@skill-map/cli 0.61.0 → 0.61.2

package/dist/cli/tutorial/sm-tutorial/SKILL.md

@@ -161,11 +161,19 @@ cd <cwd>
 > Got the second terminal open and anchored to the folder? Confirm
 > before we move on.
 
+**HARD STOP here.** This message ends at the confirmation question
+above, that is its own beat. Do NOT run step 5, do NOT render the
+menu, do NOT add a "meanwhile / mientras tanto" bridge in the same
+message. Wait for the tester to confirm the second terminal is open
+and anchored; only then continue to step 5.
+
 ### 5. Initialise state, lay the universal files, show the menu
 
-Pre-flight does NOT pre-lay any part's fixture and does NOT auto-enter
-a part. It initialises state, lays the universal files every part
-needs, then routes to the menu. All silent (backstage):
+**Run this only after the tester has confirmed step 4.** Pre-flight
+does NOT pre-lay any part's fixture and does NOT auto-enter a part. It
+initialises state, lays the universal files every part needs, then
+routes to the menu. The init + lay are silent (backstage); the menu is
+the first thing the tester sees in this post-confirmation message:
 
 - Create the state file (carries the detected provider, the running
   `sm version`, the cwd, and the tester's language):
@@ -280,8 +288,12 @@ the `__PROVIDER__` token and skip kinds the provider does not claim.
      `fixtures.js clear portfolio --provider <provider>`, `rm -rf .skill-map`.
   2. `fixtures.js seed prologue-built --provider <provider> --lang <lang>`
      (lays the six demo nodes, wires the hub, drops `private-credentials`).
+     Run this ALWAYS, even if a demo fixture from a prior prologue run is
+     on disk, so the deliberate broken reference is the pristine
+     `prologue-built` state, not whatever the tester edited in the
+     prologue (they resolve it by hand in `edit-link`).
   3. `sm init` (single `.claude/` marker, no lens prompt), then `sm scan`.
-     If the demo was already on disk and `.skill-map/` exists, just `sm scan`.
+     If `.skill-map/` already exists, skip the init and just `sm scan`.
 
 - **`seed`** (campaign parts `connect-harness`, `daily-loop`): builds
   on the accumulating portfolio, but the tester may have jumped here.

package/dist/cli/tutorial/sm-tutorial/fixtures-data/manifest.json

@@ -23,12 +23,10 @@
       "public/index.html",
       "public/about.html",
       "public/projects.html",
-      "public/posts.html",
       "public/style.css",
       "docs/STYLE.md",
       "docs/DEPLOY.md",
       "docs/DEPLOYMENT.md",
-      "docs/draft.md",
       "__PROVIDER__/agents/content-editor.md",
       "__PROVIDER__/skills/check-links",
       "__PROVIDER__/commands/publish.md",

package/dist/cli/tutorial/sm-tutorial/fixtures-data/sets/prologue/en/__PROVIDER__/agents/demo-agent.md

@@ -4,8 +4,6 @@ description: |
   Example agent that handles read and shell tasks. Solo node at
   boot; gets connected to the rest of the demo fixture during the
   Live UI step.
-tools: [Read, Bash]
-model: sonnet
 ---
 
 # demo-agent

package/dist/cli/tutorial/sm-tutorial/fixtures-data/sets/prologue/es/__PROVIDER__/agents/demo-agent.md

@@ -4,8 +4,6 @@ description: |
   Agente de ejemplo que maneja tareas de lectura y de shell. Nodo
   suelto al arranque; se conecta con el resto del set de prueba
   durante el paso de la UI en vivo.
-tools: [Read, Bash]
-model: sonnet
 ---
 
 # demo-agent

package/dist/cli/tutorial/sm-tutorial/references/_core.md

@@ -131,14 +131,16 @@ element. Decide with §Provider detection:
 
 - `provider == claude` (Claude Code, blockquotes render as a styled
   left bar): emit tester-facing messages with `> ` on every line,
-  including blank lines inside a multi-paragraph block.
-- `provider != claude` (Antigravity CLI, agent-skills, any other
-  host: most non-Claude renderers show `>` as a literal character):
-  emit **plain prose**, NO `> ` prefix anywhere.
+  including blank lines inside a multi-paragraph block. This is the
+  only active path today.
+- `provider != claude` (coming soon: Antigravity CLI, agent-skills,
+  any other host where most non-Claude renderers show `>` as a
+  literal character): emit **plain prose**, NO `> ` prefix anywhere.
+  Kept as the wiring for the coming-soon providers; not exercised
+  while the tutorial demos `claude` only.
 
 Sample messages throughout the part files are written in the Claude
-variant (with `> `); strip the prefix when the host is non-Claude,
-the wording is unchanged.
+variant (with `> `).
 
