feat(workflows): add beval behavioral evaluation workflow for dt-coach agent

.cspell.json

@@ -28,6 +28,7 @@
     "docs/docusaurus/coverage/**",
     "dependency-pinning-artifacts/**",
     "evals/results/**",
+    "evals/beval/**/results/**",
     "vally-results/**",
     "evals/agent-behavior/eval.yaml"
   ],
@@ -68,73 +69,50 @@
     "general-technical"
   ],
   "words": [
-    "ˈpræksɪs",
     "activedescendant",
     "agentic",
     "aoda",
     "ASEC",
     "atheris",
-    "cursored",
-    "networkidle",
-    "behaviour",
-    "beval",
-    "behavioural",
-    "brainwriting",
-    "clusterfuzzlite",
-    "collab",
-    "dogfooding",
-    "easyops",
-    "evals",
-    "fetchone",
-    "figjam",
-    "graphify",
-    "graphifyy",
-    "hideable",
-    "idor",
-    "learning",
-    "normaliz",
-    "packageable",
-    "prval",
-    "skillmd",
-    "smol",
-    "subcat",
-    "uncited",
-    "vally",
-    "vyta",
-    "whiteboarding",
-    "ˈpræksɪs",
-    "πρᾶξις",
-    "agentic",
-    "ystatement",
-    "underspecified",
-    "sssc",
-    "SSSC",
-    "SLSA",
-    "Sigstore",
-    "cosign",
-    "scorecard",
-    "watchlist",
     "ation",
     "BABOK",
     "backlinked",
+    "behaviour",
+    "behavioural",
+    "beval",
     "bmve",
+    "brainwriting",
     "calledprocesserror",
     "capabilit",
     "clarif",
+    "clusterfuzzlite",
     "cmdletbinding",
+    "collab",
     "consolidat",
+    "cosign",
+    "cursored",
     "dataclass",
     "deeplink",
     "desirab",
+    "dogfooding",
+    "easyops",
     "emulat",
     "erroractionpreference",
+    "evals",
     "facilitat",
     "feasib",
+    "fetchone",
+    "figjam",
     "filenotfounderror",
+    "graphify",
+    "graphifyy",
     "Graphviz",
+    "hideable",
     "hypothes",
+    "idor",
     "IIBA",
     "ISTQB",
+    "learning",
     "Linkified",
     "loadconfig",
     "mdspell",
@@ -143,28 +121,50 @@
     "modulemember",
     "monkeypatch",
     "navigations",
+    "networkidle",
+    "normaliz",
     "observ",
     "octocat",
     "onclick",
     "optimi",
+    "packageable",
+    "parseable",
     "pascalcase",
     "polic",
     "profanit",
+    "prval",
     "pylint",
     "refus",
     "scal",
+    "scorecard",
+    "Sigstore",
+    "skillmd",
+    "SLSA",
+    "smol",
     "SPOF",
+    "SSSC",
+    "sssc",
     "stimul",
+    "subcat",
     "subdirs",
     "terroris",
     "thiserror",
     "trustworthi",
+    "uncited",
     "underspecified",
+    "vally",
     "viab",
     "violen",
     "vulnerabilit",
+    "vyta",
+    "watchlist",
+    "whiteboarding",
+    "wireframes",
     "workiq",
-    "WSJF"
+    "WSJF",
+    "ystatement",
+    "πρᾶξις",
+    "ˈpræksɪs"
   ],
   "reporters": [
     "default",

.github/dependabot.yml

@@ -5,6 +5,7 @@ updates:
   directories:
   - /
   - /docs/docusaurus
+  - /evals/beval
   groups:
     npm-dependencies:
       patterns:

