feat(ocr): add built-in first-use capability

.github/workflows/release.yml

@@ -42,7 +42,7 @@ jobs:
             test "${{ inputs.release_tag }}" = "v${version}"
           fi
       - run: cargo fmt --all -- --check
-      - run: cargo test --workspace --all-features --locked
+      - run: cargo test --workspace --all-features --locked -- --test-threads=1
       - run: cargo clippy --workspace --all-targets --all-features --locked -- -D warnings
 
   binaries:
@@ -84,12 +84,13 @@ jobs:
           version="${{ needs.validate.outputs.version }}"
           archive="a3s-use-${version}-${{ matrix.name }}.tar.gz"
           stage="${RUNNER_TEMP}/a3s-use-${version}-${{ matrix.name }}"
-          install -d "${stage}/skills" "${stage}/skill-data" "${stage}/office-skills" "${stage}/dashboard"
+          install -d "${stage}/skills" "${stage}/skill-data" "${stage}/office-skills" "${stage}/ocr-skills" "${stage}/dashboard"
           install -m 0755 "target/${{ matrix.target }}/release/a3s-use" "${stage}/a3s-use"
           install -m 0755 "target/${{ matrix.target }}/release/a3s-use-browser-driver" "${stage}/a3s-use-browser-driver"
           cp -R crates/browser-driver/skills/. "${stage}/skills/"
           cp -R crates/browser-driver/skill-data/. "${stage}/skill-data/"
           cp -R crates/office/skills/. "${stage}/office-skills/"
+          cp -R crates/ocr/skills/. "${stage}/ocr-skills/"
           cp -R crates/browser-driver/dashboard/out/. "${stage}/dashboard/"
           install -m 0644 LICENSE README.md THIRD_PARTY_NOTICES.md "${stage}/"
           install -m 0644 crates/browser-driver/LICENSE-APACHE-2.0 "${stage}/LICENSE-APACHE-2.0"
@@ -110,6 +111,7 @@ jobs:
           Copy-Item -Recurse "crates/browser-driver/skills" "$stage/skills"
           Copy-Item -Recurse "crates/browser-driver/skill-data" "$stage/skill-data"
           Copy-Item -Recurse "crates/office/skills" "$stage/office-skills"
+          Copy-Item -Recurse "crates/ocr/skills" "$stage/ocr-skills"
           Copy-Item -Recurse "crates/browser-driver/dashboard/out" "$stage/dashboard"
           Copy-Item LICENSE,README.md,THIRD_PARTY_NOTICES.md $stage
           Copy-Item crates/browser-driver/LICENSE-APACHE-2.0 "$stage/LICENSE-APACHE-2.0"
@@ -128,6 +130,7 @@ jobs:
           test -x "${install_root}/a3s-use-browser-driver"
           test -f "${install_root}/skill-data/core/SKILL.md"
           test -f "${install_root}/office-skills/a3s-use-office/SKILL.md"
+          test -f "${install_root}/ocr-skills/a3s-use-ocr/SKILL.md"
           test -f "${install_root}/dashboard/index.html"
           test -f "${install_root}/LICENSE-APACHE-2.0"
           test -f "${install_root}/UPSTREAM.md"
@@ -143,6 +146,29 @@ jobs:
               "agentcore", "core", "dogfood", "electron", "slack", "vercel-sandbox"
           }
           PY
+          "${install_root}/a3s-use" ocr doctor --json > "${RUNNER_TEMP}/ocr-doctor.json"
+          python3 - "${RUNNER_TEMP}/ocr-doctor.json" <<'PY'
+          import json, pathlib, sys
+          value = json.loads(pathlib.Path(sys.argv[1]).read_text())
+          assert value["ok"] is True
+          assert value["data"]["readiness"] in {"ready", "missing", "broken", "unknown"}
+          PY
+          "${install_root}/a3s-use" capability snapshot --json > "${RUNNER_TEMP}/capabilities.json"
+          python3 - "${RUNNER_TEMP}/capabilities.json" "${install_root}" <<'PY'
+          import json, pathlib, sys
+          value = json.loads(pathlib.Path(sys.argv[1]).read_text())
+          root = pathlib.Path(sys.argv[2]).resolve()
+          ocr = next(
+              capability
+              for capability in value["data"]["registry"]["capabilities"]
+              if capability["id"] == "use/ocr"
+          )
+          assert ocr["mcp"]["target"] == "ocr-native"
+          assert len(ocr["skills"]) == 1
+          assert pathlib.Path(ocr["skills"][0]["path"]).resolve() == (
+              root / "ocr-skills" / "a3s-use-ocr" / "SKILL.md"
+          )
+          PY
           A3S_OFFICECLI_EXECUTABLE="${install_root}/must-not-be-invoked" \
             "${install_root}/a3s-use" office skills list --json > "${RUNNER_TEMP}/office-skills.json"
           python3 - "${RUNNER_TEMP}/office-skills.json" <<'PY'
@@ -179,6 +205,7 @@ jobs:
             "$root/a3s-use-browser-driver.exe",
             "$root/skill-data/core/SKILL.md",
             "$root/office-skills/a3s-use-office/SKILL.md",
+            "$root/ocr-skills/a3s-use-ocr/SKILL.md",
             "$root/dashboard/index.html",
             "$root/LICENSE-APACHE-2.0",
             "$root/UPSTREAM.md",
@@ -194,6 +221,19 @@ jobs:
           if (-not $officeSkills.ok -or $officeSkills.data.Count -ne 1 -or $officeSkills.data[0].name -ne "a3s-use-office") {
             throw "Packaged Office Skill smoke failed"
           }
+          $ocr = (& "$root/a3s-use.exe" ocr doctor --json | ConvertFrom-Json)
+          if (-not $ocr.ok -or -not $ocr.data.readiness) { throw "Built-in OCR doctor smoke failed" }
+          $capabilities = (& "$root/a3s-use.exe" capability snapshot --json | ConvertFrom-Json)
+          $ocrCapability = $capabilities.data.registry.capabilities |
+            Where-Object id -eq "use/ocr"
+          if ($ocrCapability.mcp.target -ne "ocr-native" -or $ocrCapability.skills.Count -ne 1) {
+            throw "Built-in OCR capability projection smoke failed"
+          }
+          $expectedOcrSkill = (Resolve-Path "$root/ocr-skills/a3s-use-ocr/SKILL.md").Path
+          $actualOcrSkill = (Resolve-Path $ocrCapability.skills[0].path).Path
+          if (-not [StringComparer]::OrdinalIgnoreCase.Equals($actualOcrSkill, $expectedOcrSkill)) {
+            throw "Built-in OCR Skill was not projected from the installed release"
+          }
           $requests = @(
             '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"release-smoke","version":"1"}}}',
             '{"jsonrpc":"2.0","method":"notifications/initialized","params":{}}',
@@ -223,7 +263,7 @@ jobs:
           ref: ${{ github.event_name == 'workflow_dispatch' && inputs.release_tag || github.ref }}
       - uses: dtolnay/rust-toolchain@stable
       - uses: Swatinem/rust-cache@v2
-      - name: Publish Core then Browser
+      - name: Publish Core, OCR, then Browser
         env:
           CARGO_REGISTRY_TOKEN: ${{ secrets.CARGO_TOKEN }}
           VERSION: ${{ needs.validate.outputs.version }}
@@ -262,6 +302,8 @@ jobs:
 
