@sanity/ailf 3.4.1 → 3.5.1

package/bin/ailf.js

@@ -33,9 +33,24 @@ const callerCwd = process.cwd()
 // ---------------------------------------------------------------------------
 if (existsSync(tsSrc)) {
   try {
+    // Enable the `ailf-source` export condition so @sanity/ailf-shared and
+    // @sanity/ailf-core resolve to their `src/index.ts` entrypoints rather
+    // than whatever happens to be in their `dist/` directories. Without
+    // this, running `ailf …` against a freshly pulled monorepo (or any
+    // workspace with a stale dist) fails at import time whenever the
+    // source introduces a new export that the dist hasn't caught up with.
+    const existingNodeOptions = process.env.NODE_OPTIONS ?? ""
+    const conditionFlag = "--conditions=ailf-source"
+    const nodeOptions = existingNodeOptions.includes(conditionFlag)
+      ? existingNodeOptions
+      : `${existingNodeOptions} ${conditionFlag}`.trim()
     execFileSync("npx", ["tsx", tsSrc, ...args], {
       cwd: ROOT,
-      env: { ...process.env, AILF_CALLER_CWD: callerCwd },
+      env: {
+        ...process.env,
+        AILF_CALLER_CWD: callerCwd,
+        NODE_OPTIONS: nodeOptions,
+      },
       stdio: "inherit",
     })
     process.exit(0)

package/config/airbyte/ai_literacy_framework.connector.yaml

@@ -87,6 +87,23 @@ definitions:
               summary.overall.avgInfrastructureEfficiency,
                 "promptfoo_url": provenance.promptfooUrl,
                 "promptfoo_urls": provenance.promptfooUrls[] { mode, url },
+                "classification": provenance.classification,
+                "owner_team": provenance.owner.team,
+                "owner_individual": provenance.owner.individual,
+                "executor_type": provenance.executor.type,
+                "executor_name": provenance.executor.name,
+                "executor_surface": provenance.executor.surface,
+                "executor_github_actor": provenance.executor.githubActor,
+                "purpose": provenance.purpose,
+                "labels": provenance.labels,
+                "lineage_rerun_of": provenance.lineage.rerunOf,
+                "lineage_compared_against": provenance.lineage.comparedAgainst,
+                "lineage_parent_job_id": provenance.lineage.parentJobId,
+                "tool_ailf_version": provenance.tool.ailfVersion,
+                "tool_node_version": provenance.tool.nodeVersion,
+                "host_platform": provenance.host.platform,
+                "host_arch": provenance.host.arch,
+                "host_ci": provenance.host.ci,
                 _createdAt
               }
         record_selector:
@@ -464,6 +481,103 @@ schemas:
             url:
               type: string
               description: Promptfoo share URL for this mode
+      # ----------------------------------------------------------------
+      # D0037 — run classification, ownership, executor, reproducibility
+      # ----------------------------------------------------------------
+      classification:
+        type:
+          - string
+          - "null"
+        description:
+          "Run classification (D0037): official | ad-hoc | experimental | test |
+          external. Orthogonal to trigger_type."
+      owner_team:
+        type:
+          - string
+          - "null"
+        description: Team slug this run is attributable to (free-form).
+      owner_individual:
+        type:
+          - string
+          - "null"
+        description: Individual (e.g., GH actor) this run is attributable to.
+      executor_type:
+        type:
+          - string
+          - "null"
+        description: '"user" | "system" — who/what actually invoked the run.'
+      executor_name:
+        type:
+          - string
+          - "null"
+        description:
+          For system executors the system name (e.g., "github-actions"); for
+          user executors the resolved user name.
+      executor_surface:
+        type:
+          - string
+          - "null"
+        description:
+          For user executors — origin surface ("cli" | "studio" | "api").
+      executor_github_actor:
+        type:
+          - string
+          - "null"
+        description: GitHub actor when the user invoked via a GH surface.
+      purpose:
+        type:
+          - string
+          - "null"
+        description: Human-authored "why I ran this" (AILF_PURPOSE / --purpose).
+      labels:
+        type:
+          - array
+          - "null"
+        items:
+          type: string
+        description:
+          Free-form searchable tags (release IDs, regression hunts,
+          experiments).
+      lineage_rerun_of:
+        type:
+          - string
+          - "null"
+        description: Prior RunId this run re-executes.
+      lineage_compared_against:
+        type:
+          - string
+          - "null"
+        description: Sibling RunId this run is intentionally compared against.
+      lineage_parent_job_id:
+        type:
+          - string
+          - "null"
+        description: API-gateway job ID that dispatched this run.
+      tool_ailf_version:
+        type:
+          - string
+          - "null"
+        description: "@sanity/ailf package version that produced this run."
+      tool_node_version:
+        type:
+          - string
+          - "null"
+        description: Node runtime version.
+      host_platform:
+        type:
+          - string
+          - "null"
+        description: os.platform() — darwin | linux | win32.
+      host_arch:
+        type:
+          - string
+          - "null"
+        description: os.arch() — x64 | arm64.
+      host_ci:
+        type:
+          - string
+          - "null"
+        description: CI provider when running under one (e.g., github-actions).
       _createdAt:
         type:
           - string

package/config/bigquery/README.md

@@ -22,10 +22,13 @@ BigQuery views (this directory)
 
 ## Files
 
-| File                    | Purpose                                                                         |
-| ----------------------- | ------------------------------------------------------------------------------- |
-| `views/area_scores.sql` | Flattens nested `model_scores` array into one row per area per model per report |
-| `views/reports.sql`     | Clean passthrough view with correct types and column ordering                   |
+| File                             | Purpose                                                                                                             |
+| -------------------------------- | ------------------------------------------------------------------------------------------------------------------- |
+| `views/area_scores.sql`          | Flattens nested `model_scores` array into one row per area per model per report                                     |
+| `views/reports.sql`              | Clean passthrough view with correct types and column ordering                                                       |
+| `views/official_runs.sql`        | Canonical trend series (D0037): `classification='official' AND trigger_type='scheduled' AND owner_team='core-docs'` |
+| `views/official_area_scores.sql` | `area_scores` joined to `official_runs` — inherits the official-run predicate for area-level dashboards             |
+| `views/team_runs_template.sql`   | Recipe/template for instantiating per-team filtered views                                                           |
 
 ## Setup
 
@@ -36,22 +39,55 @@ from `docs/design-docs/report-store/bigquery.md`.
 ### 1. Create the raw dataset (Airbyte writes here)
 
 ```bash
-bq mk --dataset data-platform-302218:ailf_raw
+bq --project_id=data-platform-302218 --location=EU mk --dataset ailf_raw
 ```
 
 ### 2. Create the analytics dataset (views live here)
 
 ```bash
-bq mk --dataset data-platform-302218:ailf
+bq --project_id=data-platform-302218 --location=EU mk --dataset ailf
 ```
 
