@skill-map/cli 0.35.0 → 0.37.0

package/README.md

@@ -1,6 +1,6 @@
 # skill-map
 
-Map, inspect, and manage collections of interrelated Markdown files, skills, agents, commands, hooks, and notes that compose AI-agent ecosystems (Claude Code, Codex, Gemini, Copilot, docs sites).
+Map, inspect, and manage collections of interrelated Markdown files, skills, agents, commands, hooks, and notes that compose AI-agent ecosystems (Claude Code, Codex, Antigravity, Copilot, docs sites).
 
 **Status**: pre-1.0, active development. Steps 0a–9 are complete (spec, kernel, plugin loader, full CLI surface, plugin author UX). Step 14 (Full Web UI) is in progress with sub-steps 14.1–14.4 closed (Hono BFF + REST + WebSocket broadcaster + reactive UI); 14.5–14.7 (polish + bundle budgets + responsive scope) still pending. The full deterministic scan, check, history, orphans, plugin authoring, and `sm serve` are live; the optional LLM layer (Phase B / `v0.8.0`) lands after Step 14 closes. See [`ROADMAP.md`](../ROADMAP.md) for the canonical completeness marker and full execution plan. Releases follow the standard changeset flow.
 

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

@@ -116,7 +116,7 @@ must internalise before talking to the tester:
     styled left bar): emit tester-facing messages with `> ` on
     every line, including blank lines inside a multi-paragraph
     block.
-  - `provider != claude` (Gemini CLI, agent-skills, any other
+  - `provider != claude` (Antigravity CLI, agent-skills, any other
     host, most non-Claude renderers show `>` as a literal
     character): emit **plain prose**, NO `> ` prefix anywhere.
   Sample messages in this SKILL are written in the Claude variant
@@ -191,8 +191,7 @@ Same logic as `sm-tutorial`'s §Provider detection. Recap:
 | Provider       | Base dir              | Kinds claimed                | Env-var signal                                  |
 |----------------|-----------------------|------------------------------|-------------------------------------------------|
 | `claude`       | `.claude/`            | `agent`, `command`, `skill`  | `CLAUDECODE=1` OR `AI_AGENT` starts with `claude-code` |
-| `gemini`       | `.gemini/`            | `agent`, `skill`             | `GEMINI_CLI=1` OR `AI_AGENT` starts with `gemini` |
-| `agent-skills` | `.agents/skills/`     | `skill` only                 | no formal env yet, opt-in if the tester asks   |
+| `agent-skills` | `.agents/skills/`     | `skill` only (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 asks   |
 
 **During pre-flight**, inspect the env, pick the provider, and
 persist it into `master-state.yml.master.provider`. Fallback to
@@ -203,14 +202,13 @@ host-dependent rendering rule).
 **Global substitution rule**: wherever this file (or any tour
 file under `references/tour-*.md`) says `.claude/<…>`, swap it
 for the detected `<provider_dir>`. Skip any fixture file or step
-whose kind is not in the provider's supported set (`gemini`: skip
-the `master-command`-style stub if a tour references one;
-`agent-skills`: only the skill + the markdown note are valid).
+whose kind is not in the provider's supported set (`agent-skills`
+/ Antigravity: only the skill + the markdown note are valid).
 
 **Reality check (don't mention)**: this skill ships at
 `.claude/skills/sm-master/`, so in practice Claude Code is the
 only host today. The detection wiring is here so mirrored skills
-in `.gemini/skills/` / `.agents/skills/` reuse it as-is.
+in `.agents/skills/` reuse it as-is.
 
 ## Pre-flight
 
@@ -344,8 +342,8 @@ nodes are enough. Read `references/fixture-templates.md` for the
 verbatim layout and file contents, then write each file to the cwd
 under the detected `<provider_dir>` (per §Provider detection).
 **Skip files whose kind is not in the provider's supported set**:
-on `gemini` keep agent + skill + note; on `agent-skills` keep only
-skill + note (no agent kind there). Translate the natural-language
+on `agent-skills` / Antigravity keep only skill + note (no agent
+kind there). Translate the natural-language
 prose to the tester's language; keep paths, frontmatter keys,
 identifiers, and link targets in English.
 
@@ -392,7 +390,7 @@ Read the `## State YAML` block at the bottom of
 `references/fixture-templates.md` and write it to
 `<cwd>/master-state.yml`. Substitute the four placeholders:
 `<ISO-8601 now>`, `<output of pwd>`, `<output of sm version>`,
-and the resolved `provider` (`claude` / `gemini` / `agent-skills`).
+and the resolved `provider` (`claude` / `agent-skills` / `antigravity`).
 