.github/workflows/beval.yml

@@ -0,0 +1,87 @@
+name: Behavioral Evaluation (beval)
+
+on:
+  workflow_call:
+    secrets:
+      COPILOT_TOKEN:
+        required: true
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: false
+
+jobs:
+  evaluate:
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+
+    env:
+      AGENT_REPO_ROOT: ${{ github.workspace }}
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Set up Python
+        uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+
+      - name: Install GitHub Copilot CLI
+        run: |
+          npm ci --prefix evals/beval
+          echo "${{ github.workspace }}/evals/beval/node_modules/.bin" >> "$GITHUB_PATH"
+
+      - name: Install beval
+        # beval is hosted under a personal account (vyta) while an org-owned
+        # home is evaluated. The install is pinned to a specific commit SHA to
+        # mitigate supply-chain risk in the interim.
+        run: pip install --no-cache-dir "beval[all] @ git+https://github.com/vyta/beval.git@d9f46c24f03b0b806d928a8a8ce2fc66d8e470fb#subdirectory=python"
+
+      - name: Start agent (TCP)
+        env:
+          COPILOT_GITHUB_TOKEN: ${{ secrets.COPILOT_TOKEN }}
+        run: |
+          copilot --acp --port 3000 &
+          for i in $(seq 1 30); do
+            nc -z 127.0.0.1 3000 && break
+            echo "Waiting for agent to start ($i)..."
+            sleep 2
+          done
+          nc -z 127.0.0.1 3000 || { echo "Agent failed to start"; exit 1; }
+
+      - name: Start judge (TCP)
+        env:
+          COPILOT_GITHUB_TOKEN: ${{ secrets.COPILOT_TOKEN }}
+        run: |
+          copilot --acp --port 3001 &
+          for i in $(seq 1 30); do
+            nc -z 127.0.0.1 3001 && break
+            echo "Waiting for judge to start ($i)..."
+            sleep 2
+          done
+          nc -z 127.0.0.1 3001 || { echo "Judge failed to start"; exit 1; }
+
+      - name: Run evaluations
+        run: |
+          beval \
+            -c evals/beval/dt-coach/eval.config.yaml \
+            run \
+            --cases evals/beval/dt-coach/cases/ \
+            --agent evals/beval/dt-coach/agent.yaml \
+            -m validation \
+            -o evals/beval/dt-coach/results/results.json
+
+      - name: Upload results
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        if: always()
+        with:
+          name: beval-results-${{ github.run_id }}
+          path: evals/beval/dt-coach/results/
+          retention-days: 30

.github/workflows/dependency-review.yml

@@ -65,6 +65,11 @@ jobs:
             WTFPL, LicenseRef-scancode-unicode
           # Packages with compound SPDX expressions containing GPL or MPL
           # from bundled code; distributed licenses are permissive.
+          # @github/copilot and its platform-specific binaries (including
+          # @github/copilot-linuxmusl-arm64 and @github/copilot-linuxmusl-x64
+          # added in v1.0.63) use a non-SPDX proprietary license
+          # (LicenseRef-bad-see-license-in-license.md); it is GitHub's own
+          # CLI toolchain, deliberately used in beval.yml.
           # pkg:npm/hve-core is the private root package (never published to npm).
           # pkg:pypi/certifi uses MPL-2.0 (Mozilla CA bundle).
           # pkg:pypi/charset-normalizer is MIT licensed but some older wheel
@@ -119,6 +124,15 @@ jobs:
             pkg:pypi/regex,
             pkg:npm/dompurify,
             pkg:npm/lunr-languages,
+            pkg:npm/%40github/copilot,
+            pkg:npm/%40github/copilot-darwin-arm64,
+            pkg:npm/%40github/copilot-darwin-x64,
+            pkg:npm/%40github/copilot-linux-arm64,
+            pkg:npm/%40github/copilot-linux-x64,
+            pkg:npm/%40github/copilot-win32-arm64,
+            pkg:npm/%40github/copilot-win32-x64,
+            pkg:npm/%40github/copilot-linuxmusl-arm64,
+            pkg:npm/%40github/copilot-linuxmusl-x64,
             pkg:npm/lightningcss,
             pkg:npm/lightningcss-android-arm64,
             pkg:npm/lightningcss-darwin-arm64,

.gitignore

@@ -492,3 +492,6 @@ dependency-pinning-artifacts/
 docs/docusaurus/build/
 docs/docusaurus/.docusaurus/
 docs/docusaurus/node_modules/
+
+# Beval evaluation results
+evals/beval/**/results/

evals/beval/dt-coach/agent.yaml

@@ -0,0 +1,20 @@
+name: dt-coach
+description: >
+  Design Thinking Coach — a conversational coaching agent that guides teams
+  through the 9 Design Thinking for HVE methods using a Think/Speak/Empower
+  philosophy.
+protocol: acp
+connection:
+  transport: tcp
+  host: ${AGENT_HOST:-127.0.0.1}
+  port: ${AGENT_PORT:-3000}
+  cwd: ${AGENT_REPO_ROOT:-.}
+  model: ${AGENT_MODEL:-claude-opus-4.6-1m}
+init_prompt: "Launch .github/agents/design-thinking/dt-coach.agent.md"
+timeout: 120
+retry:
+  max_attempts: 2
+  backoff: 5.0
+metadata:
+  domain: design-thinking
+  version: "0.1"