 ### 3. Create the views
 
+**Important ordering (learned 2026-04-23):** Airbyte must be redeployed with the
+current manifest **before** you run these view SQLs. Each view binds to specific
+columns on `ailf_raw.reports`; if the raw table is missing columns the Airbyte
+projection expects, the `CREATE VIEW` statement fails with
+`Unrecognized name: <column>`.
+
+If your Airbyte destination has **schema evolution enabled** ("Propagate column
+changes" in the UI), new columns appear automatically on the next incremental
+sync. If not, flip it on, trigger a resync, and confirm the expected columns
+exist before creating views:
+
+```bash
+bq --project_id=data-platform-302218 --location=EU query --use_legacy_sql=false \
+  "SELECT column_name FROM ailf_raw.INFORMATION_SCHEMA.COLUMNS WHERE table_name = 'reports' ORDER BY column_name"
+```
+
+If propagation is disabled and you can't flip it quickly, manually
+`ALTER TABLE ailf_raw.reports ADD COLUMN IF NOT EXISTS …` for each missing
+column as a stop-gap. Values will be `NULL` until Airbyte writes to them on the
+next sync.
+
+Once the raw table has the expected columns:
+
 ```bash
-bq query --use_legacy_sql=false < views/reports.sql
-bq query --use_legacy_sql=false < views/area_scores.sql
+cd packages/eval/config/bigquery
+bq --project_id=data-platform-302218 --location=EU query --use_legacy_sql=false < views/reports.sql
+bq --project_id=data-platform-302218 --location=EU query --use_legacy_sql=false < views/area_scores.sql
+bq --project_id=data-platform-302218 --location=EU query --use_legacy_sql=false < views/official_runs.sql
+bq --project_id=data-platform-302218 --location=EU query --use_legacy_sql=false < views/official_area_scores.sql
+# per-team views are optional — copy views/team_runs_template.sql,
+# fill in the slug, and run.
 ```
 
+> `--project_id` / `--location=EU` are required because `bq` needs an explicit
+> billing project and the `ailf*` datasets live in the EU multi-region. If you
+> run `bq query` from this repo regularly, consider setting the default with
+> `gcloud config set project data-platform-302218`.
+
 ## Naming conventions
 
 - **`ailf_raw.*`** — raw Airbyte-loaded tables (nested JSON, Airbyte metadata

package/config/bigquery/views/official_area_scores.sql

@@ -0,0 +1,20 @@
+-- ailf.official_area_scores — per-area scores restricted to official runs
+--
+-- Joins ailf.area_scores (UNNESTed per-model per-area scores) to
+-- ailf.official_runs on report_id so area-level trend dashboards
+-- inherit the D0037 official-run predicate without re-declaring it.
+--
+-- Source: ailf.area_scores, ailf.official_runs
+-- Target: ailf.official_area_scores (this view)
+--
+-- Usage:
+--   bq query --use_legacy_sql=false < views/official_area_scores.sql
+--
+-- @see views/official_runs.sql
+-- @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md
+
+CREATE OR REPLACE VIEW `data-platform-302218.ailf.official_area_scores` AS
+SELECT a.*
+FROM   `data-platform-302218.ailf.area_scores` AS a
+INNER JOIN `data-platform-302218.ailf.official_runs` AS r
+       USING (report_id);

package/config/bigquery/views/official_runs.sql

@@ -0,0 +1,31 @@
+-- ailf.official_runs — canonical trend series
+--
+-- Filters ailf.reports down to the runs that form the canonical
+-- core-docs scheduled evaluation series (daily-baseline, weekly-full).
+-- Dashboards that must stay stable across ad-hoc / test / external-team
+-- activity should point at this view, not at `ailf.reports` directly.
+--
+-- Predicate (D0037):
+--   classification = 'official'  — intent: tracked trend series
+--   trigger_type   = 'scheduled' — mechanism: cron, not PR/manual/webhook
+--   owner_team     = 'core-docs' — attributable to the docs team
+--
+-- Historical rows predating D0037 have classification = NULL and are
+-- excluded. Backfill is a separate one-shot; see
+-- scripts/backfill-run-classification.ts.
+--
+-- Source: ailf.reports (view over ailf_raw.reports)
+-- Target: ailf.official_runs (this view)
+--
+-- Usage:
+--   bq query --use_legacy_sql=false < views/official_runs.sql
+--
+-- @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md
+-- @see docs/design-docs/run-classification-and-ownership.md
+
+CREATE OR REPLACE VIEW `data-platform-302218.ailf.official_runs` AS
+SELECT *
+FROM   `data-platform-302218.ailf.reports`
+WHERE  classification = 'official'
+  AND  trigger_type   = 'scheduled'
+  AND  owner_team     = 'core-docs';

package/config/bigquery/views/reports.sql

@@ -45,6 +45,25 @@ SELECT
   models,
   promptfoo_url,
   promptfoo_urls,
+  -- D0037 — run classification, ownership, executor, reproducibility.
+  -- Nullable for historical rows predating the taxonomy.
+  classification,
+  owner_team,
+  owner_individual,
+  executor_type,
+  executor_name,
+  executor_surface,
+  executor_github_actor,
+  purpose,
+  labels,
+  lineage_rerun_of,
+  lineage_compared_against,
+  lineage_parent_job_id,
+  tool_ailf_version,
+  tool_node_version,
+  host_platform,
+  host_arch,
+  host_ci,
   TIMESTAMP(_createdAt)                           AS synced_at
 FROM
   `data-platform-302218.ailf_raw.reports`;

package/config/bigquery/views/team_runs_template.sql

@@ -0,0 +1,17 @@
+-- ailf.team_runs_<team> — per-team filtered view factory
+--
+-- D0037 team ownership is a free-form string column (owner_team). Rather
+-- than pre-creating one view per team, use this template as a recipe:
+-- copy this file to `team_runs_<team>.sql`, fill in the slug, and run.
+--
+-- Example for the studio team:
+--
+--   CREATE OR REPLACE VIEW `data-platform-302218.ailf.team_runs_studio` AS
+--   SELECT *
+--   FROM   `data-platform-302218.ailf.reports`
+--   WHERE  owner_team = 'studio';
+--
+-- Teams with ad-hoc filtering needs can also query ailf.reports directly
+-- with `WHERE owner_team = 'X'` rather than maintaining a view.
+--
+-- @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md