 ## Menu
 
@@ -644,8 +642,8 @@ anything**:
 
 2. If the cwd matches, read `master.provider` from the yaml and
    use it to compute `<provider_dir>` plus the subset of files
-   actually created (gemini and agent-skills drop some). Show the
-   resolved list to the tester and ask for the literal
+   actually created (agent-skills / Antigravity drop some). Show
+   the resolved list to the tester and ask for the literal
    `yes, wipe` confirmation:
 
    > Start over will delete these paths from `<cwd>`:
@@ -655,8 +653,8 @@ anything**:
    > findings.md
    > .skillmapignore
    > .skill-map/
-   > <provider_dir>/agents/master-agent.md       (claude, gemini)
-   > <provider_dir>/skills/master-skill/         (all three)
+   > <provider_dir>/agents/master-agent.md       (claude only)
+   > <provider_dir>/skills/master-skill/         (both providers)
    > .skill-map/plugins/                         (if any tour created some)
    > notes/ideas.md
    > ```

package/dist/cli/tutorial/sm-master/references/fixture-templates.md

@@ -7,21 +7,22 @@ cwd at boot, plus the initial `master-state.yml` template.
 ## Fixture layout (per provider)
 
 Per §Provider detection in `SKILL.md`, the `<provider_dir>`
-placeholder resolves to `.claude/`, `.gemini/`, or
-`.agents/skills/` depending on the detected runtime. Drop any
-file whose kind is not in the provider's supported set: on
-`gemini` the agent + skill + note are valid; on `agent-skills`
-only the skill + note are valid; on `claude` (default) all
-three apply.
+placeholder resolves to `.claude/` or `.agents/skills/` depending
+on the detected runtime (Google's Antigravity CLI, which replaced
+Gemini CLI on 2026-05-19, adopted the same open standard as
+`agent-skills`, so both share the `.agents/skills/` layout). Drop
+any file whose kind is not in the provider's supported set: on
+`agent-skills` / Antigravity only the skill + note are valid;
+on `claude` (default) all three apply.
 
