anpord 0.1.13 → 0.1.17

package/dist/{client-BfdjFtlV.mjs → client-DE3YNtWc.mjs}

@@ -1,6 +1,6 @@
-import { C as EvalTrigger, E as EvalVerify, T as EvalPrompt, _ as HarnessProfile, a as EvalPageCursor, c as EvalRunPage, d as EvalValidator, f as EvalVariables, h as StartedEval, i as EvalName, l as EvalSandbox, m as RerunCellRequest, n as EvalCellHistoryEntry, o as EvalPrepare, p as ModelCatalogue, r as EvalHarness, s as EvalRun, t as CaseCache, u as EvalSource, v as PROFILE_HARNESS_RULE, w as EvalCaseName, x as profileFitsHarness } from "./evals-Dcc_x6jW.mjs";
-import { FetchHttpClient, HttpApi, HttpApiClient, HttpApiEndpoint, HttpApiGroup, HttpApiMiddleware, HttpApiSchema, HttpApiSecurity, HttpClient, HttpClientRequest, OpenApi } from "@effect/platform";
-import { Context, Effect, Layer, Redacted, Schema } from "effect";
+import { g as NotFound, h as Forbidden, m as Conflict, p as BadRequest, r as ApiKeyAuthentication, t as PublicEvalsGroup } from "./evals-api-CsdU741J.mjs";
+import { Effect, Layer, Redacted, Schema } from "effect";
+import { FetchHttpClient, HttpApi, HttpApiClient, HttpApiEndpoint, HttpApiGroup, HttpClient, HttpClientRequest, OpenApi } from "@effect/platform";
 //#region ../schema/src/domain/prompts.ts
 const ChannelName = Schema.String.pipe(Schema.minLength(1), Schema.maxLength(36), Schema.pattern(/^[a-z0-9][a-z0-9_-]*$/, { message: () => "Channel must be lowercase alphanumeric, optionally with - or _" }), Schema.brand("ChannelName")).annotations({ description: "Addresses a version. `latest` is derived from the highest version rather than stored, so it cannot drift from the version table." });
 ChannelName.make("production");
@@ -93,93 +93,6 @@ Schema.Struct({
 	version: Schema.optional(VersionNumber)
 });
 //#endregion
-//#region ../schema/src/domain/errors.ts
-var NotFound = class extends Schema.TaggedError()("NotFound", { message: Schema.String }, HttpApiSchema.annotations({ status: 404 })) {};
-var Conflict = class extends Schema.TaggedError()("Conflict", { message: Schema.String }, HttpApiSchema.annotations({ status: 409 })) {};
-var BadRequest = class extends Schema.TaggedError()("BadRequest", { message: Schema.String }, HttpApiSchema.annotations({ status: 400 })) {};
-var Unauthorized = class extends Schema.TaggedError()("Unauthorized", { message: Schema.String }, HttpApiSchema.annotations({ status: 401 })) {};
-var Forbidden = class extends Schema.TaggedError()("Forbidden", { message: Schema.String }, HttpApiSchema.annotations({ status: 403 })) {};
-Schema.TaggedError()("InternalError", { message: Schema.String }, HttpApiSchema.annotations({ status: 500 }));
-//#endregion
-//#region ../schema/src/internal/authentication.ts
-var CurrentActor = class extends Context.Tag("@anpord/schema/CurrentActor")() {};
-HttpApiMiddleware.Tag()("@anpord/schema/Authentication", {
-	failure: Unauthorized,
-	provides: CurrentActor,
-	security: { session: HttpApiSecurity.apiKey({
-		in: "cookie",
-		key: "anpord.session_token"
-	}) }
-});
-//#endregion
-//#region ../schema/src/public/authentication.ts
-var ApiKeyAuthentication = class extends HttpApiMiddleware.Tag()("@anpord/schema/ApiKeyAuthentication", {
-	failure: Unauthorized,
-	provides: CurrentActor,
-	security: { bearer: HttpApiSecurity.bearer }
-}) {};
-//#endregion
-//#region ../schema/src/public/evals-api.ts
-const EvalRunRequest = Schema.Struct({ id: Schema.String }).annotations({
-	description: "Select an eval run by id.",
-	identifier: "EvalRunRequest"
-});
-const EvalCellRequest = Schema.Struct({ cellKey: Schema.String }).annotations({
-	description: "Select an eval cell by its stable key.",
-	identifier: "EvalCellRequest"
-});
-const EvalModelsRequest = Schema.Struct({
-	harness: EvalHarness,
-	q: Schema.optional(Schema.String)
-}).annotations({
-	description: "Select a harness whose available models should be listed.",
-	identifier: "EvalModelsRequest"
-});
-const PublicEvalSandbox = EvalSandbox.annotations({
-	description: "The hosted sandbox a task runs in.",
-	identifier: "PublicEvalSandbox"
-});
-const ListEvalsRequest = Schema.Struct({
-	cursor: Schema.optional(Schema.NullOr(EvalPageCursor)),
-	limit: Schema.optional(Schema.Int)
-}).annotations({