package/dist/_vendor/ailf-core/examples/index.d.ts

@@ -433,6 +433,6 @@ export interface ExampleRecord {
 }
 export declare const EXAMPLES: Record<ExampleType, ExampleRecord>;
 /** GitHub Actions workflow template for AI Literacy evaluation */
-export declare const workflowYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# AI Literacy Evaluation \u2014 GitHub Actions workflow\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Evaluates your documentation quality on every pull request.\n# The AILF CLI reads your .ailf/tasks/ definitions, submits them\n# to the AILF API for evaluation, and writes a score report.\n#\n# Prerequisites:\n#   Add one secret to your repository (Settings \u2192 Secrets \u2192 Actions):\n#     AILF_API_KEY \u2014 your API key (starts with ailf_live_sk_)\n#\n#   @sanity/ailf is published with public npm access, so no npm token is\n#   needed to install the CLI.\n#\n# Customization:\n#   - Narrow the trigger paths to reduce cost (see comment below)\n#   - Check debug_mode for faster iteration (fewer tests)\n#   - See: https://github.com/sanity-labs/ai-literacy-framework\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\nname: AI Literacy Eval\n\non:\n  pull_request:\n    branches: [main]\n    # Runs on every PR to main by default. To reduce cost:\n    #   paths: [\".ailf/**\", \"docs/**\"]\n\n  workflow_dispatch:\n    inputs:\n      debug_mode:\n        description: \"Run in debug mode (fewer tests, faster iteration)\"\n        type: boolean\n        default: false\n\nconcurrency:\n  group: ailf-eval-${{ github.event.pull_request.number || github.ref }}\n  cancel-in-progress: true\n\njobs:\n  evaluate:\n    name: AI Literacy Evaluation\n    runs-on: ubuntu-latest\n    permissions:\n      contents: read\n      pull-requests: write\n    steps:\n      - uses: actions/checkout@v4\n\n      - name: Run evaluation\n        id: eval\n        env:\n          AILF_API_KEY: ${{ secrets.AILF_API_KEY }}\n        run: |\n          npx @sanity/ailf@latest pipeline --remote \\\n            --output /tmp/ailf-report.md \\\n            ${{ inputs.debug_mode && '--debug' || '' }}\n\n      - name: Post PR comment\n        if: always() && github.event_name == 'pull_request'\n        uses: actions/github-script@v7\n        with:\n          script: |\n            const fs = require('fs');\n\n            // --- Constants ---\n            const MARKER = '<!-- ailf-score-report -->';\n            const HISTORY_START = '<!-- ailf-score-history -->';\n            const HISTORY_END = '<!-- /ailf-score-history -->';\n            const MAX_HISTORY = 3; // keep at most 3 prior runs\n\n            // --- Read new report ---\n            let newReport;\n            try {\n              newReport = fs.readFileSync('/tmp/ailf-report.md', 'utf-8');\n            } catch {\n              newReport = `## \u26A0\uFE0F AI Literacy Evaluation\\n\\nNo report generated. Check the [workflow logs](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}).`;\n            }\n\n            const prNumber = context.issue?.number || context.payload?.pull_request?.number;\n            if (!prNumber) {\n              console.log('No PR number found, skipping comment');\n              return;\n            }\n\n            // --- Find existing comment ---\n            const { data: comments } = await github.rest.issues.listComments({\n              owner: context.repo.owner, repo: context.repo.repo,\n              issue_number: prNumber,\n            });\n            const existing = comments.find(c => c.body?.includes(MARKER));\n\n            // --- Build history from previous comment ---\n            let historyEntries = [];\n            if (existing) {\n              const oldBody = existing.body || '';\n\n              // Collect existing collapsed history entries\n              const histStart = oldBody.indexOf(HISTORY_START);\n              const histEnd = oldBody.indexOf(HISTORY_END);\n              if (histStart !== -1 && histEnd !== -1) {\n                const historyContent = oldBody.slice(histStart + HISTORY_START.length, histEnd).trim();\n                // Split on </details> boundaries to get individual entries\n                if (historyContent) {\n                  historyEntries = historyContent\n                    .split(/<\\/details>\\s*/)\n                    .map(s => s.trim())\n                    .filter(s => s.startsWith('<details>'))\n                    .map(s => s + '\\n</details>');\n                }\n              }\n\n              // Extract the current report (will become the newest history entry)\n              let previousReport = '';\n              if (histStart !== -1) {\n                // Report is between MARKER and the \"Previous runs\" heading (or history section)\n                const markerIdx = oldBody.indexOf(MARKER);\n                // Find the --- separator before history\n                const separatorIdx = oldBody.lastIndexOf('---', histStart);\n                const endIdx = separatorIdx > markerIdx ? separatorIdx : histStart;\n                previousReport = oldBody.slice(markerIdx + MARKER.length, endIdx).trim();\n              } else {\n                // No history yet \u2014 everything after MARKER is the report\n                const markerIdx = oldBody.indexOf(MARKER);\n                if (markerIdx !== -1) {\n                  previousReport = oldBody.slice(markerIdx + MARKER.length).trim();\n                }\n              }\n\n              // Collapse the previous report into a <details> entry\n              if (previousReport) {\n                const scoreMatch = previousReport.match(/Overall:\\s*(\\d+)\\/100/);\n                const score = scoreMatch ? scoreMatch[1] : '?';\n                const dateMatch = previousReport.match(/Generated by.*?\u00B7\\s*([^\u00B7<\\n*]+)/);\n                const date = dateMatch\n                  ? dateMatch[1].trim()\n                  : new Date().toISOString().slice(0, 16).replace('T', ' ') + ' UTC';\n                const entry = `<details>\\n<summary>\uD83D\uDCDC ${date} \u2014 ${score}/100</summary>\\n\\n${previousReport}\\n\\n</details>`;\n                historyEntries.unshift(entry); // newest first\n              }\n\n              // Enforce max history limit\n              historyEntries = historyEntries.slice(0, MAX_HISTORY);\n            }\n\n            // --- Assemble final comment ---\n            const historySection = historyEntries.length > 0\n              ? `\\n\\n---\\n\\n### \uD83D\uDCDC Previous runs\\n\\n${HISTORY_START}\\n${historyEntries.join('\\n\\n')}\\n${HISTORY_END}`\n              : '';\n            const finalBody = `${MARKER}\\n${newReport}${historySection}`;\n\n            if (existing) {\n              await github.rest.issues.updateComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                comment_id: existing.id, body: finalBody,\n              });\n              console.log(`Updated comment (${historyEntries.length} history entries)`);\n            } else {\n              await github.rest.issues.createComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                issue_number: prNumber, body: finalBody,\n              });\n              console.log('Created new PR comment');\n            }\n\n      - name: Summary\n        if: always()\n        run: |\n          if [ -f /tmp/ailf-report.md ]; then\n            cat /tmp/ailf-report.md >> \"$GITHUB_STEP_SUMMARY\"\n          else\n            echo \"## \u26A0\uFE0F AI Literacy Evaluation\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"No report generated. Check the workflow logs.\" >> \"$GITHUB_STEP_SUMMARY\"\n          fi\n";
