diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 854f4334..e42f62ab 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -81,21 +81,21 @@ jobs: - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-sdk - ref: 3645e7d0a7b7755e16fafa252db4abb7ef4058c7 # @tangle-network/agent-interface@0.40.0 + ref: fc00613985c86f917a679c9bf957e6af8cbedb7c # @tangle-network/agent-interface@0.43.0 path: .cohort/agent-sdk persist-credentials: false - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-eval - ref: 55d2faacc6ba2c92ad21b42b9b607ab16ae0fc7c # v0.140.1 + ref: f379f2aa347e416fdb28c151ba54bc93b64e4f65 # v0.142.2 path: .cohort/agent-eval persist-credentials: false - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-knowledge - ref: b67218f308ff0c8df6fe0cc8d8e47fa6351d0ecb # v7.0.4 + ref: c5e2ec6ffc5cd92873ce22740e8ab14a72456d59 # v7.0.7 path: .cohort/agent-knowledge persist-credentials: false diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml index 713461fe..c815f19a 100644 --- a/.github/workflows/publish.yml +++ b/.github/workflows/publish.yml @@ -135,7 +135,7 @@ jobs: uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-sdk - ref: 3645e7d0a7b7755e16fafa252db4abb7ef4058c7 # @tangle-network/agent-interface@0.40.0 + ref: fc00613985c86f917a679c9bf957e6af8cbedb7c # @tangle-network/agent-interface@0.43.0 path: .cohort/agent-sdk persist-credentials: false @@ -143,7 +143,7 @@ jobs: uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-eval - ref: 55d2faacc6ba2c92ad21b42b9b607ab16ae0fc7c # v0.140.1 + ref: f379f2aa347e416fdb28c151ba54bc93b64e4f65 # v0.142.2 path: .cohort/agent-eval persist-credentials: false @@ -151,7 +151,7 @@ jobs: uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 with: repository: tangle-network/agent-knowledge - ref: b67218f308ff0c8df6fe0cc8d8e47fa6351d0ecb # v7.0.4 + ref: c5e2ec6ffc5cd92873ce22740e8ab14a72456d59 # v7.0.7 path: .cohort/agent-knowledge persist-credentials: false diff --git a/CHANGELOG.md b/CHANGELOG.md index 1b96b3f6..cc30f34f 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,6 +2,16 @@ ## Unreleased +## 0.123.0 + +### Current shared contracts and honest CLI accounting + +- Runtime now develops and publishes against Agent Eval 0.142.2, Agent Interface 0.43.0, Agent Knowledge 7.0.7, the profile materializer 0.10.2, and Sandbox 0.17.2. +- Runtime requires Agent Interface 0.43.x so every package in this release set uses the same profile contract. +- Runtime passes the canonical Agent Interface profile directly through Sandbox now that both packages share the same contract. +- A CLI worker that reports no usage now leaves token and dollar cost unknown instead of recording a measured zero. +- The 0.122.0 agent-graph API, which reached `main` without a published tag, ships in this release. + ## 0.122.0 ### Agent graphs: loops as data, edges you can audit diff --git a/bench/CHANGELOG.md b/bench/CHANGELOG.md index 9e67c632..6cf3c219 100644 --- a/bench/CHANGELOG.md +++ b/bench/CHANGELOG.md @@ -1,5 +1,10 @@ # Changelog +## 0.7.0 + +- Consume Runtime 0.123.0 with Eval 0.142.2, Interface 0.43.0, Knowledge 7.0.7, and