Files
awoooi/apps/api/src/services/ai_agent_interaction_learning_proof.py
Your Name 6982a674b4
All checks were successful
CD Pipeline / tests (push) Successful in 1m28s
Code Review / ai-code-review (push) Successful in 14s
CD Pipeline / build-and-deploy (push) Successful in 4m46s
CD Pipeline / post-deploy-checks (push) Successful in 2m2s
feat(governance): 顯示 Agent 互動學習證據
2026-06-11 18:53:24 +08:00

198 lines
7.8 KiB
Python

"""
AI Agent interaction and learning proof snapshot.
Loads the latest committed, read-only proof surface for how operators can see
OpenClaw, Hermes, and NemoTron communicating, handing off work, learning, and
growing. This module is intentionally truth-gated: it never starts workers,
opens Redis consumer groups, writes database migrations, sends Telegram
messages, exposes transcripts, or marks live runtime as active.
"""
from __future__ import annotations
import json
from pathlib import Path
from typing import Any
from src.services.snapshot_paths import default_evaluations_dir
_DEFAULT_EVALUATIONS_DIR = default_evaluations_dir(Path(__file__))
_SNAPSHOT_PATTERN = "ai_agent_interaction_learning_proof_*.json"
_SCHEMA_VERSION = "ai_agent_interaction_learning_proof_v1"
def load_latest_ai_agent_interaction_learning_proof(
evaluations_dir: Path | None = None,
) -> dict[str, Any]:
"""Load the newest committed AI Agent interaction learning proof snapshot."""
directory = evaluations_dir or _DEFAULT_EVALUATIONS_DIR
candidates = sorted(directory.glob(_SNAPSHOT_PATTERN))
if not candidates:
raise FileNotFoundError(
f"no AI Agent interaction learning proof snapshots found in {directory}"
)
latest = candidates[-1]
with latest.open(encoding="utf-8") as handle:
payload = json.load(handle)
if not isinstance(payload, dict):
raise ValueError(f"{latest}: expected JSON object")
_require_schema(payload, _SCHEMA_VERSION, str(latest))
_require_read_only_truth(payload, str(latest))
_require_rollup_consistency(payload, str(latest))
_require_agent_lanes(payload, str(latest))
_require_frontend_redaction(payload, str(latest))
return payload
def _require_schema(payload: dict[str, Any], expected: str, label: str) -> None:
actual = payload.get("schema_version")
if actual != expected:
raise ValueError(f"{label}: expected schema_version={expected}, got {actual!r}")
def _require_read_only_truth(payload: dict[str, Any], label: str) -> None:
program_status = payload.get("program_status") or {}
if program_status.get("read_only_mode") is not True:
raise ValueError(f"{label}: program_status.read_only_mode must be true")
if program_status.get("runtime_authority") != "proof_surface_only_no_live_worker":
raise ValueError(
f"{label}: runtime_authority must stay proof_surface_only_no_live_worker"
)
live_truth = payload.get("live_truth") or {}
live_flags = {
"runtime_loop_enabled",
"live_agent_session_readback_enabled",
"redis_consumer_group_enabled",
"telegram_send_enabled",
"learning_writeback_enabled",
}
enabled = sorted(flag for flag in live_flags if live_truth.get(flag) is not False)
if enabled:
raise ValueError(f"{label}: live truth flags must remain false: {enabled}")
live_counts = {
"active_live_agent_sessions",
"live_agent_messages_24h",
"live_handoffs_24h",
"live_learning_writes_24h",
"telegram_digest_receipts_24h",
}
non_zero = sorted(key for key in live_counts if live_truth.get(key) != 0)
if non_zero:
raise ValueError(f"{label}: live truth counts must remain zero: {non_zero}")
boundaries = payload.get("approval_boundaries") or {}
blocked_flags = {
"runtime_worker_allowed",
"db_migration_allowed",
"redis_consumer_group_allowed",
"telegram_direct_send_allowed",
"conversation_transcript_display_allowed",
"agent_private_reasoning_display_allowed",
"secret_plaintext_allowed",
