TalonT-Org · Trecek · Apr 13, 2026 · Apr 12, 2026 · Apr 12, 2026 · Apr 12, 2026
diff --git a/.gitattributes b/.gitattributes
@@ -0,0 +1,3 @@
+# Vendored binary assets — suppress diff spam on updates
+src/autoskillit/assets/**/*.js binary
+src/autoskillit/assets/**/*.css binary
diff --git a/README.md b/README.md
@@ -5,7 +5,7 @@
 AutoSkillit is a Claude Code plugin that runs YAML recipes through a two-tier
 orchestrator. Bundled recipes turn GitHub issues into merged PRs by chaining
 plan, dry-walkthrough, worktree, test, and PR-review skills against 42 MCP
-tools and 95 bundled skills.
+tools and 108 bundled skills.
 
 https://github.com/user-attachments/assets/bcd910c8-7269-46d6-a496-53b2cb24d212
 

diff --git a/Taskfile.yml b/Taskfile.yml
@@ -174,3 +174,16 @@ tasks:
     cmds:
       - uv tool install --force "autoskillit[dev] @ git+https://github.com/TalonT-Org/AutoSkillit.git@integration"
       - "$(uv tool dir)/autoskillit/bin/autoskillit install"
+
+  vendor-mermaid:
+    desc: Download latest mermaid v11 UMD bundle to src/autoskillit/assets/mermaid/
+    cmds:
+      - mkdir -p src/autoskillit/assets/mermaid
+      - |
+        VERSION=$(curl -sfL https://unpkg.com/mermaid@11/package.json \
+          | python3 -c "import sys, json; print(json.load(sys.stdin)['version'])")
+        echo "Fetching mermaid ${VERSION}..."
+        curl -sfL "https://unpkg.com/mermaid@${VERSION}/dist/mermaid.min.js" \
+          -o src/autoskillit/assets/mermaid/mermaid.min.js
+        echo "${VERSION}" > src/autoskillit/assets/mermaid/VERSION
+        echo "Vendored mermaid ${VERSION} ($(wc -c < src/autoskillit/assets/mermaid/mermaid.min.js) bytes)"
diff --git a/docs/README.md b/docs/README.md
@@ -2,7 +2,7 @@
 
 AutoSkillit is a Claude Code plugin that runs YAML recipes through a
 two-tier orchestrator. The bundled recipes implement issue → plan → worktree
-→ tests → PR → merge pipelines using 42 MCP tools and 95 bundled skills.
+→ tests → PR → merge pipelines using 42 MCP tools and 108 bundled skills.
 
 ## Start here
 

diff --git a/docs/developer/end-turn-hazards.md b/docs/developer/end-turn-hazards.md
@@ -190,7 +190,7 @@ two detectors on every SKILL.md in the project:
 | `_check_loop_boundary()` | "For each" loops with tool invocations but no anti-prose guard |
 
 These run as part of `test_no_text_then_tool_in_any_step`, a parametrized
-test that scans all 95 bundled skills. Any new skill with an unguarded
+test that scans all 108 bundled skills. Any new skill with an unguarded
 loop fails CI automatically.
 
 ## If This Gets Fixed Upstream

diff --git a/docs/examples/research-pipeline.md b/docs/examples/research-pipeline.md
@@ -20,7 +20,7 @@ fifth example, update the test fixture in the same commit.
 The first run exercised the `research` recipe end to end against an empty
 target repository. The orchestrator collected ingredients, ran
 `scope` → `plan-experiment` → `implement-experiment` → `run-experiment` →
-`write-report`, and committed both the experiment results and the auxiliary
+`generate-report`, and committed both the experiment results and the auxiliary
 artifacts in two separate PRs so reviewers could read the report without
 diffing the data.
 

diff --git a/docs/execution/architecture.md b/docs/execution/architecture.md
@@ -4,7 +4,7 @@ How AutoSkillit runs a recipe end to end: orchestrator, kitchen gating, clone an
 
 ## Overview
 
-AutoSkillit is a Claude Code plugin that orchestrates automated workflows using headless sessions. It provides 42 MCP tools and 95 bundled skills, organized into a gated visibility system.
+AutoSkillit is a Claude Code plugin that orchestrates automated workflows using headless sessions. It provides 42 MCP tools and 108 bundled skills, organized into a gated visibility system.
 