+export declare const workflowYaml = "# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n# AI Literacy Evaluation \u2014 GitHub Actions workflow\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n#\n# Evaluates your documentation quality on every pull request.\n# The AILF CLI reads your .ailf/tasks/ definitions, submits them\n# to the AILF API for evaluation, and writes a score report.\n#\n# Prerequisites:\n#   Add one secret to your repository (Settings \u2192 Secrets \u2192 Actions):\n#     AILF_API_KEY \u2014 your API key (starts with ailf_live_sk_)\n#\n#   @sanity/ailf is published with public npm access, so no npm token is\n#   needed to install the CLI.\n#\n# Customization:\n#   - Narrow the trigger paths to reduce cost (see comment below)\n#   - Check debug_mode for faster iteration (fewer tests)\n#   - See: https://github.com/sanity-labs/ai-literacy-framework\n# \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\nname: AI Literacy Eval\n\non:\n  pull_request:\n    branches: [main]\n    # Runs on every PR to main by default. To reduce cost:\n    #   paths: [\".ailf/**\", \"docs/**\"]\n\n  workflow_dispatch:\n    inputs:\n      debug_mode:\n        description: \"Run in debug mode (fewer tests, faster iteration)\"\n        type: boolean\n        default: false\n\nconcurrency:\n  group: ailf-eval-${{ github.event.pull_request.number || github.ref }}\n  cancel-in-progress: true\n\njobs:\n  evaluate:\n    name: AI Literacy Evaluation\n    runs-on: ubuntu-latest\n    permissions:\n      contents: read\n      pull-requests: write\n    steps:\n      - uses: actions/checkout@v4\n\n      - name: Run evaluation\n        id: eval\n        env:\n          AILF_API_KEY: ${{ secrets.AILF_API_KEY }}\n          # \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n          # D0037 run provenance envelope \u2014 REPLACE THE OWNER TEAM SLUG\n          # below. Unedited templates produce runs tagged with the literal\n          # placeholder so you can spot them in Studio / BigQuery and fix.\n          #\n          # AILF_CLASSIFICATION values: official | ad-hoc | experimental |\n          # test | external. External teams should use `ad-hoc` by default;\n          # `official` is reserved for the core-docs scheduled series.\n          # \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n          AILF_CLASSIFICATION: ad-hoc\n          AILF_OWNER_TEAM: \"<REPLACE-WITH-YOUR-TEAM-SLUG>\"\n          AILF_OWNER_INDIVIDUAL: ${{ github.actor }}\n        run: |\n          npx @sanity/ailf@latest pipeline --remote \\\n            --output /tmp/ailf-report.md \\\n            ${{ inputs.debug_mode && '--debug' || '' }}\n\n      - name: Post PR comment\n        if: always() && github.event_name == 'pull_request'\n        uses: actions/github-script@v7\n        with:\n          script: |\n            const fs = require('fs');\n\n            // --- Constants ---\n            const MARKER = '<!-- ailf-score-report -->';\n            const HISTORY_START = '<!-- ailf-score-history -->';\n            const HISTORY_END = '<!-- /ailf-score-history -->';\n            const MAX_HISTORY = 3; // keep at most 3 prior runs\n\n            // --- Read new report ---\n            let newReport;\n            try {\n              newReport = fs.readFileSync('/tmp/ailf-report.md', 'utf-8');\n            } catch {\n              newReport = `## \u26A0\uFE0F AI Literacy Evaluation\\n\\nNo report generated. Check the [workflow logs](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}).`;\n            }\n\n            const prNumber = context.issue?.number || context.payload?.pull_request?.number;\n            if (!prNumber) {\n              console.log('No PR number found, skipping comment');\n              return;\n            }\n\n            // --- Find existing comment ---\n            const { data: comments } = await github.rest.issues.listComments({\n              owner: context.repo.owner, repo: context.repo.repo,\n              issue_number: prNumber,\n            });\n            const existing = comments.find(c => c.body?.includes(MARKER));\n\n            // --- Build history from previous comment ---\n            let historyEntries = [];\n            if (existing) {\n              const oldBody = existing.body || '';\n\n              // Collect existing collapsed history entries\n              const histStart = oldBody.indexOf(HISTORY_START);\n              const histEnd = oldBody.indexOf(HISTORY_END);\n              if (histStart !== -1 && histEnd !== -1) {\n                const historyContent = oldBody.slice(histStart + HISTORY_START.length, histEnd).trim();\n                // Split on </details> boundaries to get individual entries\n                if (historyContent) {\n                  historyEntries = historyContent\n                    .split(/<\\/details>\\s*/)\n                    .map(s => s.trim())\n                    .filter(s => s.startsWith('<details>'))\n                    .map(s => s + '\\n</details>');\n                }\n              }\n\n              // Extract the current report (will become the newest history entry)\n              let previousReport = '';\n              if (histStart !== -1) {\n                // Report is between MARKER and the \"Previous runs\" heading (or history section)\n                const markerIdx = oldBody.indexOf(MARKER);\n                // Find the --- separator before history\n                const separatorIdx = oldBody.lastIndexOf('---', histStart);\n                const endIdx = separatorIdx > markerIdx ? separatorIdx : histStart;\n                previousReport = oldBody.slice(markerIdx + MARKER.length, endIdx).trim();\n              } else {\n                // No history yet \u2014 everything after MARKER is the report\n                const markerIdx = oldBody.indexOf(MARKER);\n                if (markerIdx !== -1) {\n                  previousReport = oldBody.slice(markerIdx + MARKER.length).trim();\n                }\n              }\n\n              // Collapse the previous report into a <details> entry\n              if (previousReport) {\n                const scoreMatch = previousReport.match(/Overall:\\s*(\\d+)\\/100/);\n                const score = scoreMatch ? scoreMatch[1] : '?';\n                const dateMatch = previousReport.match(/Generated by.*?\u00B7\\s*([^\u00B7<\\n*]+)/);\n                const date = dateMatch\n                  ? dateMatch[1].trim()\n                  : new Date().toISOString().slice(0, 16).replace('T', ' ') + ' UTC';\n                const entry = `<details>\\n<summary>\uD83D\uDCDC ${date} \u2014 ${score}/100</summary>\\n\\n${previousReport}\\n\\n</details>`;\n                historyEntries.unshift(entry); // newest first\n              }\n\n              // Enforce max history limit\n              historyEntries = historyEntries.slice(0, MAX_HISTORY);\n            }\n\n            // --- Assemble final comment ---\n            const historySection = historyEntries.length > 0\n              ? `\\n\\n---\\n\\n### \uD83D\uDCDC Previous runs\\n\\n${HISTORY_START}\\n${historyEntries.join('\\n\\n')}\\n${HISTORY_END}`\n              : '';\n            const finalBody = `${MARKER}\\n${newReport}${historySection}`;\n\n            if (existing) {\n              await github.rest.issues.updateComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                comment_id: existing.id, body: finalBody,\n              });\n              console.log(`Updated comment (${historyEntries.length} history entries)`);\n            } else {\n              await github.rest.issues.createComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                issue_number: prNumber, body: finalBody,\n              });\n              console.log('Created new PR comment');\n            }\n\n      - name: Summary\n        if: always()\n        run: |\n          if [ -f /tmp/ailf-report.md ]; then\n            cat /tmp/ailf-report.md >> \"$GITHUB_STEP_SUMMARY\"\n          else\n            echo \"## \u26A0\uFE0F AI Literacy Evaluation\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"No report generated. Check the workflow logs.\" >> \"$GITHUB_STEP_SUMMARY\"\n          fi\n";
 /** TypeScript project configuration template (ailf.config.ts) */
 export declare const ailfConfigTs = "/**\n * .ailf/ailf.config.ts \u2014 AI Literacy Framework project configuration.\n *\n * This file configures how the AILF evaluation pipeline runs in this\n * repository. Place it at .ailf/ailf.config.ts in your project root.\n *\n * Evaluations are submitted to the AILF API (ailf-api.sanity.build).\n * The API handles LLM calls, doc fetching, grading, and report\n * publishing. Your repo only needs one secret: AILF_API_KEY.\n *\n * Docs: https://github.com/sanity-labs/ai-literacy-framework\n */\n\nexport default {\n  /**\n   * Documentation source \u2014 which docs are being evaluated.\n   *\n   * This tells the pipeline which Sanity project and dataset contain\n   * the documentation under test. For most users, this is Sanity's own\n   * docs project.\n   */\n  source: {\n    /** Sanity project ID (find yours at sanity.io/manage) */\n    projectId: \"3do82whm\",\n    /** The dataset to query (e.g., \"production\", \"next\") */\n    dataset: \"next\",\n    /**\n     * The public URL of your documentation site.\n     * Used by agentic mode to test agent discoverability.\n     */\n    baseUrl: \"https://www.sanity.io/docs\",\n  },\n\n  /**\n   * Trigger configuration \u2014 when evaluations run automatically.\n   *\n   * Each key is a trigger context. The pipeline checks which trigger\n   * matches the current execution context (PR, merge, schedule, etc.)\n   * and applies its settings.\n   *\n   * Mode options:\n   *   \"validate-only\" \u2014 check that task files parse correctly (fast, no LLM calls)\n   *   \"eval\"          \u2014 run the full evaluation pipeline\n   */\n  triggers: {\n    /** On pull requests: just validate task files parse correctly. */\n    pr: {\n      mode: \"validate-only\",\n    },\n\n    /** When .ailf/ files change in a PR: run a real evaluation. */\n    \"pr-task-change\": {\n      mode: \"eval\",\n      paths: [\".ailf/**\"],\n    },\n\n    /** On merge to main: run evaluation (non-blocking). */\n    main: {\n      mode: \"eval\",\n      blocking: false,\n      notify: true,\n    },\n  },\n}\n";