-	description: "Where to read from, and how much.",
-	identifier: "ListEvalsRequest"
-});
-const PublicEvalCase = Schema.Struct({
-	cache: Schema.optional(CaseCache),
-	name: EvalCaseName,
-	prepare: Schema.optional(Schema.NullOr(EvalPrepare)),
-	source: Schema.optional(EvalSource),
-	validator: Schema.optional(Schema.NullOr(EvalValidator)),
-	variables: Schema.optional(EvalVariables),
-	verify: Schema.NullOr(EvalVerify)
-}).pipe(Schema.filter(({ validator, verify }) => !(validator !== void 0 && validator !== null && verify !== null), { message: () => "Use either validator or verify, not both." })).annotations({
-	description: "A task, workspace source, setup command, and verifier.",
-	identifier: "StartEvalCase"
-});
-const PublicEvalTask = Schema.Struct({
-	harness: EvalHarness,
-	model: Schema.String.pipe(Schema.minLength(1)),
-	profile: Schema.optional(HarnessProfile),
-	sandbox: Schema.optional(PublicEvalSandbox)
-}).pipe(Schema.filter(profileFitsHarness, { message: () => PROFILE_HARNESS_RULE })).annotations({
-	description: `A harness and model, with an optional sandbox and an optional profile layered on the harness. Omit the sandbox to use the default. ${PROFILE_HARNESS_RULE}`,
-	identifier: "StartEvalTask"
-});
-const PublicStartEvalRequest = Schema.Struct({
-	trigger: Schema.optional(EvalTrigger),
-	cases: Schema.Array(PublicEvalCase).pipe(Schema.minItems(1), Schema.maxItems(100)),
-	name: Schema.optional(EvalName),
-	prompt: EvalPrompt,
-	tasks: Schema.Array(PublicEvalTask).pipe(Schema.minItems(1), Schema.maxItems(20)),
-	trials: Schema.Int.pipe(Schema.between(1, 10))
-}).annotations({
-	description: `Start a grid with at most 100 total case, task, and trial combinations.`,
-	identifier: "StartEvalRequest"
-});
-var PublicEvalsGroup = class extends HttpApiGroup.make("evals").add(HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(EvalRunPage).annotate(OpenApi.Summary, "List eval runs").annotate(OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(StartedEval).annotate(OpenApi.Summary, "Start an eval run").annotate(OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${PROFILE_HARNESS_RULE}`)).add(HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(EvalRun).annotate(OpenApi.Summary, "Get an eval run")).add(HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(Schema.Array(EvalCellHistoryEntry)).annotate(OpenApi.Summary, "List a cell's history").annotate(OpenApi.Description, "Returns the 20 most recent results.")).add(HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(Schema.extend(EvalRunRequest, Schema.extend(EvalCellRequest, RerunCellRequest))).addSuccess(StartedEval).annotate(OpenApi.Summary, "Rerun one cell")).add(HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(ModelCatalogue).annotate(OpenApi.Summary, "List models available to the harness").annotate(OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(OpenApi.Title, "Evals").annotate(OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
-//#endregion
 //#region ../schema/src/public/requests.ts
 const GetPromptRequest = Schema.Struct({
 	channel: Schema.optional(ChannelName),

package/dist/{client-C-P2I00z.cjs → client-DziJjMFh.cjs}

@@ -1,6 +1,6 @@
-const require_evals = require("./evals-DouMsrsM.cjs");
-let _effect_platform = require("@effect/platform");
+const require_evals_api = require("./evals-api-CqhPZCr5.cjs");
 let effect = require("effect");
+let _effect_platform = require("@effect/platform");
 //#region ../schema/src/domain/prompts.ts
 const ChannelName = effect.Schema.String.pipe(effect.Schema.minLength(1), effect.Schema.maxLength(36), effect.Schema.pattern(/^[a-z0-9][a-z0-9_-]*$/, { message: () => "Channel must be lowercase alphanumeric, optionally with - or _" }), effect.Schema.brand("ChannelName")).annotations({ description: "Addresses a version. `latest` is derived from the highest version rather than stored, so it cannot drift from the version table." });
 ChannelName.make("production");
@@ -93,93 +93,6 @@ effect.Schema.Struct({
 	version: effect.Schema.optional(VersionNumber)
 });
 //#endregion
-//#region ../schema/src/domain/errors.ts
-var NotFound = class extends effect.Schema.TaggedError()("NotFound", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 404 })) {};
