Use get jobs from PMAT (#474)

prediction_market_agent/agents/microchain_agent/call_api.py

@@ -73,58 +73,7 @@ def __call__(
         return "Message sent"
 
 
-class GetTasks(Function):
-    @property
-    def description(self) -> str:
-        return "Use this function to get available tasks in a JSON dumped format."
-
-    @property
-    def example_args(self) -> list[str]:
-        return []
-
-    def __call__(self) -> str:
-        return json.dumps(
-            [
-                {
-                    "id": "1",
-                    "title": """Solve the following math problem:
-
-If there are 3 apples and you take away 2, how many do you have?
-                    
-Send the answer to Telegram Chat ID 6514789173.
-""",
-                    "reward": {
-                        "amount": 1,
-                        "currency": "wxDai",
-                    },
-                },
-                {
-                    "id": "2",
-                    "title": """
-Do a prediction on the following Omen market with id `0xa6febd6ffda06af320f5d92ec3dd562149cc2bc7`.                  
-
-Send the result to Telegram Chat ID 6514789173.
-""",
-                    "reward": {
-                        "amount": 1,
-                        "currency": "wETH",
-                    },
-                },
-                {
-                    "id": "3",
-                    "title": "Install Python library `chihuahua` and then execute the following code: `import chihuahua; print(42)`.",
-                    "reward": {
-                        "amount": 2,
-                        "currency": "wxDai",
-                    },
-                },
-            ],
-            indent=2,
-        )
-
-
 API_FUNCTIONS: list[type[Function]] = [
     CallAPI,
     SendTelegramMessage,
-    GetTasks,
 ]

prediction_market_agent/agents/microchain_agent/jobs_functions.py

@@ -0,0 +1,44 @@
+import json
+
+from microchain import Function
+from prediction_market_agent_tooling.jobs.jobs import get_jobs
+from prediction_market_agent_tooling.markets.data_models import Currency
+from prediction_market_agent_tooling.markets.markets import MarketType
+
+from prediction_market_agent.utils import APIKeys
+
+
+class JobFunction(Function):
+    def __init__(self, market_type: MarketType, keys: APIKeys) -> None:
+        self.keys = keys
+        self.market_type = market_type
+        super().__init__()
+
+    @property
+    def currency(self) -> Currency:
+        return self.market_type.market_class.currency
+
+
+class GetJobs(JobFunction):
+    @property
+    def description(self) -> str:
+        return f"""Use this function to get available jobs in a JSON dumped format.
+You need to provide max bond value in {self.currency}, that is, how much you are willing to bond on the fact that you completed the job as required in the job description.
+"""
+
+    @property
+    def example_args(self) -> list[int]:
+        return [10]
+
+    def __call__(self, max_bond: float) -> str:
+        jobs = get_jobs(self.market_type, limit=None)
+        return json.dumps(
+            [j.to_simple_job(max_bond=max_bond).model_dump() for j in jobs],
+            indent=2,
+            default=str,
+        )
+
+
+JOB_FUNCTIONS: list[type[JobFunction]] = [
+    GetJobs,
+]

prediction_market_agent/agents/microchain_agent/market_functions.py

@@ -14,7 +14,7 @@
     withdraw_wxdai_to_xdai_to_keep_balance,
 )
 from prediction_market_agent_tooling.tools.betting_strategies.kelly_criterion import (
-    get_kelly_bet,
+    get_kelly_bet_simplified,
 )
 from prediction_market_agent_tooling.tools.tavily_storage.tavily_models import (
     TavilyStorage,
@@ -420,7 +420,7 @@ def __call__(
     ) -> str:
         confidence = 0.5  # Until confidence score is available, be conservative
         max_bet = float(get_balance(self.keys, market_type=self.market_type).amount)
-        kelly_bet = get_kelly_bet(
+        kelly_bet = get_kelly_bet_simplified(
             market_p_yes=market_p_yes,
             estimated_p_yes=estimated_p_yes,
             max_bet=max_bet,

prediction_market_agent/agents/microchain_agent/microchain_agent.py

@@ -28,6 +28,7 @@
 from prediction_market_agent.agents.microchain_agent.code_functions import (
     CODE_FUNCTIONS,
 )
+from prediction_market_agent.agents.microchain_agent.jobs_functions import JOB_FUNCTIONS
 from prediction_market_agent.agents.microchain_agent.learning_functions import (
     LEARNING_FUNCTIONS,
 )
@@ -120,6 +121,9 @@ def build_agent_functions(
         functions.extend([f() for f in API_FUNCTIONS])
         functions.extend([f() for f in CODE_FUNCTIONS])
 