 **The rule, in one line**: every tester-facing line of prose carries
 the `> ` bar in the Claude variant, context sentences, intros, tips,
@@ -257,41 +259,41 @@ on-disk convention:
 | Provider       | Base dir          | Kinds it claims             | Detect via env var(s)                                  |
 |----------------|-------------------|-----------------------------|--------------------------------------------------------|
 | `claude`       | `.claude/`        | `agent`, `command`, `skill` | `CLAUDECODE=1` OR `AI_AGENT` starts with `claude-code` |
-| `agent-skills` | `.agents/skills/` | `skill` only (vendor-neutral; also the on-disk home for Google's Antigravity CLI, which replaced the Gemini CLI on 2026-05-19 and adopted this open standard) | no formal env yet; opt-in if the tester says so |
-| `openai`       | `.codex/`         | `agent` (`.codex/agents/*.toml`) | no formal env yet; informational today |
-
-**Decision logic, applied silently during pre-flight**:
-
-1. Inspect the agent's environment (`process.env`).
-2. Claude-flavoured var present → `provider = claude`,
-   `<provider_dir> = .claude`, kinds = `{agent, command, skill}`.
-3. Else if the tester says Antigravity / agent-skills (opt-in) →
-   `provider = agent-skills`, `<provider_dir> = .agents`, kinds =
-   `{skill}`.
-4. Else → **fallback to claude** AND surface one short message so
-   they can correct course (render `> ` if it turns out to be
-   Claude, plain prose if they correct you):
-
-   > Heads up: I couldn't detect which agent runtime is hosting me,
-   > so I'll demo skill-map's Claude provider (`.claude/`). If you
-   > actually use Antigravity or agent-skills, tell me and I swap
-   > the fixture to `.agents/skills/`.
+| `agent-skills` | `.agents/skills/` | `skill` only (vendor-neutral; also the on-disk home for Google's Antigravity CLI, which replaced the Gemini CLI on 2026-05-19 and adopted this open standard) | coming soon (not selectable in the tutorial yet) |
+| `openai`       | `.codex/`         | `agent` (`.codex/agents/*.toml`) | coming soon (not selectable in the tutorial yet) |
+
+**Decision logic, applied silently during pre-flight**: the tutorial
+demonstrates the `claude` provider only. `agent-skills` and `openai`
+are **coming soon** and are not selectable here, so there is no
+runtime to detect or opt into.
+
+1. `provider = claude`, `<provider_dir> = .claude`, kinds =
+   `{agent, command, skill}`. Always.
+2. Do NOT offer Antigravity / agent-skills / openai as an alternative,
+   and do NOT ask the tester which runtime hosts them. If a tester
+   says they use another runtime, acknowledge it briefly and explain
+   that those providers are coming soon, the tutorial demos `claude`
+   (`.claude/`) today:
+
+   > Heads up: skill-map also reads Antigravity, agent-skills and
+   > Codex projects, but those providers are coming soon in this
+   > tutorial. We'll demo skill-map's Claude provider (`.claude/`)
+   > today.
 
 Persist `provider` into the state file (`tutorial.provider`) so a
-resumed session does not re-detect.
+resumed session does not re-detect. (It is always `claude` for now.)
 
 **Global substitution rule**: the fixture scripts do the file-level
 work. You pass `--provider <p>` (the value persisted in
 `tutorial.provider`) and `--lang <l>`, and they resolve the
 `__PROVIDER__` path token, skip files whose kind the provider does
 not claim, and report the adjusted `nodeCount` plus the `skipped`
-list in their summary. Your job is the **narration**: wherever a part
-file's tester-facing prose says `.claude/`, swap it for the detected
-`<provider_dir>`, and narrate the node count from the script summary
-(on `agent-skills` / Antigravity only the skill + markdown notes
-exist, so the count is lower and the agent / command beats fold
-away). The campaign cross-link chapters target `claude` today (see
-the reality check below).
+list in their summary. Today `provider` is always `claude`, so the
+narration uses `.claude/` throughout; the `--provider` plumbing stays
+wired so the coming-soon providers (`agent-skills` / Antigravity,
+`openai`) drop in later without a narrative rewrite. The campaign
+cross-link chapters target `claude` today (see the reality check
+below).
 
 **Reality check (don't mention to the tester)**: this skill ships
 at `.claude/skills/sm-tutorial/`, so Claude Code is the only host
@@ -338,6 +340,11 @@ For every chapter:
    Inviolable rule #6, NO "¿seguimos?"). `pace` only decides batching:
    `per-step` presents one chapter per exchange, `auto-advance` may
    chain chapters that need no tester action into one response.
+   **Either pace still emits every chapter's `Capítulo S.N` Announcement
+   (step 1).** `auto-advance` drops only the inter-chapter "¿seguimos?"
+   pause, never the per-chapter announcement, so any chapter with a
+   tester-facing beat always opens with its number, even when it
+   follows straight on from the previous one.
 