package/dist/_vendor/ailf-core/examples/index.js

@@ -630,7 +630,7 @@ export const EXAMPLES = {
 // Raw file exports (non-data files, exported as raw strings)
 // ---------------------------------------------------------------------------
 /** GitHub Actions workflow template for AI Literacy evaluation */
-export const workflowYaml = "# ──────────────────────────────────────────────────────────────────────\n# AI Literacy Evaluation — GitHub Actions workflow\n# ──────────────────────────────────────────────────────────────────────\n#\n# Evaluates your documentation quality on every pull request.\n# The AILF CLI reads your .ailf/tasks/ definitions, submits them\n# to the AILF API for evaluation, and writes a score report.\n#\n# Prerequisites:\n#   Add one secret to your repository (Settings → Secrets → Actions):\n#     AILF_API_KEY — your API key (starts with ailf_live_sk_)\n#\n#   @sanity/ailf is published with public npm access, so no npm token is\n#   needed to install the CLI.\n#\n# Customization:\n#   - Narrow the trigger paths to reduce cost (see comment below)\n#   - Check debug_mode for faster iteration (fewer tests)\n#   - See: https://github.com/sanity-labs/ai-literacy-framework\n# ──────────────────────────────────────────────────────────────────────\n\nname: AI Literacy Eval\n\non:\n  pull_request:\n    branches: [main]\n    # Runs on every PR to main by default. To reduce cost:\n    #   paths: [\".ailf/**\", \"docs/**\"]\n\n  workflow_dispatch:\n    inputs:\n      debug_mode:\n        description: \"Run in debug mode (fewer tests, faster iteration)\"\n        type: boolean\n        default: false\n\nconcurrency:\n  group: ailf-eval-${{ github.event.pull_request.number || github.ref }}\n  cancel-in-progress: true\n\njobs:\n  evaluate:\n    name: AI Literacy Evaluation\n    runs-on: ubuntu-latest\n    permissions:\n      contents: read\n      pull-requests: write\n    steps:\n      - uses: actions/checkout@v4\n\n      - name: Run evaluation\n        id: eval\n        env:\n          AILF_API_KEY: ${{ secrets.AILF_API_KEY }}\n        run: |\n          npx @sanity/ailf@latest pipeline --remote \\\n            --output /tmp/ailf-report.md \\\n            ${{ inputs.debug_mode && '--debug' || '' }}\n\n      - name: Post PR comment\n        if: always() && github.event_name == 'pull_request'\n        uses: actions/github-script@v7\n        with:\n          script: |\n            const fs = require('fs');\n\n            // --- Constants ---\n            const MARKER = '<!-- ailf-score-report -->';\n            const HISTORY_START = '<!-- ailf-score-history -->';\n            const HISTORY_END = '<!-- /ailf-score-history -->';\n            const MAX_HISTORY = 3; // keep at most 3 prior runs\n\n            // --- Read new report ---\n            let newReport;\n            try {\n              newReport = fs.readFileSync('/tmp/ailf-report.md', 'utf-8');\n            } catch {\n              newReport = `## ⚠️ AI Literacy Evaluation\\n\\nNo report generated. Check the [workflow logs](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}).`;\n            }\n\n            const prNumber = context.issue?.number || context.payload?.pull_request?.number;\n            if (!prNumber) {\n              console.log('No PR number found, skipping comment');\n              return;\n            }\n\n            // --- Find existing comment ---\n            const { data: comments } = await github.rest.issues.listComments({\n              owner: context.repo.owner, repo: context.repo.repo,\n              issue_number: prNumber,\n            });\n            const existing = comments.find(c => c.body?.includes(MARKER));\n\n            // --- Build history from previous comment ---\n            let historyEntries = [];\n            if (existing) {\n              const oldBody = existing.body || '';\n\n              // Collect existing collapsed history entries\n              const histStart = oldBody.indexOf(HISTORY_START);\n              const histEnd = oldBody.indexOf(HISTORY_END);\n              if (histStart !== -1 && histEnd !== -1) {\n                const historyContent = oldBody.slice(histStart + HISTORY_START.length, histEnd).trim();\n                // Split on </details> boundaries to get individual entries\n                if (historyContent) {\n                  historyEntries = historyContent\n                    .split(/<\\/details>\\s*/)\n                    .map(s => s.trim())\n                    .filter(s => s.startsWith('<details>'))\n                    .map(s => s + '\\n</details>');\n                }\n              }\n\n              // Extract the current report (will become the newest history entry)\n              let previousReport = '';\n              if (histStart !== -1) {\n                // Report is between MARKER and the \"Previous runs\" heading (or history section)\n                const markerIdx = oldBody.indexOf(MARKER);\n                // Find the --- separator before history\n                const separatorIdx = oldBody.lastIndexOf('---', histStart);\n                const endIdx = separatorIdx > markerIdx ? separatorIdx : histStart;\n                previousReport = oldBody.slice(markerIdx + MARKER.length, endIdx).trim();\n              } else {\n                // No history yet — everything after MARKER is the report\n                const markerIdx = oldBody.indexOf(MARKER);\n                if (markerIdx !== -1) {\n                  previousReport = oldBody.slice(markerIdx + MARKER.length).trim();\n                }\n              }\n\n              // Collapse the previous report into a <details> entry\n              if (previousReport) {\n                const scoreMatch = previousReport.match(/Overall:\\s*(\\d+)\\/100/);\n                const score = scoreMatch ? scoreMatch[1] : '?';\n                const dateMatch = previousReport.match(/Generated by.*?·\\s*([^·<\\n*]+)/);\n                const date = dateMatch\n                  ? dateMatch[1].trim()\n                  : new Date().toISOString().slice(0, 16).replace('T', ' ') + ' UTC';\n                const entry = `<details>\\n<summary>📜 ${date} — ${score}/100</summary>\\n\\n${previousReport}\\n\\n</details>`;\n                historyEntries.unshift(entry); // newest first\n              }\n\n              // Enforce max history limit\n              historyEntries = historyEntries.slice(0, MAX_HISTORY);\n            }\n\n            // --- Assemble final comment ---\n            const historySection = historyEntries.length > 0\n              ? `\\n\\n---\\n\\n### 📜 Previous runs\\n\\n${HISTORY_START}\\n${historyEntries.join('\\n\\n')}\\n${HISTORY_END}`\n              : '';\n            const finalBody = `${MARKER}\\n${newReport}${historySection}`;\n\n            if (existing) {\n              await github.rest.issues.updateComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                comment_id: existing.id, body: finalBody,\n              });\n              console.log(`Updated comment (${historyEntries.length} history entries)`);\n            } else {\n              await github.rest.issues.createComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                issue_number: prNumber, body: finalBody,\n              });\n              console.log('Created new PR comment');\n            }\n\n      - name: Summary\n        if: always()\n        run: |\n          if [ -f /tmp/ailf-report.md ]; then\n            cat /tmp/ailf-report.md >> \"$GITHUB_STEP_SUMMARY\"\n          else\n            echo \"## ⚠️ AI Literacy Evaluation\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"No report generated. Check the workflow logs.\" >> \"$GITHUB_STEP_SUMMARY\"\n          fi\n";