Sandbox 0.17.2 as one compatible dependency set. +- Release under a new minor because Interface and Sandbox are pre-1.0 dependencies moving across minor boundaries. + ## 0.6.0 - Consume Runtime 0.121.0 with Sandbox 0.16.0 so a clean install no longer requires the superseded Sandbox 0.15 line. diff --git a/bench/package.json b/bench/package.json index c6a0efa3..21c1975c 100644 --- a/bench/package.json +++ b/bench/package.json @@ -1,6 +1,6 @@ { "name": "@tangle-network/agent-bench", - "version": "0.6.0", + "version": "0.7.0", "type": "module", "description": "Benchmark adapters and execution for agent-runtime across coding, tool-use, RAG, memory, browser, and terminal tasks.", "repository": { diff --git a/docs/api/primitive-catalog.md b/docs/api/primitive-catalog.md index 1dccb0b8..58052dfd 100644 --- a/docs/api/primitive-catalog.md +++ b/docs/api/primitive-catalog.md @@ -7,7 +7,7 @@ # Primitive catalog — the never-stale anti-reinvention inventory -> **GENERATED** from `@tangle-network/agent-runtime@0.122.0` and `@tangle-network/agent-eval@0.140.1` by `scripts/gen-primitive-catalog.mjs`. Do NOT hand-edit — run `pnpm run docs:api`. This is the mechanical companion to the JUDGMENT in `canonical-api.md` (§2 decision table + §1.5 AgentProfile law): that doc says WHICH primitive to reach for and what NOT to build; this catalog proves WHAT exists. Per-symbol signatures + `file:line` live in the per-module pages under `docs/api/`. +> **GENERATED** from `@tangle-network/agent-runtime@0.123.0` and `@tangle-network/agent-eval@0.142.2` by `scripts/gen-primitive-catalog.mjs`. Do NOT hand-edit — run `pnpm run docs:api`. This is the mechanical companion to the JUDGMENT in `canonical-api.md` (§2 decision table + §1.5 AgentProfile law): that doc says WHICH primitive to reach for and what NOT to build; this catalog proves WHAT exists. Per-symbol signatures + `file:line` live in the per-module pages under `docs/api/`. ## 1. agent-runtime — own public surface diff --git a/docs/canonical-api.md b/docs/canonical-api.md index e95b4e3a..001cd83a 100644 --- a/docs/canonical-api.md +++ b/docs/canonical-api.md @@ -4,11 +4,11 @@ Generated signatures and the complete export list live in docs/api/. Run pnpm docs:freshness after editing this file. --> -> **Version 0.122.0.** +> **Version 0.123.0.** > [`docs/api/primitive-catalog.md`](./api/primitive-catalog.md) lists every export and import path. -> `agent-eval` must satisfy `>=0.140.1 <0.141.0`. -> `sandbox` must satisfy `>=0.16.0 <0.17.0`. -> Portable profile and tool-part types come from `@tangle-network/agent-interface` `>=0.40.0 <0.41.0`. +> `agent-eval` must satisfy `>=0.142.2 <0.143.0`. +> `sandbox` must satisfy `>=0.17.2 <0.18.0`. +> Portable profile and tool-part types come from `@tangle-network/agent-interface` `>=0.43.0 <0.44.0`. > > **`./kernel` is the execution kernel**: `package.json` maps it to `src/runtime/index.ts`. Everything below labelled `/kernel` lives there — the recursive atom (`Scope`/`Supervisor`), the executor registry, budget conservation, the finalizer seam, analyst wiring, and the round-synchronous loop. > diff --git a/package.json b/package.json index 00db954b..2357b07f 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "@tangle-network/agent-runtime", - "version": "0.122.0", + "version": "0.123.0", "description": "Shared task-lifecycle skeleton for agents: a recursive loop kernel for chat turns, one-shot tasks, and multi-attempt loops, with trace capture and eval-gated self-improvement. Domain behavior lives in adapters; scoring and ship-gates in @tangle-network/agent-eval.", "homepage": "https://github.com/tangle-network/agent-runtime#readme", "repository": { @@ -169,9 +169,9 @@ "license": "MIT", "packageManager": "pnpm@11.17.0", "peerDependencies": { - "@tangle-network/agent-eval": ">=0.140.1 <0.141.0", - "@tangle-network/agent-interface": ">=0.40.0 <0.41.0", - "@tangle-network/sandbox": ">=0.16.0 <0.17.0", + "@tangle-network/agent-eval": ">=0.142.2 <0.143.0", + "@tangle-network/agent-interface": ">=0.43.0 <0.44.0", + "@tangle-network/sandbox": ">=0.17.2 <0.18.0", "playwright": "^1.40.0" }, "peerDependenciesMeta": { diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml index bed59f9a..e7e27027 100644 --- a/pnpm-lock.yaml +++ b/pnpm-lock.yaml @@ -10,23 +10,23 @@ catalogs: specifier: 0.18.5 version: 0.18.5 '@tangle-network/agent-eval': - specifier: 0.140.1 - version: 0.140.1 + specifier: 0.142.2 + version: 0.142.2 '@tangle-network/agent-interface': - specifier: 0.40.0 - version: 0.40.0 + specifier: 0.43.0 + version: 0.43.0 '@tangle-network/agent-knowledge': - specifier: 7.0.4 - version: 7.0.4 + specifier: 7.0.7 + version: 7.0.7 '@tangle-network/agent-profile-materialize': - specifier: 0.9.4 - version: 0.9.4 + specifier: 0.10.2 + version: 0.10.2 '@tangle-network/agent-trace-contract': specifier: ^1.0.2 version: 1.0.2 '@tangle-network/sandbox': - specifier: 0.16.0 - version: 0.16.0 + specifier: 0.17.2 + version: 0.17.2 '@types/node': specifier: 26.1.1 version: 26.1.1 @@ -49,10 +49,10 @@ importers: dependencies: '@tangle-network/agent-knowledge': specifier: 'catalog:' - version: 7.0.4 + version: 7.0.7 '@tangle-network/agent-profile-materialize': specifier: 'catalog:' - version: 0.9.4(@tangle-network/agent-interface@0.40.0) + version: 0.10.2(@tangle-network/agent-interface@0.43.0) '@tangle-network/agent-trace-contract': specifier: 'catalog:' version: 1.0.2 @@ -68,13 +68,13 @@ importers: version: 2.5.5 '@tangle-network/agent-eval': specifier: 'catalog:' - version: 0.140.1 + version: 0.142.2 '@tangle-network/agent-interface': specifier: 'catalog:' - version: 0.40.0 + version: 0.43.0 '@tangle-network/sandbox': specifier: 'catalog:' - version: 0.16.0(viem@2.54.6(typescript@6.0.3)(zod@4.4.3)) + version: 0.17.2(viem@2.54.6(typescript@6.0.3)(zod@4.4.3)) '@types/node': specifier: 'catalog:' version: 26.1.1 @@ -119,19 +119,19 @@ importers: dependencies: '@tangle-network/agent-eval': specifier: 'catalog:' - version: 0.140.1 + version: 0.142.2 '@tangle-network/agent-interface': specifier: 'catalog:' - version: 0.40.0 + version: 0.43.0 '@tangle-network/agent-knowledge': specifier: 'catalog:' - version: 7.0.4 + version: 7.0.7 '@tangle-network/agent-runtime': specifier: workspace:* version: link:.. '@tangle-network/sandbox': specifier: 'catalog:' - version: 0.16.0(viem@2.54.6(typescript@6.0.3)(zod@4.4.3)) + version: 0.17.2(viem@2.54.6(typescript@6.0.3)(zod@4.4.3)) devDependencies: '@arethetypeswrong/cli': specifier: 'catalog:' @@ -1105,38 +1105,32 @@ packages: '@standard-schema/spec@1.1.0': resolution: {integrity: sha512-l2aFy5jALhniG5HgqrD6jXLi/rUWrKvqN/qJx6yoJsgKhblVd+iqqU4RCXavm/jPityDo5TCvKMnpjKnOriy0w==} - '@tangle-network/agent-core@0.4.27': - resolution: {integrity: sha512-gD3z0p+h8/XB2/vWQF6/RtKTLcoJOSrB85+ndYyX/3/l0pkUuVBTjm/RISiQBJyNVozgZz5iqhIV2O136keclA==} + '@tangle-network/agent-core@0.4.33': + resolution: {integrity: sha512-if3DeIo4e2c9wELJjrWOu4SsKv3WnWdDrp+kezP6JrpiQ1DXValRRlN/aTppicFHepw67OD5JQI9i351Z/hvDQ==} - '@tangle-network/agent-core@0.4.29': - resolution: {integrity: sha512-8RaZSTYU/HXZt48Mr0KfwLla0j3K6EH8H8xc0lI5qFggAMsNBic8Tk951CnFaNX680VSXGOma499zS1NgowqfA==} - - '@tangle-network/agent-eval@0.140.1': - resolution: {integrity: sha512-fuC+2haGKI93z96mcTcSumsFa4gIsuOOqA3BCeeLPQkasoSLgej3aJ6yDMim0vvbyDhtaW6jYSQ+fdGSPYnqjg==} + '@tangle-network/agent-eval@0.142.2': + resolution: {integrity: sha512-uejOBoCarTEOh520Zz5JzHkRN3YXqDZnsjEwGLF22WUnpbxzSSLUC4AuuAIOUPsvsI5kjs/tyR0lp6ad/H7P+g==} engines: {node: '>=20'} hasBin: true - '@tangle-network/agent-interface@0.38.0': - resolution: {integrity: sha512-4awwTtCeUyCAOlEqr4SKyqbj7ykvoRob6GXe9ZRk6KglTgnAU6VuEoMu1KudVvEYjmTHjBCCIHa3yL1FKgKJaA==} - - '@tangle-network/agent-interface@0.40.0': - resolution: {integrity: sha512-4A6+bb5Ub22WJsVQRNuNIr6BSsfQG4ZZT436GZ2qUcZocjJ7186MEbN7kOmYq9TQ9qeMSUh/HbyYw2nsok5b+w==} + '@tangle-network/agent-interface@0.43.0': + resolution: {integrity: sha512-t35nGOJ3sWouHoUM/5A8eUmsR+8JcuRF6VFQbT9cPzxIMUV3NsFMJgYXPGmyIj65vtvs73O6Mvijp47FAAGbuQ==} - '@tangle-network/agent-knowledge@7.0.4': - resolution: {integrity: sha512-/7FBk6OI14I/L7kc4sDmFwR7hDyndKajOjeQGdP6iSiPE/q8DldQ5UNSjJlUTeMfE3YBmLZe1/IuJEuiaqtzvg==} + '@tangle-network/agent-knowledge@7.0.7': + resolution: {integrity: sha512-h0ukEgQsRd5zUhM8ibVLaHAssBKjcFQTZny9UUP+frL+ID8JtukH9MQSGjHlkKxbeqfA2MOVU6QCU2bcyt5OBw==} engines: {node: '>=20.19.0'} hasBin: true - '@tangle-network/agent-profile-materialize@0.9.4': - resolution: {integrity: sha512-LCKfohFcjzeDz+WrPx6lXyZvn0CYyw0Y3/sBCaJ6fqj2Sgg7KwG178eg/LhlqNPz6xJzBh0g4iHxPge1kNn4nw==} + '@tangle-network/agent-profile-materialize@0.10.2': + resolution: {integrity: sha512-u3MtUy8BD5odTWu/nK0b2XuieBSZMMx1kR9izwqIoR+c8pQaIRQiENNms9qeyQFLqqm7ewC1/Qq5oTtuihs/5g==} peerDependencies: - '@tangle-network/agent-interface': '>=0.38.0 <0.41.0' + '@tangle-network/agent-interface': '>=0.38.0 <0.44.0' '@tangle-network/agent-trace-contract@1.0.2': resolution: {integrity: sha512-v7uMh56jkEp4vckevEU9xKsIatbs5dqzGPp69dFLSSXUVit0RP6VD6EANMXVlTCUk+6wVKBLHJx23XspVCEiIA==} - '@tangle-network/sandbox@0.16.0': - resolution: {integrity: sha512-4L2Xvmy/9i9udI/ISV0YuB6/f6i5OICIvkiw/cKx1lxDNbnEIkk8l2X4OReGjUkPvFSI8TnOD7Gq7LADlYdmeQ==} + '@tangle-network/sandbox@0.17.2': + resolution: {integrity: