Wave A: pi-native tool layer — delete LhTool/toPiTool, TypeBox everywhere, terminate on questions
- src/agent/lh-tool.ts deleted: the zod-wrapped LhTool + toPiTool bridge (z.toJSONSchema per call, clientQuestion flag, hand-rolled result mapping) is gone. Tools are pi-agent-core native AgentTools now. - New src/agent/tool-execute.ts owns what the wrapper added: the abort guard (runWithAbortGuard: run signal + 30s per-call timeout + 250ms grace, 'Operation aborted' / timeout wording preserved), ToolInputError (the retryable '- problem at path' format), result mapping (string verbatim / object JSON / image parts, details rides the record's tool part output), and questionResult() — placeholder + terminate:true (owner Option A: belt of terminate + slim makeQuestionStopper with the clean-question rule; maxTurns deleted from the teaching turn entirely per owner). - All 11 tool files + tools.ts registries converted to TypeBox parameters (typebox@1.3.7, pinned to pi-agent-core's copy). zod refinements (quizInputProblems, askOptionProblems, live-nodeId) moved from schemas into execute — the model sees the same retryable error shape. - question-core + contracts/probe TypeBox-authored (single source shared with the UI); tool-parts parse seam runs the boundary's full chain (Convert → Check → Clean → refinement), fixing the live-smoke lockout: pi's boundary Value.Converts correctAnswer: 17 → '17' but the record keeps raw args, so a parse without Convert rejected a posed quiz (invisible pending question, turn ended). Pinned with the real smoke emission shape. - Subagents converted: demo-builder/researcher AgentTool-native, visual-maker/pdf-reader declaration-only pi-ai Tools with local execute fns, extraction submit schemas re-authored in TypeBox (owner option 2: zod defaults → Type.Optional + fill at the submit seam). - Accepted delta, flagged: TypeBox-native schemas skip pi's JSON-Schema primitive coercion fast path... (boundary still Value.Converts), and prober-quiz unknown keys ride through (Value.Clean strips at display). - Tests: 988 passing (was 967 at the recon baseline); tool-abort and read-tool rewritten against the native seams; question-core/question- tools/tool-parts/demos/web-search/graph-tools/plan-tree/probe-done/ tool-toggles converted to Value.Check/execute pins. Live smoke (PORT=3004, real model): native ls turn streams end-to-end; a real quiz pose (first attempt missing correctAnswer → tool-input-error → model retry → posed) terminates the turn cleanly and persists pending.
This commit is contained in:
amosblomqvist
+70
-42
@@ -1,6 +1,9 @@
|
||||
import { lhTool, type LhTool } from "./lh-tool.ts";
|
||||
import { Type } from "typebox";
|
||||
import type { AgentTool } from "@earendil-works/pi-agent-core";
|
||||
import { questionResult, serverExecute, ToolInputError } from "./tool-execute.ts";
|
||||
import {
|
||||
askInputSchema,
|
||||
askOptionProblems,
|
||||
probeQuizInputSchema,
|
||||
quizInputObject,
|
||||
quizInputProblems,
|
||||
@@ -25,14 +28,22 @@ export type { ProbeDoneContext } from "./tools/probe.ts";
|
||||
* The agent-split tool composition (docs/probe-teach-split.md §1): the
|
||||
* PROBER and the TEACHER are different agents with different tool sets.
|
||||
* The question tools share a NAME but are genuinely DIFFERENT tools —
|
||||
* the prober's quiz has NO nodeId field at all (its schema is built from
|
||||
* quizInputBase; zod strips the key if the model tries anyway), the
|
||||
* the prober's quiz has NO nodeId field at all (its declaration is built
|
||||
* from quizInputBase; the model's schema never mentions nodeId), the
|
||||
* teacher's carries the validated position field.
|
||||
*
|
||||
* - Question tools (quiz / ask_user_question): CLIENT-SIDE (no `execute`).
|
||||
* The call streams to the browser in `input-available` state; the
|
||||
* learner answers in the dock; the client grades and sends the verdict
|
||||
* with addToolOutput. Input schemas live in core/question-core.
|
||||
* - Question tools (quiz / ask_user_question): CLIENT-ANSWERED. Their
|
||||
* execute validates the authoring refinements (the cross-field rules no
|
||||
* schema carries — quizInputProblems / askOptionProblems; a failure
|
||||
* throws ToolInputError, which pi's loop converts to an error tool
|
||||
* result the model retries on), then returns the synthetic
|
||||
* "question posed" placeholder (questionResult: terminate: true — the
|
||||
* turn ends; the call streams to the browser in `input-available`
|
||||
* state; the learner answers in the dock; the client grades and sends
|
||||
* the verdict with addToolOutput). Input schemas live in
|
||||
* core/question-core. Question tools opt into SEQUENTIAL execution so
|
||||
* parallel calls resolve in the model's call order (the dock's subject
|
||||
* can never flip mid-turn).