 ## Routing + menu (orchestrator)
 

package/dist/cli/tutorial/sm-tutorial/references/_manifest.json

@@ -69,7 +69,7 @@
         },
         {
           "id": "manual",
-          "title": "The handbook and CLAUDE.md (@AGENTS.md)",
+          "title": "The handbook (AGENTS.md) and CLAUDE.md",
           "est_min": 2
         },
         {
@@ -132,19 +132,14 @@
           "est_min": 3
         },
         {
-          "id": "add-page",
-          "title": "Add a page with your agent",
-          "est_min": 4
-        },
-        {
-          "id": "orphan-draft",
-          "title": "A page nobody links to yet",
+          "id": "preview",
+          "title": "Bring it up and see your site",
           "est_min": 2
         },
         {
-          "id": "wire-and-improve",
-          "title": "Wire the draft in",
-          "est_min": 3
+          "id": "add-page",
+          "title": "Add a page with your agent",
+          "est_min": 4
         },
         {
           "id": "broken-ref",
@@ -162,8 +157,8 @@
           "est_min": 4
         },
         {
-          "id": "sidecar",
-          "title": "Annotate the handbook (.sm and consent)",
+          "id": "stability",
+          "title": "Set a node's stability (and the `.sm` sidecar)",
           "est_min": 3
         },
         {

package/dist/cli/tutorial/sm-tutorial/references/_manifest.yml

@@ -113,7 +113,7 @@ parts:
     status: active
     chapters:
       - id: kickoff     ; title: "Start the portfolio (sm init on the real skeleton)" ; est_min: 2
-      - id: manual      ; title: "The handbook and CLAUDE.md (@AGENTS.md)" ; est_min: 2
+      - id: manual      ; title: "The handbook (AGENTS.md) and CLAUDE.md" ; est_min: 2
       - id: first-agent ; title: "The first harness agent (content-editor)" ; est_min: 2
       - id: real-kinds  ; title: "The real kinds in context" ; est_min: 2
 
@@ -144,13 +144,12 @@ parts:
     status: active
     chapters:
       - id: setup            ; title: "Make it yours, make it presentable" ; est_min: 3
+      - id: preview          ; title: "Bring it up and see your site" ; est_min: 2
       - id: add-page         ; title: "Add a page with your agent" ; est_min: 4
-      - id: orphan-draft     ; title: "A page nobody links to yet" ; est_min: 2
-      - id: wire-and-improve ; title: "Wire the draft in" ; est_min: 3
       - id: broken-ref       ; title: "A rename breaks a link" ; est_min: 4
       - id: reserved         ; title: "A reserved name collides" ; est_min: 2
       - id: publish          ; title: "Ship it: run /publish for real" ; est_min: 4
-      - id: sidecar          ; title: "Annotate the handbook (.sm and consent)" ; est_min: 3
+      - id: stability        ; title: "Set a node's stability (and the `.sm` sidecar)" ; est_min: 3
       - id: golive           ; title: "Your portfolio, live next to the graph" ; est_min: 3
 
   # ----- parked: MCP returns later as its own iteration (body kept in part-mcp.md) -----

package/dist/cli/tutorial/sm-tutorial/references/fixtures.md

@@ -70,9 +70,8 @@ entered out of order.
 
 `manifest.json#footprints` lists the full on-disk reach of each
 fixture, INCLUDING files a part's later chapters add (the daily loop's
-`docs/draft.md`, `public/style.css` + generated pages, the renamed
-`new-page` command, `AGENTS.sm`; the portfolio's `DEPLOYMENT.md`
-rename). `fixtures.js clear <footprint>` (part-entry resets) and
+`public/style.css` + generated pages, the renamed `new-page` command,
+`AGENTS.sm`; the portfolio's `DEPLOYMENT.md` rename). `fixtures.js clear <footprint>` (part-entry resets) and
 `state.js wipe` (start-over) both read it, so the per-fixture path list
 lives in ONE place. Add or drop a harness file there.
 

package/dist/cli/tutorial/sm-tutorial/references/part-connect-harness.md

@@ -32,7 +32,7 @@ Wait for confirmation. Mark `check-links`: done.
 
 On `agent-skills` / Antigravity there is no `command` kind, so skip this whole chapter and fold its purpose into the prose of the next one.
 