sha512-e+p/Uet2nVej3pLLCOoxeRN29t38QfWEWpcnnJsC2b0BNsQEPUg6xNlEgdeT9Eo0U6++QyR+GUrDovJ/houd2g==} peerDependencies: '@mastra/core': ^1.36.0 '@modelcontextprotocol/sdk': ^1.29.0 @@ -2799,56 +2793,46 @@ snapshots: '@standard-schema/spec@1.1.0': {} - '@tangle-network/agent-core@0.4.27': - dependencies: - '@tangle-network/agent-interface': 0.38.0 - zod: 4.4.3 - - '@tangle-network/agent-core@0.4.29': + '@tangle-network/agent-core@0.4.33': dependencies: - '@tangle-network/agent-interface': 0.40.0 + '@tangle-network/agent-interface': 0.43.0 zod: 4.4.3 - '@tangle-network/agent-eval@0.140.1': + '@tangle-network/agent-eval@0.142.2': dependencies: '@asteasolutions/zod-to-openapi': 9.1.0(zod@4.4.3) '@hono/node-server': 2.0.12(hono@4.12.32) - '@tangle-network/agent-core': 0.4.29 - '@tangle-network/agent-interface': 0.40.0 + '@tangle-network/agent-core': 0.4.33 + '@tangle-network/agent-interface': 0.43.0 + '@tangle-network/agent-trace-contract': 1.0.2 hono: 4.12.32 linear-sum-assignment: 1.0.9 re2js: 2.8.6 zod: 4.4.3 - '@tangle-network/agent-interface@0.38.0': - dependencies: - '@noble/hashes': 1.8.0 - spdx-expression-parse: 5.0.0 - zod: 4.4.3 - - '@tangle-network/agent-interface@0.40.0': + '@tangle-network/agent-interface@0.43.0': dependencies: '@noble/hashes': 1.8.0 spdx-expression-parse: 5.0.0 zod: 4.4.3 - '@tangle-network/agent-knowledge@7.0.4': + '@tangle-network/agent-knowledge@7.0.7': dependencies: - '@tangle-network/agent-eval': 0.140.1 - '@tangle-network/agent-interface': 0.40.0 + '@tangle-network/agent-eval': 0.142.2 + '@tangle-network/agent-interface': 0.43.0 proper-lockfile: 4.1.2 zod: 4.4.3 - '@tangle-network/agent-profile-materialize@0.9.4(@tangle-network/agent-interface@0.40.0)': + '@tangle-network/agent-profile-materialize@0.10.2(@tangle-network/agent-interface@0.43.0)': dependencies: - '@tangle-network/agent-interface': 0.40.0 + '@tangle-network/agent-interface': 0.43.0 '@tangle-network/agent-trace-contract@1.0.2': {} - '@tangle-network/sandbox@0.16.0(viem@2.54.6(typescript@6.0.3)(zod@4.4.3))': + '@tangle-network/sandbox@0.17.2(viem@2.54.6(typescript@6.0.3)(zod@4.4.3))': dependencies: - '@tangle-network/agent-core': 0.4.27 - '@tangle-network/agent-interface': 0.38.0 + '@tangle-network/agent-core': 0.4.33 + '@tangle-network/agent-interface': 0.43.0 zod: 4.4.3 optionalDependencies: viem: 2.54.6(typescript@6.0.3)(zod@4.4.3) diff --git a/pnpm-workspace.yaml b/pnpm-workspace.yaml index 728f450e..2d4a6e80 100644 --- a/pnpm-workspace.yaml +++ b/pnpm-workspace.yaml @@ -18,12 +18,12 @@ allowBuilds: catalog: '@arethetypeswrong/cli': 0.18.5 '@types/node': 26.1.1 - '@tangle-network/agent-eval': 0.140.1 - '@tangle-network/agent-interface': 0.40.0 - '@tangle-network/agent-knowledge': 7.0.4 - '@tangle-network/agent-profile-materialize': 0.9.4 + '@tangle-network/agent-eval': 0.142.2 + '@tangle-network/agent-interface': 0.43.0 + '@tangle-network/agent-knowledge': 7.0.7 + '@tangle-network/agent-profile-materialize': 0.10.2 '@tangle-network/agent-trace-contract': ^1.0.2 - '@tangle-network/sandbox': 0.16.0 + '@tangle-network/sandbox': 0.17.2 publint: 0.3.22 tsdown: 0.22.14 tsx: 4.23.1 diff --git a/src/runtime/environment-provider.ts b/src/runtime/environment-provider.ts index 52a79f05..4a0dc686 100644 --- a/src/runtime/environment-provider.ts +++ b/src/runtime/environment-provider.ts @@ -37,7 +37,6 @@ import type { ExecResult as SandboxExecResult, SandboxInstance, } from '@tangle-network/sandbox' -import { profileAsSandboxProfile, sandboxProfileAsProfile } from './sandbox-backend' import type { Executor, ExecutorContext, @@ -452,9 +451,7 @@ function createInputFromSandboxOptions( ...