 ## Core Concepts
 

diff --git a/docs/recipes/overview.md b/docs/recipes/overview.md
@@ -96,7 +96,7 @@ Two-phase technical research recipe. Phase 1 scopes a research question and open
 | `base_branch` | `main` | Branch to target for the research PR |
 | `setup_phases` | `false` | When `true`, decompose experiment into sequenced setup phases |
 
-**Requires pack:** `research` (pack members: `scope`, `plan-experiment`, `implement-experiment`, `run-experiment`, `write-report`)
+**Requires pack:** `research` (pack members: `scope`, `plan-experiment`, `implement-experiment`, `run-experiment`, `generate-report`)
 
 ## Project Recipes
 

diff --git a/docs/skills/catalog.md b/docs/skills/catalog.md
@@ -1,7 +1,7 @@
 # Skill catalog
 
-The complete list of bundled skills (95 total: 3 in `src/autoskillit/skills/`,
-92 in `src/autoskillit/skills_extended/`). Filesystem walk this directory if
+The complete list of bundled skills (109 total: 3 in `src/autoskillit/skills/`,
+106 in `src/autoskillit/skills_extended/`). Filesystem walk this directory if
 you need an exhaustive listing; this catalog groups by purpose.
 
 ## Tier 1 — free range (3)
@@ -33,7 +33,7 @@ Located under `src/autoskillit/skills_extended/`. Grouped by purpose:
 
 ### Experiment family
 `scope`, `plan-experiment`, `implement-experiment`, `run-experiment`,
-`write-report`, `troubleshoot-experiment`
+`generate-report`, `troubleshoot-experiment`
 
 ## Tier 3 — pipeline / automation
 
@@ -91,6 +91,27 @@ answers a specific question about the system:
 | `exp-lens-validity-threats` | Validity threats |
 | `exp-lens-variance-stability` | Variance stability |
 
+## vis-lens family (12)
+
+12 visualization-planning lenses orchestrated by `plan-visualization`, under
+`skills_extended/vis-lens-*/`. Each answers a specific question about a figure or the
+figure set:
+
+| # | Skill | Cognitive Mode | Primary Question | Priority |
+|---|-------|---------------|------------------|----------|
+| 1 | `vis-lens-always-on` | Composite | Is everything correct by default? | P0 |
+| 2 | `vis-lens-antipattern` | Diagnostic | What visualization antipatterns are present? | P0 |
+| 3 | `vis-lens-chart-select` | Typological | What chart type fits this data shape? | P0 |
+| 4 | `vis-lens-domain-norms` | Normative | Does the figure follow domain conventions? | P0 |
+| 5 | `vis-lens-uncertainty` | Probabilistic | Is uncertainty properly communicated? | P0 |
+| 6 | `vis-lens-color-access` | Chromatic | Is the color encoding accessible and perceptually uniform? | P1 |
+| 7 | `vis-lens-figure-table` | Decisional | Should this result be a figure or a table? | P1 |
+| 8 | `vis-lens-multi-compare` | Comparative | Are multi-condition comparisons statistically sound? | P1 |
+| 9 | `vis-lens-temporal` | Sequential | Is temporal/sequential data displayed correctly? | P1 |
+| 10 | `vis-lens-caption-annot` | Annotative | Are figure captions and axis labels fully self-contained? | P2 |
+| 11 | `vis-lens-story-arc` | Narrative | Do the figures tell a coherent story across the report? | P2 |
+| 12 | `vis-lens-reproducibility` | Replicative | Can the figures be reproduced from the data and code? | P2 |
+
 ## Rectify doctrine
 
 Several Tier 2 skills (`rectify`, `audit-bugs`, `design-guards`,
@@ -100,8 +121,8 @@ symptom, and the audit suite is updated so the same class of bug cannot
 recur. Commit messages prefix with `Rectify:` for traceability; the count of
 `Rectify:` commits is reported in `docs/developer/contributing.md`.
 
-## Total: 95
+## Total: 109
 
-3 (Tier 1) + 92 (`skills_extended/`) = 95 bundled skills. The total is
+3 (Tier 1) + 106 (`skills_extended/`) = 109 bundled skills. The total is
 verified by `tests/docs/test_doc_counts.py` against a filesystem walk so any
 addition or removal is caught immediately.
diff --git a/docs/skills/visibility.md b/docs/skills/visibility.md
@@ -2,7 +2,7 @@
 
 ## Overview
 
-AutoSkillit's 95 bundled skills are organized into three tiers that control when and where
+AutoSkillit's 108 bundled skills are organized into three tiers that control when and where
 they appear as slash commands. The tier system is orthogonal to subset categories — you can
 disable a subset across all tiers simultaneously, or reclassify individual skills between
 tiers. See [Subset Categories](subsets.md) for subset configuration.