-Tell the tester to create the file themselves (it is their project's file, Inviolable rule #2). Substitute `<provider_dir>` per `_core.md` in the path you give them. Backstage, get the content: `node .claude/skills/sm-tutorial/scripts/fixtures.js cat harness --file "__PROVIDER__/commands/publish.md" --provider <provider> --lang <lang>`, then render it in the fenced block the tester copies. The frontmatter fence (`---`) MUST sit on column 0 with no leading spaces: present the block below exactly as written, and if the tester pastes it indented, have them strip the leading whitespace. An indented `---` does not parse as YAML, so the `publish` node would land without its `name` or `description`.
+Tell the tester to create the file themselves (it is their project's file, Inviolable rule #2). Substitute `<provider_dir>` per `_core.md` in the path you give them. Backstage, get the content: `node .claude/skills/sm-tutorial/scripts/fixtures.js cat harness --file "__PROVIDER__/commands/publish.md" --provider <provider> --lang <lang>`, then render it as a **top-level fenced code block**: at column 0, NOT inside the `> ` blockquote and with NO leading indentation, so the tester's copy keeps every line flush left. The frontmatter fences (`---`) MUST land on column 0. If the block is rendered (or pasted) indented, the opening and closing `---` shift off column 0, the YAML never parses, and the `publish` node loads body-only without its `name` / `description` (`sm check` then warns `frontmatter-malformed`). Present the block below exactly as written.
 
 > Create `.claude/commands/publish.md` with exactly this content (the first line is `---`, nothing before it):
 
@@ -78,7 +78,7 @@ Continue the tester message:
 >
 > Did the three arrows appear?
 
-Wait for confirmation. You MAY use `Read` on the file afterwards to verify it landed. Mark `publish`: done.
+Wait for confirmation. You MAY use `Read` on the file afterwards to verify it landed, in particular that the opening and closing `---` are flush at column 0. If a later `sm check` flags `frontmatter-malformed` on `publish.md`, the fences got indented on paste: have the tester re-align every line flush left (strip the leading spaces so `---` is at column 0), then the next scan reads it clean. Mark `publish`: done.
 
 ## Chapter `links` - The handbook becomes the hub (~4 min)
 
@@ -122,7 +122,7 @@ Wait for confirmation. You MAY use `Read` on the two files afterwards to verify
 
 ## Chapter `confidence` - How sure is each link (~3 min)
 
-**Context**: skill-map records how sure it is of every connection and shows that as opacity. In this harness every link resolves to a real node, so they all read solid (1.00); the broken-reference case is the one the tester met in the prologue (the bare `@demo-guideline` mention that resolved to no agent, drawn as no arrow and flagged `reference-broken` at 0.50). Here we open the Inspector on a real harness node, read the all-solid numbers, and point back to that prologue contrast. Mirrors the prologue's connectors beat on the portfolio.
+**Context**: skill-map records how sure it is of every connection and shows that as opacity. In this harness every link resolves to a real node, so they all read solid (1.00); the broken-reference case is the one the tester met in the prologue (the `@demo-guideline` reference skill-map could not resolve, drawn as no arrow and flagged `reference-broken` at 0.50, then resolved by hand with `.md`). Here we open the Inspector on a real harness node, read the all-solid numbers, and point back to that prologue contrast. Mirrors the prologue's connectors beat on the portfolio.
 
 No file edits in this chapter, pure observation on the graph the tester just built.
 
@@ -147,14 +147,13 @@ Tell the tester:
 >
 > Your whole harness reads solid because every link lands on a real
 > node, that is what a clean, fully wired graph looks like. So what
-> does a link that does NOT resolve look like? You saw one back in the
-> prologue: `@demo-guideline` was a bare `@`-mention, and a bare
-> `@handle` only resolves to an agent. `demo-guideline` is a note, so
-> it had nothing to land on: skill-map drew no arrow and flagged it as
-> a **broken reference**, its confidence knocked down to **0.50** by
-> the broken penalty. The fix there was one character:
-> `@demo-guideline2.md`, the same handle plus a `.md`, resolved to the
-> real file and drew a solid arrow at **1.00**.
+> does a link that does NOT resolve look like? You met one back in the
+> prologue: `@demo-guideline` was a reference skill-map could not
+> resolve, it had nothing to land on, so skill-map drew no arrow and
+> flagged it as a **broken reference**, its confidence knocked down to
+> **0.50** by the broken penalty. The fix was one character: adding
+> `.md` (`@demo-guideline.md`) turned it into a file reference to the
+> real `demo-guideline.md`, and it drew a solid arrow at **1.00**.
 >
 > So confidence here is really about resolution: **1.00** for a link
 > that lands on a real node, **0.50** for one flagged broken. There is