+export const workflowYaml = "# ──────────────────────────────────────────────────────────────────────\n# AI Literacy Evaluation — GitHub Actions workflow\n# ──────────────────────────────────────────────────────────────────────\n#\n# Evaluates your documentation quality on every pull request.\n# The AILF CLI reads your .ailf/tasks/ definitions, submits them\n# to the AILF API for evaluation, and writes a score report.\n#\n# Prerequisites:\n#   Add one secret to your repository (Settings → Secrets → Actions):\n#     AILF_API_KEY — your API key (starts with ailf_live_sk_)\n#\n#   @sanity/ailf is published with public npm access, so no npm token is\n#   needed to install the CLI.\n#\n# Customization:\n#   - Narrow the trigger paths to reduce cost (see comment below)\n#   - Check debug_mode for faster iteration (fewer tests)\n#   - See: https://github.com/sanity-labs/ai-literacy-framework\n# ──────────────────────────────────────────────────────────────────────\n\nname: AI Literacy Eval\n\non:\n  pull_request:\n    branches: [main]\n    # Runs on every PR to main by default. To reduce cost:\n    #   paths: [\".ailf/**\", \"docs/**\"]\n\n  workflow_dispatch:\n    inputs:\n      debug_mode:\n        description: \"Run in debug mode (fewer tests, faster iteration)\"\n        type: boolean\n        default: false\n\nconcurrency:\n  group: ailf-eval-${{ github.event.pull_request.number || github.ref }}\n  cancel-in-progress: true\n\njobs:\n  evaluate:\n    name: AI Literacy Evaluation\n    runs-on: ubuntu-latest\n    permissions:\n      contents: read\n      pull-requests: write\n    steps:\n      - uses: actions/checkout@v4\n\n      - name: Run evaluation\n        id: eval\n        env:\n          AILF_API_KEY: ${{ secrets.AILF_API_KEY }}\n          # ────────────────────────────────────────────────────────────\n          # D0037 run provenance envelope — REPLACE THE OWNER TEAM SLUG\n          # below. Unedited templates produce runs tagged with the literal\n          # placeholder so you can spot them in Studio / BigQuery and fix.\n          #\n          # AILF_CLASSIFICATION values: official | ad-hoc | experimental |\n          # test | external. External teams should use `ad-hoc` by default;\n          # `official` is reserved for the core-docs scheduled series.\n          # ────────────────────────────────────────────────────────────\n          AILF_CLASSIFICATION: ad-hoc\n          AILF_OWNER_TEAM: \"<REPLACE-WITH-YOUR-TEAM-SLUG>\"\n          AILF_OWNER_INDIVIDUAL: ${{ github.actor }}\n        run: |\n          npx @sanity/ailf@latest pipeline --remote \\\n            --output /tmp/ailf-report.md \\\n            ${{ inputs.debug_mode && '--debug' || '' }}\n\n      - name: Post PR comment\n        if: always() && github.event_name == 'pull_request'\n        uses: actions/github-script@v7\n        with:\n          script: |\n            const fs = require('fs');\n\n            // --- Constants ---\n            const MARKER = '<!-- ailf-score-report -->';\n            const HISTORY_START = '<!-- ailf-score-history -->';\n            const HISTORY_END = '<!-- /ailf-score-history -->';\n            const MAX_HISTORY = 3; // keep at most 3 prior runs\n\n            // --- Read new report ---\n            let newReport;\n            try {\n              newReport = fs.readFileSync('/tmp/ailf-report.md', 'utf-8');\n            } catch {\n              newReport = `## ⚠️ AI Literacy Evaluation\\n\\nNo report generated. Check the [workflow logs](${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}).`;\n            }\n\n            const prNumber = context.issue?.number || context.payload?.pull_request?.number;\n            if (!prNumber) {\n              console.log('No PR number found, skipping comment');\n              return;\n            }\n\n            // --- Find existing comment ---\n            const { data: comments } = await github.rest.issues.listComments({\n              owner: context.repo.owner, repo: context.repo.repo,\n              issue_number: prNumber,\n            });\n            const existing = comments.find(c => c.body?.includes(MARKER));\n\n            // --- Build history from previous comment ---\n            let historyEntries = [];\n            if (existing) {\n              const oldBody = existing.body || '';\n\n              // Collect existing collapsed history entries\n              const histStart = oldBody.indexOf(HISTORY_START);\n              const histEnd = oldBody.indexOf(HISTORY_END);\n              if (histStart !== -1 && histEnd !== -1) {\n                const historyContent = oldBody.slice(histStart + HISTORY_START.length, histEnd).trim();\n                // Split on </details> boundaries to get individual entries\n                if (historyContent) {\n                  historyEntries = historyContent\n                    .split(/<\\/details>\\s*/)\n                    .map(s => s.trim())\n                    .filter(s => s.startsWith('<details>'))\n                    .map(s => s + '\\n</details>');\n                }\n              }\n\n              // Extract the current report (will become the newest history entry)\n              let previousReport = '';\n              if (histStart !== -1) {\n                // Report is between MARKER and the \"Previous runs\" heading (or history section)\n                const markerIdx = oldBody.indexOf(MARKER);\n                // Find the --- separator before history\n                const separatorIdx = oldBody.lastIndexOf('---', histStart);\n                const endIdx = separatorIdx > markerIdx ? separatorIdx : histStart;\n                previousReport = oldBody.slice(markerIdx + MARKER.length, endIdx).trim();\n              } else {\n                // No history yet — everything after MARKER is the report\n                const markerIdx = oldBody.indexOf(MARKER);\n                if (markerIdx !== -1) {\n                  previousReport = oldBody.slice(markerIdx + MARKER.length).trim();\n                }\n              }\n\n              // Collapse the previous report into a <details> entry\n              if (previousReport) {\n                const scoreMatch = previousReport.match(/Overall:\\s*(\\d+)\\/100/);\n                const score = scoreMatch ? scoreMatch[1] : '?';\n                const dateMatch = previousReport.match(/Generated by.*?·\\s*([^·<\\n*]+)/);\n                const date = dateMatch\n                  ? dateMatch[1].trim()\n                  : new Date().toISOString().slice(0, 16).replace('T', ' ') + ' UTC';\n                const entry = `<details>\\n<summary>📜 ${date} — ${score}/100</summary>\\n\\n${previousReport}\\n\\n</details>`;\n                historyEntries.unshift(entry); // newest first\n              }\n\n              // Enforce max history limit\n              historyEntries = historyEntries.slice(0, MAX_HISTORY);\n            }\n\n            // --- Assemble final comment ---\n            const historySection = historyEntries.length > 0\n              ? `\\n\\n---\\n\\n### 📜 Previous runs\\n\\n${HISTORY_START}\\n${historyEntries.join('\\n\\n')}\\n${HISTORY_END}`\n              : '';\n            const finalBody = `${MARKER}\\n${newReport}${historySection}`;\n\n            if (existing) {\n              await github.rest.issues.updateComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                comment_id: existing.id, body: finalBody,\n              });\n              console.log(`Updated comment (${historyEntries.length} history entries)`);\n            } else {\n              await github.rest.issues.createComment({\n                owner: context.repo.owner, repo: context.repo.repo,\n                issue_number: prNumber, body: finalBody,\n              });\n              console.log('Created new PR comment');\n            }\n\n      - name: Summary\n        if: always()\n        run: |\n          if [ -f /tmp/ailf-report.md ]; then\n            cat /tmp/ailf-report.md >> \"$GITHUB_STEP_SUMMARY\"\n          else\n            echo \"## ⚠️ AI Literacy Evaluation\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"\" >> \"$GITHUB_STEP_SUMMARY\"\n            echo \"No report generated. Check the workflow logs.\" >> \"$GITHUB_STEP_SUMMARY\"\n          fi\n";
 // ---------------------------------------------------------------------------
 // TypeScript template exports (for ailf init --output-format ts)
 // ---------------------------------------------------------------------------