+    if functions_config.include_job_functions:
+        functions.extend([f(market_type=market_type, keys=keys) for f in JOB_FUNCTIONS])
+
     if functions_config.include_learning_functions:
         functions.extend([f() for f in LEARNING_FUNCTIONS])
 

prediction_market_agent/agents/microchain_agent/prompts.py

@@ -49,14 +49,14 @@
 Make 'Reasoning' calls frequently - at least every other call. You need to reason step by step.
 """
 
-# Experimental system prompt for task-solving agent.
-TASK_AGENT_SYSTEM_PROMPT = f"""Act as a task-solving agents that picks up available tasks and solves them for getting rewards.
+# Experimental system prompt for job-solving agent.
+JOB_AGENT_SYSTEM_PROMPT = f"""Act as a job-solving agents that picks up available jobs and completes them for getting rewards.
 
-Pick up available task that's returned from GetTasks and pick one that you can solve and it's worth solving.
+Pick up available job that's returned from GetJobs and pick one that you can complete and it's worth finishing.
 
-While solving a task, reason step by step and write down thoroughly the process. You can use the 'Reasoning' function for that.
+While working on a job, reason step by step and write down thoroughly the process. You can use the 'Reasoning' function for that.
 
-Don't do anything else, just solve the task, and then pick up another one.
+Don't do anything else, just complete the job, and then pick up another one.
 
 {NON_UPDATABLE_DIVIDOR}
 
@@ -91,7 +91,7 @@ class SystemPromptChoice(str, Enum):
     JUST_BORN = "just_born"
     TRADING_AGENT = "trading_agent"
     TRADING_AGENT_MINIMAL = "trading_agent_minimal"
-    TASK_AGENT = "task_agent"
+    JOB_AGENT = "job_agent"
 
 
 class FunctionsConfig(BaseModel):
@@ -100,6 +100,7 @@ class FunctionsConfig(BaseModel):
     include_trading_functions: bool
     include_universal_functions: bool
     include_agent_functions: bool
+    include_job_functions: bool
 
     @staticmethod
     def from_system_prompt_choice(
@@ -109,6 +110,7 @@ def from_system_prompt_choice(
         include_learning_functions = False
         include_universal_functions = False
         include_agent_functions = False
+        include_job_functions = False
 
         if system_prompt_choice == SystemPromptChoice.JUST_BORN:
             include_learning_functions = True
@@ -121,22 +123,24 @@ def from_system_prompt_choice(
         ]:
             include_trading_functions = True
 
-        elif system_prompt_choice == SystemPromptChoice.TASK_AGENT:
+        elif system_prompt_choice == SystemPromptChoice.JOB_AGENT:
             include_universal_functions = True
             include_agent_functions = True
             include_trading_functions = True
+            include_job_functions = True
 
         return FunctionsConfig(
             include_trading_functions=include_trading_functions,
             include_learning_functions=include_learning_functions,
             include_universal_functions=include_universal_functions,
             include_agent_functions=include_agent_functions,
+            include_job_functions=include_job_functions,
         )
 
 
 SYSTEM_PROMPTS: dict[SystemPromptChoice, str] = {
     SystemPromptChoice.JUST_BORN: SYSTEM_PROMPT,
     SystemPromptChoice.TRADING_AGENT: TRADING_AGENT_SYSTEM_PROMPT,
-    SystemPromptChoice.TASK_AGENT: TASK_AGENT_SYSTEM_PROMPT,
+    SystemPromptChoice.JOB_AGENT: JOB_AGENT_SYSTEM_PROMPT,
     SystemPromptChoice.TRADING_AGENT_MINIMAL: TRADING_AGENT_SYSTEM_PROMPT_MINIMAL,
 }

pyproject.toml

@@ -34,7 +34,7 @@ poetry = "^1.7.1"
 poetry-plugin-export = "^1.6.0"
 functions-framework = "^3.5.0"
 cron-validator = "^1.0.8"
-prediction-market-agent-tooling = { version = "^0.48.14", extras = ["langchain", "google"] }
+prediction-market-agent-tooling = { version = "^0.48.15", extras = ["langchain", "google"] }
 pydantic-settings = "^2.1.0"
 autoflake = "^2.2.1"
 isort = "^5.13.2"