Agent tooling
Bounded TypeSafe Jev workflows for coding agents.
import { TypeSafeClient } from "@typesafe-ai/sdk";The link points at the commit we read, so the line number still holds.
These question sets are lifted from this project's source exactly as written.
from examples/workflows/stale-todo-audit.ts:28
stalechoiceIs this TODO comment still actionable?
actionable — Describes concrete remaining work.stale — Refers to work that is clearly done or no longer relevant.cannot_tell — The comment alone does not say.import { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
stale: { type: 'choice', instructions: 'Is this TODO comment still actionable?',
criteria: { actionable: 'Describes concrete remaining work.', stale: 'Refers to work that is clearly done or no longer relevant.', cannot_tell: 'The comment alone does not say.' } },
},
});from src/workflows/analyze-diff.ts:359
evidence_sufficienttrue/falseIs the shown diff evidence sufficient for the selected classification?
import { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
evidence_sufficient: { type: 'boolean', instructions: 'Is the shown diff evidence sufficient for the selected classification?' },
},
});from src/workflows/check-task.ts:195
weakens_expectationtrue/falseDoes test hunk ${hunk.id} weaken what the test expects?
expectation_change_stated_in_tasktrue/falseDoes the requested behavior require or clearly entail that the old expectation in hunk ${hunk.id} should no longer hold?
import { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
weakens_expectation: { type: 'boolean', instructions: 'Does test hunk ${hunk.id} weaken what the test expects?' },
expectation_change_stated_in_task: { type: 'boolean', instructions: 'Does the requested behavior require or clearly entail that the old expectation in hunk ${hunk.id} should no longer hold?' },
},
});