-var Conflict = class extends effect.Schema.TaggedError()("Conflict", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 409 })) {};
-var BadRequest = class extends effect.Schema.TaggedError()("BadRequest", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 400 })) {};
-var Unauthorized = class extends effect.Schema.TaggedError()("Unauthorized", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 401 })) {};
-var Forbidden = class extends effect.Schema.TaggedError()("Forbidden", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 403 })) {};
-effect.Schema.TaggedError()("InternalError", { message: effect.Schema.String }, _effect_platform.HttpApiSchema.annotations({ status: 500 }));
-//#endregion
-//#region ../schema/src/internal/authentication.ts
-var CurrentActor = class extends effect.Context.Tag("@anpord/schema/CurrentActor")() {};
-_effect_platform.HttpApiMiddleware.Tag()("@anpord/schema/Authentication", {
-	failure: Unauthorized,
-	provides: CurrentActor,
-	security: { session: _effect_platform.HttpApiSecurity.apiKey({
-		in: "cookie",
-		key: "anpord.session_token"
-	}) }
-});
-//#endregion
-//#region ../schema/src/public/authentication.ts
-var ApiKeyAuthentication = class extends _effect_platform.HttpApiMiddleware.Tag()("@anpord/schema/ApiKeyAuthentication", {
-	failure: Unauthorized,
-	provides: CurrentActor,
-	security: { bearer: _effect_platform.HttpApiSecurity.bearer }
-}) {};
-//#endregion
-//#region ../schema/src/public/evals-api.ts
-const EvalRunRequest = effect.Schema.Struct({ id: effect.Schema.String }).annotations({
-	description: "Select an eval run by id.",
-	identifier: "EvalRunRequest"
-});
-const EvalCellRequest = effect.Schema.Struct({ cellKey: effect.Schema.String }).annotations({
-	description: "Select an eval cell by its stable key.",
-	identifier: "EvalCellRequest"
-});
-const EvalModelsRequest = effect.Schema.Struct({
-	harness: require_evals.EvalHarness,
-	q: effect.Schema.optional(effect.Schema.String)
-}).annotations({
-	description: "Select a harness whose available models should be listed.",
-	identifier: "EvalModelsRequest"
-});
-const PublicEvalSandbox = require_evals.EvalSandbox.annotations({
-	description: "The hosted sandbox a task runs in.",
-	identifier: "PublicEvalSandbox"
-});
-const ListEvalsRequest = effect.Schema.Struct({
-	cursor: effect.Schema.optional(effect.Schema.NullOr(require_evals.EvalPageCursor)),
-	limit: effect.Schema.optional(effect.Schema.Int)
-}).annotations({
-	description: "Where to read from, and how much.",
-	identifier: "ListEvalsRequest"
-});
-const PublicEvalCase = effect.Schema.Struct({
-	cache: effect.Schema.optional(require_evals.CaseCache),
-	name: require_evals.EvalCaseName,
-	prepare: effect.Schema.optional(effect.Schema.NullOr(require_evals.EvalPrepare)),
-	source: effect.Schema.optional(require_evals.EvalSource),
-	validator: effect.Schema.optional(effect.Schema.NullOr(require_evals.EvalValidator)),
-	variables: effect.Schema.optional(require_evals.EvalVariables),
-	verify: effect.Schema.NullOr(require_evals.EvalVerify)
-}).pipe(effect.Schema.filter(({ validator, verify }) => !(validator !== void 0 && validator !== null && verify !== null), { message: () => "Use either validator or verify, not both." })).annotations({
-	description: "A task, workspace source, setup command, and verifier.",
-	identifier: "StartEvalCase"
-});
-const PublicEvalTask = effect.Schema.Struct({
-	harness: require_evals.EvalHarness,
-	model: effect.Schema.String.pipe(effect.Schema.minLength(1)),
-	profile: effect.Schema.optional(require_evals.HarnessProfile),
-	sandbox: effect.Schema.optional(PublicEvalSandbox)
-}).pipe(effect.Schema.filter(require_evals.profileFitsHarness, { message: () => require_evals.PROFILE_HARNESS_RULE })).annotations({
-	description: `A harness and model, with an optional sandbox and an optional profile layered on the harness. Omit the sandbox to use the default. ${require_evals.PROFILE_HARNESS_RULE}`,
-	identifier: "StartEvalTask"
-});
-const PublicStartEvalRequest = effect.Schema.Struct({
-	trigger: effect.Schema.optional(require_evals.EvalTrigger),
