Implement PredictProbabilityForQuestion function for microchain agent

.gitignore

@@ -6,3 +6,6 @@ __pycache__
 logs
 .mypy_cache
 .pytest_cache
+
+# Created on mech-client `interact` call
+acn_cert.txt

prediction_market_agent/agents/known_outcome_agent/known_outcome_agent.py

@@ -1,5 +1,3 @@
-import json
-import typing as t
 from datetime import datetime
 from enum import Enum
 
@@ -12,6 +10,7 @@
 from prediction_market_agent.tools.web_scrape.basic_summary import _summary
 from prediction_market_agent.tools.web_scrape.markdown import web_scrape
 from prediction_market_agent.tools.web_search.tavily import web_search
+from prediction_market_agent.utils import completion_str_to_json
 
 
 class Result(str, Enum):
@@ -141,26 +140,6 @@ def has_known_result(self) -> bool:
 """
 
 
-def completion_str_to_json(completion: str) -> dict[str, t.Any]:
-    """
-    Cleans completion JSON in form of a string:
-
-    ```json
-    {
-        ...
-    }
-    ```
-
-    into just { ... }
-    ```
-    """
-    start_index = completion.find("{")
-    end_index = completion.rfind("}")
-    completion = completion[start_index : end_index + 1]
-    completion_dict: dict[str, t.Any] = json.loads(completion)
-    return completion_dict
-
-
 def summarize_if_required(content: str, model: str, question: str) -> str:
     """
     If the content is too long to fit in the model's context, summarize it.

prediction_market_agent/agents/microchain_agent/functions.py

@@ -1,7 +1,9 @@
+import json
 import typing as t
 from decimal import Decimal
 
 from eth_utils import to_checksum_address
+from mech_client.interact import interact
 from microchain import Function
 from prediction_market_agent_tooling.markets.agent_market import AgentMarket
 from prediction_market_agent_tooling.markets.data_models import Currency, TokenAmount
@@ -11,16 +13,21 @@
     OmenSubgraphHandler,
 )
 
+from prediction_market_agent.agents.microchain_agent.mech import (
+    prediction_with_research_report,
+)
 from prediction_market_agent.agents.microchain_agent.utils import (
+    MechResult,
     MicroMarket,
     get_balance,
     get_binary_markets,
     get_boolean_outcome,
     get_example_market_id,
     get_no_outcome,
     get_yes_outcome,
+    saved_str_to_tmpfile,
 )
-from prediction_market_agent.utils import APIKeys
+from prediction_market_agent.utils import APIKeys, completion_str_to_json
 
 
 class Sum(Function):
@@ -101,19 +108,71 @@ def __call__(self, market_id: str) -> list[str]:
 class PredictPropabilityForQuestion(MarketFunction):
     @property
     def description(self) -> str:
-        return "Use this function to research the probability of an event occuring"
+        return (
+            "Use this function to research perform research and predict the "
+            "probability of an event occuring. Returns the probability. The "
+            "one parameter is the market id of the prediction market you want "
+            "to predict the probability of. Note, this costs money to run."
+        )
 
     @property
     def example_args(self) -> list[str]:
-        return ["Will Joe Biden get reelected in 2024?"]
+        return [get_example_market_id(self.market_type)]
+
+    def __call__(self, question: str) -> str:
+        private_key = APIKeys().bet_from_private_key.get_secret_value()
+        # 0.01 xDai is hardcoded cost for an interaction with the mech-client
+        MECH_CALL_XDAI_LIMIT = 0.011
+        account_balance = float(get_balance(market_type=self.market_type).amount)
+        if account_balance < MECH_CALL_XDAI_LIMIT:
+            return (
+                f"Your balance of {self.currency} ({account_balance}) is not "
+                f"large enough to make a mech call (min required "
+                f"{MECH_CALL_XDAI_LIMIT})."
+            )
+        with saved_str_to_tmpfile(private_key) as tmpfile_path:
+            response = interact(
+                prompt=question,
+                # Taken from https://github.com/valory-xyz/mech?tab=readme-ov-file#examples-of-deployed-mechs
+                agent_id=3,
+                private_key_path=tmpfile_path,
+                # To see a list of available tools, comment out the tool parameter
+                # and run the function. You will be prompted to select a tool.
+                tool="claude-prediction-online",
+            )
+            result = json.loads(response["result"])
+            return str(MechResult.model_validate(result).p_yes)
 
-    def __call__(self, a: str) -> float:
-        if a == "Will Joe Biden get reelected in 2024?":
-            return 0.41
-        if a == "Will Bitcoin hit 100k in 2024?":
-            return 0.22
 
-        return 0.0
+class PredictPropabilityForQuestionLocal(MarketFunction):
+    @property
+    def description(self) -> str:
+        return (
+            "Use this function to research perform research and predict the "
+            "probability of an event occuring. Returns the probability. The "
+            "one parameter is the market id of the prediction market you want "
+            "to predict the probability of."
+        )
+
+    @property
+    def example_args(self) -> list[str]:
+        return [get_example_market_id(self.market_type)]
+
+    def __call__(self, question: str) -> str:
+        keys = APIKeys()
+        openai_api_key = keys.openai_api_key.get_secret_value()
+        tavily_api_key = keys.tavily_api_key.get_secret_value()
+        kwargs = {
+            "tool": "prediction-with-research-conservative",
+            "prompt": question,
+            "api_keys": {
+                "openai": openai_api_key,
+                "tavily": tavily_api_key,
+            },
+        }
+        response = prediction_with_research_report.run(**kwargs)  # type: ignore
+        result = completion_str_to_json(str(response[0]))
+        return str(MechResult.model_validate(result).p_yes)
 
 
 class BuyTokens(MarketFunction):