diff --git a/pyproject.toml b/pyproject.toml
@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "autoskillit"
-version = "0.7.84"
+version = "0.8.0"
 description = "MCP server for orchestrating automated skill-driven workflows with Claude Code"
 readme = "README.md"
 license = {text = "MIT"}
@@ -29,6 +29,7 @@ dependencies = [
     "fastmcp>=3.2.0,<4.0",
     "httpx>=0.28,<0.29",
     "igraph>=1.0,<2.0",
+    "markdown-it-py>=3.0",
     "packaging>=23.0",
     "psutil>=7.2.0",
     "pygments>=2.20.0",

diff --git a/scripts/check_doc_counts.py b/scripts/check_doc_counts.py
@@ -22,13 +22,21 @@
 
 
 def count_skills() -> int:
-    """Count directories in skills/ and skills_extended/ that contain a SKILL.md."""
+    """Count public skills in skills/ and skills_extended/.
+
+    Only counts skills whose SKILL.md begins with YAML frontmatter (``---``),
+    matching the behaviour of DefaultSkillResolver().list_all() which excludes
+    internal bootstrap documents such as sous-chef.
+    """
     count = 0
     for skills_dir in (SKILLS_DIR, SKILLS_EXTENDED_DIR):
         if skills_dir.is_dir():
-            count += sum(
-                1 for d in skills_dir.iterdir() if d.is_dir() and (d / "SKILL.md").exists()
-            )
+            for d in skills_dir.iterdir():
+                skill_md = d / "SKILL.md"
+                if d.is_dir() and skill_md.exists():
+                    text = skill_md.read_text(encoding="utf-8")
+                    if text.startswith("---\n"):
+                        count += 1
     return count
 
 

diff --git a/src/autoskillit/.claude-plugin/plugin.json b/src/autoskillit/.claude-plugin/plugin.json
@@ -1,5 +1,5 @@
 {
   "name": "autoskillit",
-  "version": "0.7.84",
+  "version": "0.8.0",
   "description": "Orchestrated skill-driven workflows using Claude Code headless sessions"
 }
diff --git a/src/autoskillit/assets/mermaid/LICENSE.mermaid b/src/autoskillit/assets/mermaid/LICENSE.mermaid
@@ -0,0 +1,21 @@
+mermaid.js — https://mermaid.js.org
+Copyright (c) 2014-2025 Knut Sveidqvist and contributors
+License: MIT (https://opensource.org/licenses/MIT)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/src/autoskillit/assets/mermaid/VERSION b/src/autoskillit/assets/mermaid/VERSION
@@ -0,0 +1 @@
+11.14.0
diff --git a/src/autoskillit/assets/mermaid/mermaid.min.js b/src/autoskillit/assets/mermaid/mermaid.min.js
diff --git a/src/autoskillit/config/defaults.yaml b/src/autoskillit/config/defaults.yaml
@@ -198,8 +198,22 @@ skills:
     - plan-experiment
     - implement-experiment
     - run-experiment
-    - write-report
+    - generate-report
     - validate-audit
+    - vis-lens-always-on
+    - vis-lens-antipattern
+    - vis-lens-chart-select
+    - vis-lens-color-access
+    - vis-lens-domain-norms
+    - vis-lens-figure-table
+    - vis-lens-multi-compare
+    - vis-lens-temporal
+    - vis-lens-uncertainty
+    - vis-lens-caption-annot
+    - vis-lens-reproducibility
+    - vis-lens-story-arc
+    - plan-visualization
+    - bundle-local-report
   tier3:
     - prepare-pr
     - compose-pr

diff --git a/src/autoskillit/core/_type_constants.py b/src/autoskillit/core/_type_constants.py
@@ -169,6 +169,7 @@ class PackDef(NamedTuple):
     "audit": PackDef(True, "Codebase audit skills"),
     "research": PackDef(False, "Research recipe and experiment skills"),
     "exp-lens": PackDef(False, "Experimental design audit lenses"),
+    "vis-lens": PackDef(False, "Visualization planning lenses"),
 }
 