           publish_once a3s-use-core
           wait_until_visible a3s-use-core
+          publish_once a3s-use-ocr
+          wait_until_visible a3s-use-ocr
           publish_once a3s-use-browser
 
   release:

Cargo.toml

@@ -5,6 +5,7 @@ members = [
     "crates/browser-driver",
     "crates/office",
     "crates/extension",
+    "crates/ocr",
 ]
 resolver = "2"
 
@@ -48,7 +49,7 @@ license.workspace = true
 repository.workspace = true
 authors.workspace = true
 rust-version = "1.85"
-description = "Typed Browser, Office, and external application capabilities for A3S"
+description = "Typed Browser, Office, OCR, and external application capabilities for A3S"
 
 [lib]
 name = "a3s_use"
@@ -59,14 +60,15 @@ name = "a3s-use"
 path = "src/main.rs"
 
 [features]
-default = ["browser", "office", "extensions", "mcp"]
+default = ["browser", "office", "ocr", "extensions", "mcp"]
 browser = ["dep:a3s-use-browser"]
 office = [
     "dep:a3s-use-office",
     "dep:axum",
     "dep:futures-util",
     "dep:getrandom",
 ]
+ocr = ["dep:a3s-use-ocr"]
 extensions = ["dep:a3s-use-extension"]
 mcp = [
     "dep:axum",
@@ -84,6 +86,7 @@ lightpanda = ["browser", "a3s-use-browser/lightpanda"]
 a3s-use-core = { version = "0.1.1", path = "crates/core" }
 a3s-use-browser = { version = "0.1.1", path = "crates/browser", optional = true }
 a3s-use-office = { version = "0.1.1", path = "crates/office", optional = true }
+a3s-use-ocr = { version = "0.1.1", path = "crates/ocr", optional = true }
 a3s-use-extension = { version = "0.1.1", path = "crates/extension", optional = true }
 anyhow.workspace = true
 axum = { workspace = true, optional = true }