package/dist/_vendor/ailf-core/ports/context.d.ts

@@ -11,6 +11,7 @@
  * Fields marked optional are transitional — they will become required
  * as downstream consumers are converted to use them.
  */
+import type { RunClassification, RunExecutorSurface } from "../../ailf-shared/index.d.ts";
 import type { RunId } from "../types/branded-ids.js";
 import type { DebugOptions, EvalMode, PluginRegistry } from "../types/index.js";
 import type { ArtifactWriter } from "./artifact-writer.js";
@@ -138,6 +139,30 @@ export interface ResolvedConfig {
         repo: string;
         sha?: string;
     };
+    /**
+     * Caller-provided D0037 provenance envelope (classification, owner,
+     * executor, purpose, labels). Set on --remote submissions so the
+     * caller's `AILF_*` env vars / CLI flags survive the API boundary.
+     * When set, buildRunContext prefers these over server-env detection —
+     * same pattern as callerGit.
+     *
+     * @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md
+     */
+    callerEnvelope?: {
+        classification?: RunClassification;
+        owner?: {
+            team: string;
+            individual?: string;
+        };
+        executor?: {
+            type: "user";
+            surface: RunExecutorSurface;
+            name?: string;
+            githubActor?: string;
+        };
+        purpose?: string;
+        labels?: string[];
+    };
     /** Callback URL configuration for API-triggered evaluations */
     callback?: {
         url: string;