 CATEGORY_TAGS: frozenset[str] = frozenset(PACK_REGISTRY.keys())

diff --git a/src/autoskillit/migrations/0.7.77-to-0.8.0.yaml b/src/autoskillit/migrations/0.7.77-to-0.8.0.yaml
@@ -0,0 +1,125 @@
+from_version: "0.7.77"
+to_version: "0.8.0"
+description: >
+  Research recipe: write-report renamed to generate-report, output_mode ingredient added
+  (default 'local'), bundle lifecycle refactored, plan-visualization step and vis-lens
+  pack requirement added.
+changes:
+  - id: research-rename-write-report
+    description: "write-report skill and steps renamed to generate-report"
+    instruction: |
+      In any research.yaml recipe:
+      1. Rename step 'write_report' → 'generate_report'.
+      2. Rename step 'write_report_inconclusive' → 'generate_report_inconclusive'.
+      3. Rename step 're_write_report' → 're_generate_report'.
+      4. In each step's skill_command, replace '/autoskillit:write-report' with
+         '/autoskillit:generate-report'.
+      5. Update all on_success, on_failure, on_result routes that reference the old names.
+    example_before: |
+      write_report:
+        tool: run_skill
+        with:
+          skill_command: "/autoskillit:write-report ${{ context.worktree_path }} ..."
+        on_success: test
+      retest:
+        on_failure: re_write_report
+    example_after: |
+      generate_report:
+        tool: run_skill
+        with:
+          skill_command: "/autoskillit:generate-report ${{ context.worktree_path }} ..."
+        on_success: test
+      retest:
+        on_failure: re_generate_report
+
+  - id: research-add-output-mode-ingredient
+    description: "Add output_mode ingredient (default 'local') — BREAKING: old default was implicit pr"
+    instruction: |
+      In any research.yaml recipe, add to the ingredients block (after audit_claims):
+
+        output_mode:
+          required: false
+          default: "local"
+          description: >
+            Where research outputs land. 'local' (default) produces a self-contained
+            report.html bundle under {source_dir}/research-bundles/{date-slug}/ with
+            no GitHub interaction. 'pr' opens a GitHub PR alongside the HTML bundle
+            and runs the full review/archival flow.
+
+      BREAKING CHANGE: Prior to 0.8.0 the recipe always opened a GitHub PR. The new
+      default is 'local'. To restore previous behavior, set output_mode: "pr" in your
+      recipe or pass it as an override to open_kitchen.
+    example_before: |
+      ingredients:
+        audit_claims:
+          required: false
+          default: "false"
+    example_after: |
+      ingredients:
+        audit_claims:
+          required: false
+          default: "false"
+        output_mode:
+          required: false
+          default: "local"
+          description: "..."
+
+  - id: research-replace-commit-with-stage-and-finalize
+    description: "commit_research_artifacts replaced by stage_bundle + finalize_bundle lifecycle"
+    instruction: |
+      In any research.yaml recipe:
+      1. Remove the 'commit_research_artifacts' step.
+      2. Add 'stage_bundle' after run_experiment_lenses (idempotent cp, no commit).
+      3. Add 're_stage_bundle' after 're_generate_report'.
+      4. Add 'finalize_bundle' in the archival phase (after re_push_research.on_success).
+         finalize_bundle conditionally renames/commits in pr mode and keeps browsable files
+         intact in local mode.
+      5. Update all route edges that referenced commit_research_artifacts.
+    example_before: |
+      run_experiment_lenses:
+        on_success: commit_research_artifacts
+      commit_research_artifacts:
+        tool: run_cmd
+        with:
+          cmd: "... git commit ..."
+        on_success: push_branch
+    example_after: |
+      run_experiment_lenses:
+        on_success: stage_bundle
+      stage_bundle:
+        tool: run_cmd
+        with:
+          cmd: "... cp (no commit) ..."
+        on_success: route_pr_or_local
+
+  - id: research-add-plan-visualization-step
+    description: "Add plan_visualization step after design review GO verdict"
+    instruction: |
+      In any research.yaml recipe, wire a 'plan_visualization' step between the design
+      review GO verdict and create_worktree. The step invokes
+      '/autoskillit:plan-visualization' and captures visualization_plan_path and
+      report_plan_path.
+    example_before: |
+      revise_design:
+        on_result:
+          - when: "${{ result.verdict }} == GO"
+            route: create_worktree
+    example_after: |
+      revise_design:
+        on_result:
+          - when: "${{ result.verdict }} == GO"
+            route: plan_visualization
+      plan_visualization:
+        tool: run_skill
+        with:
+          skill_command: "/autoskillit:plan-visualization ..."
+        on_success: create_worktree
+
+  - id: research-add-vis-lens-pack-requirement
+    description: "Add vis-lens to requires_packs"
+    instruction: |
+      In any research.yaml recipe, ensure 'vis-lens' appears in the requires_packs list.
+    example_before: |
+      requires_packs: [research, exp-lens]
+    example_after: |
+      requires_packs: [research, exp-lens, vis-lens]