-	cases: effect.Schema.Array(PublicEvalCase).pipe(effect.Schema.minItems(1), effect.Schema.maxItems(100)),
-	name: effect.Schema.optional(require_evals.EvalName),
-	prompt: require_evals.EvalPrompt,
-	tasks: effect.Schema.Array(PublicEvalTask).pipe(effect.Schema.minItems(1), effect.Schema.maxItems(20)),
-	trials: effect.Schema.Int.pipe(effect.Schema.between(1, 10))
-}).annotations({
-	description: `Start a grid with at most 100 total case, task, and trial combinations.`,
-	identifier: "StartEvalRequest"
-});
-var PublicEvalsGroup = class extends _effect_platform.HttpApiGroup.make("evals").add(_effect_platform.HttpApiEndpoint.post("list", "/evals.list").setPayload(ListEvalsRequest).addSuccess(require_evals.EvalRunPage).annotate(_effect_platform.OpenApi.Summary, "List eval runs").annotate(_effect_platform.OpenApi.Description, "Newest first. Pass the `next` cursor from a response to read the page after it; a null `next` means there are no more.")).add(_effect_platform.HttpApiEndpoint.post("start", "/evals.start").setPayload(PublicStartEvalRequest).addSuccess(require_evals.StartedEval).annotate(_effect_platform.OpenApi.Summary, "Start an eval run").annotate(_effect_platform.OpenApi.Description, `Starts the grid and returns its id while trials continue in the background. ${require_evals.PROFILE_HARNESS_RULE}`)).add(_effect_platform.HttpApiEndpoint.post("get", "/evals.get").setPayload(EvalRunRequest).addSuccess(require_evals.EvalRun).annotate(_effect_platform.OpenApi.Summary, "Get an eval run")).add(_effect_platform.HttpApiEndpoint.post("cellHistory", "/evals.cellHistory").setPayload(EvalCellRequest).addSuccess(effect.Schema.Array(require_evals.EvalCellHistoryEntry)).annotate(_effect_platform.OpenApi.Summary, "List a cell's history").annotate(_effect_platform.OpenApi.Description, "Returns the 20 most recent results.")).add(_effect_platform.HttpApiEndpoint.post("rerunCell", "/evals.rerunCell").setPayload(effect.Schema.extend(EvalRunRequest, effect.Schema.extend(EvalCellRequest, require_evals.RerunCellRequest))).addSuccess(require_evals.StartedEval).annotate(_effect_platform.OpenApi.Summary, "Rerun one cell")).add(_effect_platform.HttpApiEndpoint.post("models", "/evals.models").setPayload(EvalModelsRequest).addSuccess(require_evals.ModelCatalogue).annotate(_effect_platform.OpenApi.Summary, "List models available to the harness").annotate(_effect_platform.OpenApi.Description, "The command harness has no catalogue of its own, so its list is empty: the model is whatever the profile's run command reads from ANPORD_MODEL.")).addError(BadRequest).addError(Conflict).addError(Forbidden).addError(NotFound).middleware(ApiKeyAuthentication).annotate(_effect_platform.OpenApi.Title, "Evals").annotate(_effect_platform.OpenApi.Description, "Run cases across harness, model, and sandbox combinations and compare the results with their baselines.") {};
-//#endregion
 //#region ../schema/src/public/requests.ts
 const GetPromptRequest = effect.Schema.Struct({
 	channel: effect.Schema.optional(ChannelName),
@@ -285,10 +198,10 @@ const PromptList = effect.Schema.Struct({ data: effect.Schema.Array(PublicPrompt
 });
 //#endregion
 //#region ../schema/src/public/prompts-api.ts
-var PublicPromptsGroup = class extends _effect_platform.HttpApiGroup.make("prompts").add(_effect_platform.HttpApiEndpoint.post("get", "/prompts.get").setPayload(GetPromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Resolve a prompt").annotate(_effect_platform.OpenApi.Description, "Returns the content a caller should send to a model. With no selector this follows the organization's default channel.")).add(_effect_platform.HttpApiEndpoint.post("list", "/prompts.list").setPayload(ListPromptsRequest).addSuccess(PromptList).annotate(_effect_platform.OpenApi.Summary, "List prompts").annotate(_effect_platform.OpenApi.Description, "Up to 100 prompts in the organization, without content.")).add(_effect_platform.HttpApiEndpoint.post("create", "/prompts.create").setPayload(CreatePromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Create a prompt").annotate(_effect_platform.OpenApi.Description, "Creates the prompt and its first version in one call.")).add(_effect_platform.HttpApiEndpoint.post("update", "/prompts.update").setPayload(UpdatePromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Add