diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 235b553..bbed27f 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -43,6 +43,11 @@ jobs:
       - name: Verify packed package lifecycle
         run: pnpm verify:package
 
+      - name: Verify packed package with Eval 0.174
+        run: node scripts/verify-package.mjs
+        env:
+          AGENT_KNOWLEDGE_EVAL_VERSION: 0.174.0
+
   official-optimizers:
     runs-on: ubuntu-latest
     timeout-minutes: 20
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 5069543..6d203ae 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,10 @@
 # Changelog
 
+## 14.0.3 — 2026-09-07
+
+The Eval peer range admits both 0.174 and 0.175, restoring compatibility with Eval 0.174 consumers.
+Runtime integration examples use the maintained package root import.
+
 ## 14.0.2 — 2026-09-07
 
 ### Changed
diff --git a/README.md b/README.md
index f2191c0..9a1a3b3 100644
--- a/README.md
+++ b/README.md
@@ -4,7 +4,7 @@ Build, search, test, and improve source-backed knowledge bases in TypeScript.
 
 The package manages source records, Markdown knowledge, indexes, retrieval tests, candidate workspaces, and memory adapters.
 It does not browse, call models, or run agents.
-Supply application callbacks for those decisions, or use `@tangle-network/agent-runtime/knowledge` to run them with agents.
+Supply application callbacks for those decisions, or use `@tangle-network/agent-runtime` to run them with agents.
 
 ## Install
 
@@ -29,7 +29,7 @@ Requires Node.js 20.19 or later.
 | Compare providers or optimize memory configuration | `AgentMemoryAdapter`, `runAgentMemoryImprovement` | `/memory` |
 | Read from external authorities | `KnowledgeSource` and source adapters | `/sources` |
 | Run retrieval, answer, KB, or memory benchmark cases | `runKnowledgeBenchmarkSuite` | `/benchmarks` |
-| Use live research or coding agents | `runKnowledgeImprovementJob` | `@tangle-network/agent-runtime/knowledge` |
+| Use live research or coding agents | `runKnowledgeImprovementJob` | `@tangle-network/agent-runtime` |
 
 ## Create and search a knowledge base
 
diff --git a/package.json b/package.json
index bb5262c..cdee568 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-knowledge",
-  "version": "14.0.2",
+  "version": "14.0.3",
   "description": "Build, search, evaluate, and improve source-backed knowledge bases.",
   "homepage": "https://github.com/tangle-network/agent-knowledge#readme",
   "repository": {
@@ -84,7 +84,7 @@
     "zod": "4.5.4"
   },
   "peerDependencies": {
-    "@tangle-network/agent-eval": ">=0.175.0 <0.176.0",
+    "@tangle-network/agent-eval": ">=0.174.0 <0.176.0",
     "@tangle-network/agent-interface": "^2.0.0"
   },
   "devDependencies": {
diff --git a/scripts/verify-official-optimizers.mjs b/scripts/verify-official-optimizers.mjs
index 4a0071c..335184f 100644
--- a/scripts/verify-official-optimizers.mjs
+++ b/scripts/verify-official-optimizers.mjs
@@ -18,7 +18,7 @@ const agentEvalVersion = sourcePackage.devDependencies?.['@tangle-network/agent-
 if (!/^\d+\.\d+\.\d+$/.test(agentEvalVersion)) {
   throw new Error('@tangle-network/agent-eval must have one exact development pin')
 }
-const expectedEvalPeerRange = expectedPeerRange(agentEvalVersion)
+const expectedEvalPeerRange = '>=0.174.0 <0.176.0'
 if (sourcePackage.peerDependencies?.['@tangle-network/agent-eval'] !== expectedEvalPeerRange) {
   throw new Error(
     `@tangle-network/agent-eval peer range must be ${expectedEvalPeerRange} to match the development pin`,
diff --git a/scripts/verify-package.mjs b/scripts/verify-package.mjs
index 4e2629c..301626b 100644
--- a/scripts/verify-package.mjs
+++ b/scripts/verify-package.mjs
@@ -66,8 +66,12 @@ const agentCorePackage = '@tangle-network/agent-core'
 const agentInterfacePackage = '@tangle-network/agent-interface'
 const repoRoot = resolve(dirname(fileURLToPath(import.meta.url)), '..')
 const sourcePackage = JSON.parse(readFileSync(join(repoRoot, 'package.json'), 'utf8'))
-const agentEvalVersion = exactDevelopmentPin(sourcePackage, agentEvalPackage)
-const agentEvalPeerRange = expectedPeerRange(agentEvalVersion)
+const agentEvalVersion =
+  process.env.AGENT_KNOWLEDGE_EVAL_VERSION ?? exactDevelopmentPin(sourcePackage, agentEvalPackage)
+if (!['0.174.0', '0.175.0'].includes(agentEvalVersion)) {
+  throw new Error(`unsupported Eval compatibility test version: ${agentEvalVersion}`)
+}
+const agentEvalPeerRange = '>=0.174.0 <0.176.0'
 const agentInterfaceVersion = exactDevelopmentPin(sourcePackage, agentInterfacePackage)
 const agentInterfacePeerRange = expectedPeerRange(agentInterfaceVersion)
 const zodVersion = exactVersion(sourcePackage.dependencies?.zod, 'zod runtime dependency')
diff --git a/skills/build-with-agent-knowledge/SKILL.md b/skills/build-with-agent-knowledge/SKILL.md
index 3ab6a44..5d40014 100644
--- a/skills/build-with-agent-knowledge/SKILL.md
+++ b/skills/build-with-agent-knowledge/SKILL.md
@@ -46,7 +46,7 @@ The agent's current answer is not a gold answer.
 It does not own model choice, prompts, browsing, agent scheduling, product authorization, or product storage transactions.
 
 Supply callbacks for research, retrieval, answer generation, and scoring.
-Use `@tangle-network/agent-runtime/knowledge` when those callbacks should run agents.
+Use `@tangle-network/agent-runtime` when those callbacks should run agents.
 Use existing vector, graph, search, and memory systems through adapters instead of rebuilding their databases here.
 
 ## Build The Smallest Complete Path