|
||||
* - probe_done (src/agent/tools/probe.ts): SERVER-SIDE seam tool.
|
||||
* - planTools (src/agent/tools/plan.ts): SERVER-SIDE — author-once
|
||||
* plan_init. Teacher only.
|
||||
@@ -40,22 +51,28 @@ export type { ProbeDoneContext } from "./tools/probe.ts";
|
||||
* only.
|
||||
*/
|
||||
|
||||
export const askUserQuestionTool: LhTool<any, any> = {
|
||||
export const askUserQuestionTool: AgentTool<typeof askInputSchema> = {
|
||||
name: "ask_user_question",
|
||||
clientQuestion: true,
|
||||
label: "ask_user_question",
|
||||
description:
|
||||
"Ask the learner a single question and pause until they answer. NOT graded — there is no correct answer. " +
|
||||
"Use when requirements are ambiguous, preferences are needed, a decision would materially affect the lesson, " +
|
||||
"or you need an open-ended response. ONE question per turn: call a single ask_user_question, then end your " +
|
||||
"turn — his answer determines the next.",
|
||||
schema: askInputSchema,
|
||||
parameters: askInputSchema,
|
||||
executionMode: "sequential",
|
||||
execute: async (_toolCallId, input) => {
|
||||
const problems = askOptionProblems(input.options ?? []);
|
||||
if (problems.length > 0) throw new ToolInputError("ask_user_question", problems);
|
||||
return questionResult();
|
||||
},
|
||||
};
|
||||
|
||||
// ─── the prober's question tools (no nodeId — no map exists yet) ─────────────
|
||||
|
||||
export const probeQuizTool: LhTool<any, any> = {
|
||||
export const probeQuizTool: AgentTool<typeof probeQuizInputSchema> = {
|
||||
name: "quiz",
|
||||
clientQuestion: true,
|
||||
label: "quiz",
|
||||
description:
|
||||
"Ask the learner a GRADED question with a known correct answer, then instantly grade and give feedback. " +
|
||||
"Unlike ask_user_question (which collects preferences/decisions with no right answer), quiz always has a " +
|
||||
@@ -67,7 +84,13 @@ export const probeQuizTool: LhTool<any, any> = {
|
||||
"free-text note to ANY answer; it reaches you only when non-empty. No free-text answers — for non-graded " +
|
||||
"questions use ask_user_question instead. ONE question per turn: call a single quiz, then end your turn — " +
|
||||
"his answer determines the next question to ask.",
|
||||
schema: probeQuizInputSchema,
|
||||
parameters: probeQuizInputSchema,
|
||||
executionMode: "sequential",
|
||||
execute: async (_toolCallId, input) => {
|
||||
const problems = quizInputProblems(input);
|
||||
if (problems.length > 0) throw new ToolInputError("quiz", problems);
|
||||
return questionResult();
|
||||
},
|
||||
};
|
||||
|
||||
/** The prober's registry context: the probe's own needs (probe_done —
|
||||
@@ -98,7 +121,7 @@ export interface ProberToolContext extends ProbeDoneContext {
|
||||
export function createProberTools(
|
||||
ctx: ProberToolContext,
|
||||
enabled?: EnabledTools,
|
||||
): Record<string, LhTool<any, any>> {
|
||||
): Record<string, AgentTool<any>> {
|
||||
// probe_done is re-added AFTER the filter, always-on by construction — it
|
||||
// is how the phase ENDS; a prober that can never conclude is a lockout.
|
||||
const done = probeDoneTool(ctx);
|
||||
@@ -146,36 +169,31 @@ export function createProberTools(
|
||||
|
||||
// ─── the teacher's question tools (quiz carries the position field) ──────────
|
||||
|
||||
/** The nodeId-against-live-tree check the teacher's quiz chains after the
|
||||
* shared authoring contract (the old superRefine, verbatim). */
|
||||
function nodeIdProblem(ctx: Pick<PlanToolContext, "chatId" | "chatStore">, nodeId: string): string | null {
|
||||
const record = ctx.chatStore.getChat(ctx.chatId);
|
||||
const known = record?.tree?.nodes.map((n) => n.id) ?? [];
|
||||
if (!known.includes(nodeId)) {
|
||||
return (
|
||||
`nodeId "${nodeId}" matches no node in this session's plan map. ` +
|
||||
(known.length > 0
|
||||
? `Copy one of your plan_init slugs exactly (${known.slice(0, 5).join(", ")}${known.length > 5 ? ", …" : ""}), or omit nodeId for an ad-hoc check.`
|
||||
: "No plan map exists yet — omit nodeId (ad-hoc checks only until you call plan_init).")