a version").annotate(_effect_platform.OpenApi.Description, "Content is versioned, so updating a prompt appends a version rather than overwriting one. Earlier versions stay readable.")).add(_effect_platform.HttpApiEndpoint.post("promote", "/prompts.promote").setPayload(PromotePromptRequest).addSuccess(Ok).annotate(_effect_platform.OpenApi.Summary, "Promote a version to a channel").annotate(_effect_platform.OpenApi.Description, "Points a channel, such as production, at a version. This is how a version goes live without callers changing anything.")).addError(BadRequest).addError(Forbidden).addError(Conflict).addError(NotFound).middleware(ApiKeyAuthentication).annotate(_effect_platform.OpenApi.Title, "Prompts").annotate(_effect_platform.OpenApi.Description, "Read and write prompts. Content is versioned, so writes append rather than overwrite, and channels decide which version callers receive.") {};
+var PublicPromptsGroup = class extends _effect_platform.HttpApiGroup.make("prompts").add(_effect_platform.HttpApiEndpoint.post("get", "/prompts.get").setPayload(GetPromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Resolve a prompt").annotate(_effect_platform.OpenApi.Description, "Returns the content a caller should send to a model. With no selector this follows the organization's default channel.")).add(_effect_platform.HttpApiEndpoint.post("list", "/prompts.list").setPayload(ListPromptsRequest).addSuccess(PromptList).annotate(_effect_platform.OpenApi.Summary, "List prompts").annotate(_effect_platform.OpenApi.Description, "Up to 100 prompts in the organization, without content.")).add(_effect_platform.HttpApiEndpoint.post("create", "/prompts.create").setPayload(CreatePromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Create a prompt").annotate(_effect_platform.OpenApi.Description, "Creates the prompt and its first version in one call.")).add(_effect_platform.HttpApiEndpoint.post("update", "/prompts.update").setPayload(UpdatePromptRequest).addSuccess(PublicPromptWithVersions).annotate(_effect_platform.OpenApi.Summary, "Add a version").annotate(_effect_platform.OpenApi.Description, "Content is versioned, so updating a prompt appends a version rather than overwriting one. Earlier versions stay readable.")).add(_effect_platform.HttpApiEndpoint.post("promote", "/prompts.promote").setPayload(PromotePromptRequest).addSuccess(Ok).annotate(_effect_platform.OpenApi.Summary, "Promote a version to a channel").annotate(_effect_platform.OpenApi.Description, "Points a channel, such as production, at a version. This is how a version goes live without callers changing anything.")).addError(require_evals_api.BadRequest).addError(require_evals_api.Forbidden).addError(require_evals_api.Conflict).addError(require_evals_api.NotFound).middleware(require_evals_api.ApiKeyAuthentication).annotate(_effect_platform.OpenApi.Title, "Prompts").annotate(_effect_platform.OpenApi.Description, "Read and write prompts. Content is versioned, so writes append rather than overwrite, and channels decide which version callers receive.") {};
 //#endregion
 //#region ../schema/src/public/api.ts
-var PublicApi = class extends _effect_platform.HttpApi.make("anpord-public").add(PublicEvalsGroup).add(PublicPromptsGroup).prefix("/v1").annotate(_effect_platform.OpenApi.Title, "Anpord API").annotate(_effect_platform.OpenApi.Version, "1.0.0").annotate(_effect_platform.OpenApi.Description, "Run agent evals and manage prompts. Every endpoint takes a JSON body over POST and authenticates with a bearer API key.").annotate(_effect_platform.OpenApi.Servers, [{
+var PublicApi = class extends _effect_platform.HttpApi.make("anpord-public").add(require_evals_api.PublicEvalsGroup).add(PublicPromptsGroup).prefix("/v1").annotate(_effect_platform.OpenApi.Title, "Anpord API").annotate(_effect_platform.OpenApi.Version, "1.0.0").annotate(_effect_platform.OpenApi.Description, "Run agent evals and manage prompts. Every endpoint takes a JSON body over POST and authenticates with a bearer API key.").annotate(_effect_platform.OpenApi.Servers, [{
 	description: "Production",
 	url: "https://api.anpord.com"
 }]) {};