(options?.git?.ref ? { gitRef: options.git.ref } : {}), } return { - // sandboxProfileAsProfile: sandbox 0.16.0 types this profile against - // interface 0.38; it crosses as data (see sandbox-backend.ts). - ...(profile !== undefined ? { profile: sandboxProfileAsProfile(profile) } : {}), + ...(profile !== undefined ? { profile } : {}), ...(backend ? { backend } : {}), ...(Object.keys(workspace).length > 0 ? { workspace } : {}), ...(options?.resources ? { resources: options.resources as ResourceRequest } : {}), @@ -497,9 +494,7 @@ async function sandboxOptionsFromCreateInput( backend: { ...baseBackend, type: backendType, - // profileAsSandboxProfile: reverse hop of the interface 0.38/0.40 skew - // (see sandbox-backend.ts); the profile crosses as data. - profile: profileAsSandboxProfile(profile), + profile, }, } } diff --git a/src/runtime/sandbox-backend.ts b/src/runtime/sandbox-backend.ts index 3f8b262d..54856fbc 100644 --- a/src/runtime/sandbox-backend.ts +++ b/src/runtime/sandbox-backend.ts @@ -14,28 +14,6 @@ import type { CreateSandboxOptions } from '@tangle-network/sandbox' type BackendType = NonNullable['type'] type BackendOverride = NonNullable -/** The AgentProfile type as @tangle-network/sandbox declares it — compiled - * against agent-interface 0.38, not this package's 0.40. */ -export type SandboxAgentProfile = NonNullable - -/** - * Version-skew boundary: sandbox 0.16.0 types profiles against interface 0.38 - * (MCP `args`/`env`/`headers` as plain strings), while this runtime is on 0.40 - * (`AgentProfileConfigValue` objects). No published sandbox release is typed - * against 0.40 yet, and the profile crosses this boundary as data — the values - * are forwarded, not reinterpreted — so the only conversion is the nominal type - * hop. This is the ONE sanctioned cast for that skew; remove both adapters when - * sandbox releases against interface 0.40. - */ -export function profileAsSandboxProfile(profile: AgentProfile): SandboxAgentProfile { - return profile as unknown as SandboxAgentProfile -} - -/** Reverse hop of the same skew boundary — see `profileAsSandboxProfile`. */ -export function sandboxProfileAsProfile(profile: SandboxAgentProfile): AgentProfile { - return profile as unknown as AgentProfile -} - /** * Harnesses the sandbox accepts as a `backend.type`. `gemini` is a * `HarnessType` with no sandbox backend, so it is absent here and a profile @@ -110,7 +88,7 @@ export function buildBackendOptions( ...base, backend: { type: resolveBackendType(profile, overrideBackend), - profile: profileAsSandboxProfile(profile), + profile, ...(overrideBackend?.model ? { model: overrideBackend.model } : {}), ...