|
||||
);
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
/** Build the teach-side quiz tool for one request (exported for tests): the authoring contract
|
||||
* + the per-session plan-nodeId validation (chained superRefines — the
|
||||
* + the per-session plan-nodeId validation (both run in execute — the
|
||||
* grading contract is untouched). Reads the LIVE chat record, so rewrites
|
||||
* mid-turn are honored, and with no tree yet every nodeId is rejected.
|
||||
* Omitting nodeId is legal — an ad-hoc check that moves nothing. */
|
||||
export function teachQuizToolFor(ctx: Pick<PlanToolContext, "chatId" | "chatStore">) {
|
||||
const inputSchema = quizInputObject
|
||||
.superRefine((input, add) => {
|
||||
for (const problem of quizInputProblems(input)) {
|
||||
add.addIssue({ code: "custom", message: problem });
|
||||
}
|
||||
})
|
||||
.superRefine((input, add) => {
|
||||
if (typeof input.nodeId !== "string" || input.nodeId.length === 0) return;
|
||||
const record = ctx.chatStore.getChat(ctx.chatId);
|
||||
const known = record?.tree?.nodes.map((n) => n.id) ?? [];
|
||||
if (!known.includes(input.nodeId)) {
|
||||
add.addIssue({
|
||||
code: "custom",
|
||||
message:
|
||||
`nodeId "${input.nodeId}" matches no node in this session's plan map. ` +
|
||||
(known.length > 0
|
||||
? `Copy one of your plan_init slugs exactly (${known.slice(0, 5).join(", ")}${known.length > 5 ? ", …" : ""}), or omit nodeId for an ad-hoc check.`
|
||||
: "No plan map exists yet — omit nodeId (ad-hoc checks only until you call plan_init)."),
|
||||
});
|
||||
}
|
||||
});
|
||||
return lhTool({
|
||||
export function teachQuizToolFor(ctx: Pick<PlanToolContext, "chatId" | "chatStore">): AgentTool<typeof quizInputObject> {
|
||||
return {
|
||||
name: "quiz",
|
||||
clientQuestion: true,
|
||||
label: "quiz",
|
||||
description:
|
||||
"Ask the learner a GRADED question with a known correct answer, then instantly grade and give feedback. " +
|
||||
"Unlike ask_user_question (which collects preferences/decisions with no right answer), quiz always has a " +
|
||||
@@ -187,8 +205,18 @@ export function teachQuizToolFor(ctx: Pick<PlanToolContext, "chatId" | "chatStor
|
||||
"An always-present optional note field lets the learner attach a free-text note to ANY answer; it reaches you " +
|
||||
"only when non-empty. No free-text answers — for non-graded questions use ask_user_question instead. " +
|
||||
"ONE question per turn: call a single quiz, then end your turn — his answer determines the next.",
|
||||
schema: inputSchema,
|
||||
});
|
||||
parameters: quizInputObject,
|
||||
executionMode: "sequential",
|
||||
execute: async (_toolCallId, input) => {
|
||||
const problems = [...quizInputProblems(input)];
|
||||
if (typeof input.nodeId === "string" && input.nodeId.length > 0) {
|
||||
const nodeIdProblemText = nodeIdProblem(ctx, input.nodeId);
|
||||
if (nodeIdProblemText) problems.push(nodeIdProblemText);
|
||||
}
|
||||
if (problems.length > 0) throw new ToolInputError("quiz", problems);
|
||||
return questionResult();
|
||||
},
|
||||
};
|
||||
}
|
||||
|
||||
export interface TeacherToolContext extends PlanToolContext {
|
||||
@@ -219,7 +247,7 @@ export interface TeacherToolContext extends PlanToolContext {
|
||||
export function createTeacherTools(
|
||||
ctx: TeacherToolContext,
|
||||
enabled?: EnabledTools,
|
||||
): Record<string, LhTool<any, any>> {
|
||||
): Record<string, AgentTool<any>> {
|
||||
return filterEnabledTools(
|
||||
{
|
||||
quiz: teachQuizToolFor(ctx),
|
||||
|
||||
Reference in New Issue
Block a user