package/dist/_vendor/ailf-core/schemas/pipeline-request.d.ts

@@ -87,6 +87,29 @@ export declare const PipelineRequestSchema: z.ZodObject<{
         full: "full";
     }>>;
     presets: z.ZodOptional<z.ZodArray<z.ZodString>>;
+    classification: z.ZodOptional<z.ZodEnum<{
+        external: "external";
+        official: "official";
+        "ad-hoc": "ad-hoc";
+        experimental: "experimental";
+        test: "test";
+    }>>;
+    owner: z.ZodOptional<z.ZodObject<{
+        team: z.ZodString;
+        individual: z.ZodOptional<z.ZodString>;
+    }, z.core.$strip>>;
+    executor: z.ZodOptional<z.ZodObject<{
+        type: z.ZodLiteral<"user">;
+        surface: z.ZodEnum<{
+            cli: "cli";
+            studio: "studio";
+            api: "api";
+        }>;
+        name: z.ZodOptional<z.ZodString>;
+        githubActor: z.ZodOptional<z.ZodString>;
+    }, z.core.$strict>>;
+    purpose: z.ZodOptional<z.ZodString>;
+    labels: z.ZodOptional<z.ZodArray<z.ZodString>>;
 }, z.core.$strip>;
 /** Inferred TypeScript type for a pipeline request payload. */
 export type PipelineRequest = z.infer<typeof PipelineRequestSchema>;

package/dist/_vendor/ailf-core/schemas/pipeline-request.js

@@ -13,7 +13,7 @@
  * @see packages/eval/src/pipeline/map-request-to-config.ts — maps to ResolvedConfig
  */
 import { z } from "zod";
-import { LITERACY_VARIANTS, RAW_EVAL_MODES } from "../../ailf-shared/index.js";
+import { LITERACY_VARIANTS, RAW_EVAL_MODES, RUN_CLASSIFICATIONS, RUN_EXECUTOR_SURFACES, } from "../../ailf-shared/index.js";
 // ---------------------------------------------------------------------------
 // Debug options — boolean shorthand or structured object
 // ---------------------------------------------------------------------------
@@ -49,6 +49,47 @@ const CallerGitSchema = z.object({
     sha: z.string().optional(),
 });
 // ---------------------------------------------------------------------------
+// Caller envelope (D0037) — for --remote evaluations from external repos
+// ---------------------------------------------------------------------------
+/**
+ * Caller-provided D0037 provenance envelope.
+ *
+ * When the CLI submits a PipelineRequest via `--remote`, the pipeline
+ * runs server-side (Cloud Run / dispatched GH Actions) where the
+ * caller's local env vars don't exist. Carrying the envelope on the
+ * request lets the caller's `AILF_CLASSIFICATION` / `AILF_OWNER_TEAM`
+ * / explicit CLI flags survive the API boundary so provenance
+ * attributes to the caller, not the server runtime.
+ *
+ * Only caller-identity fields cross the wire. Server-environment
+ * facts (`executor.email`, `tool.ailfVersion`, `tool.nodeVersion`,
+ * `host.*`) stay server-inferred on the receiving side.
+ *
+ * @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md
+ */
+const CallerOwnerSchema = z.object({
+    team: z.string().min(1),
+    individual: z.string().optional(),
+});
+/**
+ * Executor payload — user variant only. System executors (github-actions,
+ * cloud-run) are always server-detected and must not be spoofable by a
+ * caller, so the wire format excludes them.
+ *
+ * `.strict()` makes unknown keys fail the parse rather than be silently
+ * stripped. Critical for PII: a caller that accidentally sends `email`
+ * gets an immediate 400 instead of a silent drop that looks like success.
+ * Server-inferred fields (`email`, `tool`, `host`) are NOT accepted here.
+ */
+const CallerExecutorSchema = z
+    .object({
+    type: z.literal("user"),
+    surface: z.enum(RUN_EXECUTOR_SURFACES),
+    name: z.string().optional(),
+    githubActor: z.string().optional(),
+})
+    .strict();
+// ---------------------------------------------------------------------------
 // Pipeline Request — the universal invocation contract
 // ---------------------------------------------------------------------------
 export const PipelineRequestSchema = z.object({
@@ -103,4 +144,21 @@ export const PipelineRequestSchema = z.object({
     variant: z.enum(LITERACY_VARIANTS).optional(),
     /** External preset file paths or npm package names to load */
     presets: z.array(z.string()).optional(),
+    // -------------------------------------------------------------------------
+    // D0037 caller envelope — classification, attribution, and intent
+    // -------------------------------------------------------------------------
+    /**
+     * How this run should be treated for reporting and trend tracking.
+     * Orthogonal to `trigger.type` (captured server-side). When omitted,
+     * the server defaults to `"ad-hoc"`.
+     */
+    classification: z.enum(RUN_CLASSIFICATIONS).optional(),
+    /** Team and (optionally) individual this run is attributable to. */
+    owner: CallerOwnerSchema.optional(),
+    /** Caller executor identity (user variant only). */
+    executor: CallerExecutorSchema.optional(),
+    /** Human-authored "why I ran this". */
+    purpose: z.string().optional(),
+    /** Free-form searchable tags (release IDs, regression hunts, experiments). */
+    labels: z.array(z.string().min(1)).optional(),
 });

package/dist/_vendor/ailf-shared/index.d.ts

@@ -14,5 +14,7 @@ export * from "./feature-flags.js";
 export * from "./score-grades.js";
 export * from "./noise-threshold.js";
 export * from "./eval-modes.js";
+export * from "./owner-teams.js";
+export * from "./run-classification.js";
 export * from "./run-trigger.js";
 export * from "./run-context.js";

package/dist/_vendor/ailf-shared/index.js

@@ -14,5 +14,7 @@ export * from "./feature-flags.js";
 export * from "./score-grades.js";
 export * from "./noise-threshold.js";
 export * from "./eval-modes.js";
+export * from "./owner-teams.js";
+export * from "./run-classification.js";
 export * from "./run-trigger.js";
 export * from "./run-context.js";

package/dist/_vendor/ailf-shared/owner-teams.d.ts

@@ -0,0 +1,26 @@
+/**
+ * Known owner-team slugs and soft-normalization helper.
+ *
+ * `RunOwner.team` is free-form by design (external teams name themselves
+ * and internal names drift). This module provides two things to keep UX
+ * polished without blocking new entrants:
+ *
+ *   - `KNOWN_OWNER_TEAMS` — a seed list of canonical slugs that populates
+ *     Studio filter comboboxes as suggestions. Not a closed enum.
+ *   - `normalizeOwnerTeam()` — maps a handful of common aliases to
+ *     canonical slugs. Warn-only: returns the original string when no
+ *     mapping applies. Adding an alias here is a one-liner.
+ *
+ * @see docs/decisions/D0037-run-classification-and-ownership-taxonomy.md
+ */
+export declare const KNOWN_OWNER_TEAMS: readonly string[];
+/**
+ * Normalize a free-form team slug to its canonical form.
+ *
+ * - Trims and lowercases.
+ * - Maps known aliases to canonical slugs.
+ * - Passes unknown values through unchanged (warn-only at the UI layer).
+ * - Returns `"unknown"` for empty input.
+ */
+export declare function normalizeOwnerTeam(value: string | undefined | null): string;
+export declare function isKnownOwnerTeam(value: string): boolean;