(overrideBackend?.server ? { server: overrideBackend.server } : {}), }, diff --git a/src/testing/fixtures/agent-improvement-proposal.json b/src/testing/fixtures/agent-improvement-proposal.json index d6242803..c60b85d0 100644 --- a/src/testing/fixtures/agent-improvement-proposal.json +++ b/src/testing/fixtures/agent-improvement-proposal.json @@ -1,6 +1,6 @@ { "changedSurfaces": ["prompt"], - "digest": "sha256:c6584f37a70c0564292ae3781a52048991d15816f07adb69f96f39635208dfcd", + "digest": "sha256:0f4c0fe5e1339dd5f387766056e7797cbd9064cab15ba4c3edc922569a2685b4", "evaluation": { "decision": { "contributingChecks": [ @@ -17,7 +17,7 @@ "passed": true }, { - "name": "candidate-task-pass", + "name": "no-task-regression", "passed": true }, { @@ -4810,7 +4810,7 @@ ], "metadata": { "fixture": "agent-improvement-proposal", - "runtimeVersion": "0.122.0" + "runtimeVersion": "0.123.0" }, "objectives": [ { @@ -4921,8 +4921,8 @@ "baselineContentHash": "sha256:5c21ee53e513fc604cb09754e21c392b24a424da0ef37dbf8f1ee4a8a0b08f09", "candidateContentHash": "sha256:60fcbb1c728194bd51d7d19cb732d1c3f1881dce7e0a6266b41c8b98cfd65693", "kind": "agent-eval-loop", - "recordDigest": "sha256:a9eb16d91ec7a0b4b8b818bb553afddf02a606f7309845265226d71594f6538f", - "runId": "agent-runtime-0.122.0-proposal-fixture", + "recordDigest": "sha256:ac7412b9a1e554d1b89aaaa80fcea7901cf5d258e8ae8c3777f70701c8b86d6f", + "runId": "agent-runtime-0.123.0-proposal-fixture", "schema": "agent-candidate-experiment" } }, @@ -4949,5 +4949,5 @@ ], "kind": "agent-improvement-proposal", "proposedAt": "2026-07-10T01:00:00.000Z", - "runId": "agent-runtime-0.122.0-proposal-fixture" + "runId": "agent-runtime-0.123.0-proposal-fixture" } diff --git a/src/testing/fixtures/agent-profile-improvement-proposal.json b/src/testing/fixtures/agent-profile-improvement-proposal.json index 05151d63..75ac9326 100644 --- a/src/testing/fixtures/agent-profile-improvement-proposal.json +++ b/src/testing/fixtures/agent-profile-improvement-proposal.json @@ -1,6 +1,6 @@ { "changedSurfaces": ["prompt", "skills"], - "digest": "sha256:a497966c3f7b0e73da1a9602bf673e3f05948a9a572982cbdcb61242fdac169f", + "digest": "sha256:58231ee3d872020cce4c8f1dc8179b11e598c5daf1566c643677eb5dc231e453", "evaluation": { "decision": { "contributingChecks": [ @@ -17,7 +17,7 @@ "passed": true }, { - "name": "candidate-task-pass", + "name": "no-task-regression", "passed": true }, { @@ -1715,7 +1715,7 @@ ], "metadata": { "fixture": "agent-profile-improvement-proposal", - "runtimeVersion": "0.122.0" + "runtimeVersion": "0.123.0" }, "objectives": [ { @@ -1826,7 +1826,7 @@ "baselineContentHash": "sha256:21c495a37c418c10bde64fbaa188beddeed31f1f051ea60a6a6582a9ee0db704", "candidateContentHash": "sha256:103f77bc8481601eef1ad5fe6ba84a40dffabc3a44f421f8c8559121edab84e9", "kind": "agent-eval-loop", - "recordDigest": "sha256:6c8ab52b1abc1c996614751b2a59b8a86593c5783d74361d8cef87ed73fe7a26", + "recordDigest": "sha256:f3bbd6c1b07feacc0679a0c38b9c1fbf2eea5fa8c35ddaf950a4960a2e7e3d0f", "runId": "profile-improvement-1", "schema": "agent-profile-improvement-experiment" }