
return [self._workflow_runs[run_id] for run_id in run_ids]

def list_recent_audit_events(self, page_number: int = 1, page_size: int = 25) -> list[dict[str, Any]]:
"""Return recent audit events in newest-first order."""
def list_recent_audit_events(
self,
page_number: int = 1,
page_size: int = 25,
*,
role: str | None = None,
purpose: str | None = None,
) -> list[dict[str, Any]]:
"""Return recent audit events, decrypting PII only for authorized replay."""
if page_number < 1 or page_size < 1: # pragma: no cover
raise ValueError("page_number/page_size must be >= 1")
events = list(self._audit_events)
Expand All @@ -2815,15 +2919,60 @@ def list_recent_audit_events(self, page_number: int = 1, page_size: int = 25) ->
total = len(events)
left = max(0, total - end)
right = max(0, total - start)
selected = list(reversed(events[left:right]))
if role != "admin" or purpose != "audit_replay":
return selected
restored: list[dict[str, Any]] = []
encryptors: dict[str, Any] = {}
for event in selected:
detail = event.get("event_detail")
if not is_encrypted_detail(detail):
restored.append(event)
continue
restored_event = dict(event)
Comment thread
devin-ai-integration[bot] marked this conversation as resolved.
Resolved
try:
metadata = detail.get(ENCRYPTED_FIELDS_KEY)
key_name = metadata.get("key_name") if isinstance(metadata, dict) else self._pii_key_name
if not isinstance(key_name, str) or not key_name:
raise PiiProtectionError("encrypted field metadata has no valid key name")
encryptor = encryptors.get(key_name)
if encryptor is None:
encryptor = load_pii_encryptor(key_name)
encryptors[key_name] = encryptor
restored_event["event_detail"] = encryptor.decrypt_fields(detail)
except PiiProtectionError:
restored_event["event_detail"] = {
**detail,
"__pii_protection_error__": "unavailable",
}
restored.append(restored_event)
return restored
Comment on lines +3254 to +3280

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

📝 Info: PII plaintext exposure limited to admin audit_replay

Decryption in list_recent_audit_events is gated on role admin and purpose audit_replay, and the purpose is chosen by the route via _admin_purpose(path) rather than request data, so callers cannot escalate. Only /admin/state maps to audit_replay; the other admin_state() caller passes no role/purpose and sees ciphertext. Consequence worth noting: every /admin/state GET returns plaintext PII to any admin-token holder and forces a synchronous durable authorization write that returns 503 if it fails.

Open in Devin Review

Was this helpful? React with 👍 or 👎 to provide feedback.


def list_recent_authorization_decisions(self, page_number: int = 1, page_size: int = 25) -> list[dict[str, Any]]:
"""Return recent secret-free authorization decisions in newest-first order."""
if page_number < 1 or page_size < 1: # pragma: no cover
raise ValueError("page_number/page_size must be >= 1")
events = list(self._authorization_events)
start = (page_number - 1) * page_size
end = start + page_size
total = len(events)
left = max(0, total - end)
right = max(0, total - start)
return list(reversed(events[left:right]))

def record_analytics_event(self, event_name: str, detail: dict[str, Any]) -> None:
def record_analytics_event(
self,
event_name: str,
detail: dict[str, Any],
*,
pii_fields: Iterable[str] = (),
) -> None:
"""Record a compact in-memory analytics event without prompt or output text."""
require_object_name(event_name, "analytics.event_name")
event = {
"event_time": int(time.time()),
"event_name": event_name,
"event_detail": redact_value(detail),
"event_detail": redact_value(self._protected_event_detail(detail, pii_fields)),
}
self._analytics_events.append(event)
if self._store is not None:
Comment thread
seonghobae marked this conversation as resolved.
Resolved
Comment thread
seonghobae marked this conversation as resolved.
Resolved
Comment on lines +3294 to 3309

@devin-ai-integration devin-ai-integration Bot Aug 25, 2026 •

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

📝 Info: Analytics-encrypted PII is never decryptable

record_analytics_event encrypts declared pii_fields, but only list_recent_audit_events has a decrypt/replay path (admin + audit_replay). Analytics PII envelopes have no recovery surface, so PII placed there is effectively write-only. Appears deliberate; noted for awareness.

Open in Devin Review

Was this helpful? React with 👍 or 👎 to provide feedback.

Expand Down Expand Up @@ -9006,7 +9155,7 @@ def section(
},
}

def admin_state(self) -> dict[str, Any]:
def admin_state(self, *, role: str | None = None, purpose: str | None = None) -> dict[str, Any]:
"""Build the admin console state payload from agents, policy, and audit data."""
agent_page_size = max(1, len(self.candidates))
return {
Expand All @@ -9017,7 +9166,8 @@ def admin_state(self) -> dict[str, Any]:
"complex_hints": list(self.COMPLEX_HINTS),
},
"recent_workflow_runs": [self._shorten_run(run) for run in self.list_recent_runs(page_size=max(1, len(self._run_order)))],
"recent_audit_events": self.list_recent_audit_events(),
"recent_audit_events": self.list_recent_audit_events(role=role, purpose=purpose),
"recent_authorization_decisions": self.list_recent_authorization_decisions(),
Comment thread
seonghobae marked this conversation as resolved.
Resolved
"spend": self.spend_analytics(),
}

Expand Down
Loading
Loading