Tool tree
gradeNot graded
get_game_stateAnswer
Read the current puzzle: guesses played so far, per-letter feedback, what has been ruled in or out, attempts left, and whose turn it is. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.
View tool JSON
{
"name": "get_game_state",
"kind": "answer",
"impl": "imperative",
"description": "Read the current puzzle: guesses played so far, per-letter feedback, what has been ruled in or out, attempts left, and whose turn it is. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}suggest_guessesAction
Rank candidate guesses. Pool words still fit every clue and can win now; play one of those to solve. A probe cannot win this turn — only use it to split the pool. why is at most 60 characters. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.
View tool JSON
{
"name": "suggest_guesses",
"kind": "act",
"impl": "imperative",
"description": "Rank candidate guesses. Pool words still fit every clue and can win now; play one of those to solve. A probe cannot win this turn — only use it to split the pool. why is at most 60 characters. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.",
"inputSchema": {
"type": "object",
"properties": {
"count": {
"type": "integer",
"minimum": 1,
"maximum": 5,
"default": 3,
"description": "How many candidates to return (1-5)."
}
},
"additionalProperties": false
}
}propose_guessAction
Stage a five-letter word for the player to accept or reject. Does not play the word. Replaces any existing proposal. Blocked when the player has the agent on observe-only.
View tool JSON
{
"name": "propose_guess",
"kind": "act",
"impl": "imperative",
"description": "Stage a five-letter word for the player to accept or reject. Does not play the word. Replaces any existing proposal. Blocked when the player has the agent on observe-only.",
"inputSchema": {
"type": "object",
"properties": {
"word": {
"type": "string",
"description": "A five-letter word to stage for the player's approval."
},
"reason": {
"type": "string",
"description": "One short sentence on why this word is a good move."
}
},
"required": [
"word"
],
"additionalProperties": false
}
}submit_guessSensitive Action
Play a five-letter word now. Blocked in co-pilot (use propose_guess) and observe-only. Updates the board immediately. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.
View tool JSON
{
"name": "submit_guess",
"kind": "transact",
"impl": "imperative",
"description": "Play a five-letter word now. Blocked in co-pilot (use propose_guess) and observe-only. Updates the board immediately. Feedback is a five-character string: G = correct position, Y = present elsewhere, . = absent. Positions are 1-based.",
"inputSchema": {
"type": "object",
"properties": {
"word": {
"type": "string",
"description": "The five-letter word to play."
}
},
"required": [
"word"
],
"additionalProperties": false
}
}new_gameAction
Start a new random puzzle. Clears the board and any proposal. Keeps the current collaboration mode. Six attempts.
View tool JSON
{
"name": "new_game",
"kind": "act",
"impl": "imperative",
"description": "Start a new random puzzle. Clears the board and any proposal. Keeps the current collaboration mode. Six attempts.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}