"autonomous_self_modify_allowed",
}
allowed = sorted(flag for flag in blocked_flags if boundaries.get(flag) is not False)
if allowed:
raise ValueError(f"{label}: approval boundaries must remain false: {allowed}")
def _require_rollup_consistency(payload: dict[str, Any], label: str) -> None:
rollups = payload.get("rollups") or {}
proof_ladder = payload.get("proof_ladder") or []
proof_signals = payload.get("proof_signals") or []
operator_surfaces = payload.get("operator_surfaces") or []
runtime_gates = payload.get("runtime_gates") or []
expected_counts = {
"proof_level_count": len(proof_ladder),
"signal_count": len(proof_signals),
"operator_surface_count": len(operator_surfaces),
"runtime_gate_count": len(runtime_gates),
}
mismatched = {
key: {"expected": expected, "actual": rollups.get(key)}
for key, expected in expected_counts.items()
if rollups.get(key) != expected
}
if mismatched:
raise ValueError(f"{label}: rollup counts must match payload sections: {mismatched}")
contract_ready_ids = sorted(
level.get("level_id")
for level in proof_ladder
if level.get("status") in {"contract_ready", "proof_surface_ready"}
)
if rollups.get("contract_ready_level_count") != len(contract_ready_ids):
raise ValueError(f"{label}: rollups.contract_ready_level_count mismatch")
live_pending_ids = sorted(
level.get("level_id")
for level in proof_ladder
if level.get("status") in {"live_pending", "blocked_by_gate"}
)
if sorted(rollups.get("live_pending_level_ids") or []) != live_pending_ids:
raise ValueError(f"{label}: rollups.live_pending_level_ids mismatch")
live_signal_count = sum(
1 for signal in proof_signals if signal.get("current_state") == "live_verified"
)
if rollups.get("live_signal_count") != live_signal_count:
raise ValueError(f"{label}: rollups.live_signal_count mismatch")
blocked_gate_ids = sorted(
gate.get("gate_id")
for gate in runtime_gates
if gate.get("status") in {"blocked", "approval_required"}
)
if sorted(rollups.get("blocked_gate_ids") or []) != blocked_gate_ids:
raise ValueError(f"{label}: rollups.blocked_gate_ids mismatch")
live_truth = payload.get("live_truth") or {}
for key in (
"active_live_agent_sessions",
"live_agent_messages_24h",
"live_handoffs_24h",
"live_learning_writes_24h",
"telegram_digest_receipts_24h",
):
if rollups.get(key) != live_truth.get(key):
raise ValueError(f"{label}: rollups.{key} must mirror live_truth.{key}")
def _require_agent_lanes(payload: dict[str, Any], label: str) -> None:
lanes = payload.get("agent_lanes") or []
lane_ids = {lane.get("agent_id") for lane in lanes}
required_lanes = {"openclaw", "hermes", "nemotron"}
if not required_lanes.issubset(lane_ids):
raise ValueError(f"{label}: missing required agent lanes: {sorted(required_lanes - lane_ids)}")
missing_visible_signal = [
lane.get("agent_id")
for lane in lanes
if not lane.get("visible_signals")
]
if missing_visible_signal:
raise ValueError(f"{label}: every agent lane needs visible_signals: {missing_visible_signal}")
unsafe_lanes = [
lane.get("agent_id")
for lane in lanes
if "conversation_transcript" in set(lane.get("visible_signals") or [])
]
if unsafe_lanes:
raise ValueError(f"{label}: visible signals must not expose transcripts: {unsafe_lanes}")
def _require_frontend_redaction(payload: dict[str, Any], label: str) -> None:
redaction = payload.get("frontend_redaction") or {}
if redaction.get("operator_conversation_display_allowed") is not False:
raise ValueError(f"{label}: operator conversation display must stay false")
if redaction.get("agent_private_reasoning_display_allowed") is not False:
raise ValueError(f"{label}: agent private reasoning display must stay false")
if redaction.get("raw_prompt_display_allowed") is not False:
raise ValueError(f"{label}: raw prompt display must stay false")