 Canonical layout (substitute `<provider_dir>` per detection):
 
 ```
 <cwd>/
 ├── <provider_dir>/
-│   ├── agents/                    (claude, gemini)
+│   ├── agents/                    (claude only)
 │   │   └── master-agent.md
-│   └── skills/                    (all three)
+│   └── skills/                    (both providers)
 │       └── master-skill/
 │           └── SKILL.md
 ├── notes/
@@ -138,7 +139,7 @@ master:
   started_at: "<ISO-8601 now>"
   cwd: "<output of pwd>"
   sm_version: "<output of sm version>"
-  provider: "<claude | gemini | agent-skills>"   # filled from §Provider detection
+  provider: "<claude | agent-skills | antigravity>"   # filled from §Provider detection
 tours:
   # Add a new tour by appending another entry here, mirroring the
   # shape below: status, estimated_min, and a steps[] array whose

package/dist/cli/tutorial/sm-master/references/tour-plugins.md

@@ -70,7 +70,7 @@ Mark `tour-1-intro: done`.
 >    Decides what kind each `.md` file is. The `claude` provider,
 >    for instance, walks `.claude/` and types each file it finds
 >    (agent, command, or skill).
->    Examples: `claude`, `gemini`, `agent-skills`.
+>    Examples: `claude`, `antigravity`, `openai`, `agent-skills`.
 >
 > 🔍 **extractor**
 >    Reads a node's body and emits structured findings (links,
@@ -176,8 +176,8 @@ Mark `tour-3-explore: done`.
 > All set. You now know:
 >
 > - What plugins, extensions, bundles, and the six kinds are.
-> - Four bundles ship pre-installed (`claude`, `gemini`,
->   `agent-skills`, `core`).
+> - Five bundles ship pre-installed (`claude`, `antigravity`,
+>   `openai`, `agent-skills`, `core`).
 > - How to list, inspect, diagnose, and toggle extensions from
 >   the CLI (and the same lives in the UI).
 >

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

@@ -112,7 +112,7 @@ optional second phase (~20-30 min) covering the rest of the CLI.
     a styled left bar): emit tester-facing messages with `> `
     prefix on every line, including blank lines inside a
     multi-paragraph block (the standard Markdown blockquote shape).
-  - `provider != claude` (Gemini CLI, agent-skills, any other
+  - `provider != claude` (Antigravity CLI, agent-skills, any other
     host: most non-Claude renderers show `>` as a literal
     character and the visual styling is lost): emit **plain
     prose**, NO `> ` prefix anywhere.
@@ -214,9 +214,9 @@ its own 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` |
-| `gemini`       | `.gemini/`            | `agent`, `skill`               | `GEMINI_CLI=1` OR `AI_AGENT` starts with `gemini` (any equivalent vendor var) |
+| `antigravity`  | none (metadata-only)  | none — Antigravity adopted the open standard, skills land under `.agents/skills/` and route through `agent-skills` | no formal env detection yet; Antigravity CLI replaced Gemini CLI on 2026-05-19 and reuses the open-standard layout, so detection collapses into the `agent-skills` row |
 | `openai`       | `.codex/`             | `agent` (`.codex/agents/*.toml`) | no formal env detection yet; the OpenAI Codex CLI does not host this tutorial today (this SKILL.md lives under `.claude/skills/`), so the row is informational. Add when `.codex/skills/<name>/SKILL.md` mirroring lands. |
-| `agent-skills` | `.agents/skills/`     | `skill` only (vendor-neutral)  | no formal env yet; treat as opt-in if the tester says so |
+| `agent-skills` | `.agents/skills/`     | `skill` only (vendor-neutral, also the on-disk home for Antigravity skills) | no formal env yet; treat as opt-in if the tester says so |
 
 **Decision logic, applied silently during pre-flight**:
 
@@ -224,24 +224,29 @@ its own on-disk convention:
 2. If a Claude-flavoured var is present → `provider = claude`,
    `<provider_dir> = .claude`, supported kinds = `{agent, command,
    skill}`.
-3. Else if a Gemini-flavoured var is present → `provider = gemini`,
-   `<provider_dir> = .gemini`, supported kinds = `{agent, skill}`.
+3. Else if the tester says they use Antigravity OR agent-skills
+   (no env var, opt-in) → `provider = agent-skills`,
+   `<provider_dir> = .agents`, supported kinds = `{skill}`. The
+   `antigravity` lens is the same fixture (Google adopted the open
+   standard); offer it as a manual `sm config set activeProvider
+   antigravity` after the fixture is created if the tester wants
+   Antigravity-flavoured lens identity in the UI.
 4. Else → **fallback to claude** AND surface one short message
    to the tester so they can correct course (render with `> ` if
    the fallback turns out to actually be Claude, plain prose if
-   they correct you to Gemini or agent-skills):
+   they correct you to agent-skills / Antigravity):
 
    > 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 Gemini or agent-skills, tell me and I
-   > swap the fixture to `.gemini/` or `.agents/skills/`.
+   > If you actually use Antigravity or agent-skills, tell me and
+   > I swap the fixture to `.agents/skills/`.
 
 **Reality check (do not mention to the tester unless asked)**:
 this SKILL.md lives at `.claude/skills/sm-tutorial/SKILL.md`, so
 in practice only Claude Code loads it today. The detection logic
 is wired so that the day mirrored skills land at
-`.gemini/skills/sm-tutorial/` and `.agents/skills/sm-tutorial/`,
-they reuse this same body and the fixture follows automatically.
+`.agents/skills/sm-tutorial/`, they reuse this same body and the
+fixture follows automatically.
 
 ### Global substitution rule
 
@@ -251,11 +256,9 @@ because that is the 100% case today. **Wherever you see
 the fixture, when showing the tester commands, when computing the
 expected node count, and when listing files for the start-over
 wipe.** Also: **skip any sub-step whose kind is not in the
-provider's supported set** (e.g. on `gemini`, skip the
-`demo-command` file in Step 3 and the `notes/todo → demo-command`
-connector in Step 5; on `agent-skills`, skip both `demo-agent`
-and `demo-command` and demo only the skill + the two markdown
-notes plus the connectors that target them).
+provider's supported set** (e.g. on `agent-skills` / Antigravity,
+skip both `demo-agent` and `demo-command` and demo only the skill
+plus the two markdown notes plus the connectors that target them).
 
 Persist `provider` into `tutorial-state.yml` (top-level
 `provider: <id>` field) so a resumed session does not have to
@@ -551,7 +554,7 @@ tutorial:
   started_at: "<ISO-8601 now>"
   cwd: "<output of pwd>"
   sm_version: "<output of sm version>"
-  provider: "<claude | gemini | agent-skills>"   # filled from §Provider detection
+  provider: "<claude | agent-skills | antigravity>"   # filled from §Provider detection
 tester:
   level: 2   # default; only asked if they advance into the deep-dive
 route:
@@ -763,9 +766,9 @@ hub in Step 5 will point at via a real `references` link.
 Create these four files (with `Write`), exactly in this order.
 Per §Provider detection, **substitute `.claude/` with the
 detected `<provider_dir>` and skip files whose kind is not in the
-provider's supported set** (`gemini`: skip `demo-command`, three
-new nodes land; `agent-skills`: skip both `demo-agent` and
-`demo-command`, only the skill + the two markdown notes remain).
+provider's supported set** (`agent-skills` / Antigravity: skip
+both `demo-agent` and `demo-command`, only the skill + the two
+markdown notes remain).
 Adjust the node count, the "four new nodes" message, and the file
 list shown to the tester in the sample below accordingly:
 
@@ -930,10 +933,9 @@ because both the command and the skill are addressed by slash).
 Apply with `Edit` on `notes/todo.md` (do not rewrite the file).
 Per §Provider detection, **substitute `.claude/` with the detected
 `<provider_dir>` and drop any bullet whose target node was not
-created in Step 3** (on `gemini` there is no command → skip the
-`/demo-command` bullet, three connectors land; on `agent-skills`
-there is no agent and no command → skip the `@demo-agent` and
-`/demo-command` bullets, two connectors land).
+created in Step 3** (on `agent-skills` / Antigravity there is no
+agent and no command → skip the `@demo-agent` and `/demo-command`
+bullets, two connectors land).
 
 **Edit `notes/todo.md`**: append these bullets after the
 `# Pending` heading:
@@ -965,11 +967,21 @@ Tell the tester:
 >
 > Fijate también que los conectores tienen distinta transparencia.
 > Skill-map estima qué tan seguro está de cada conexión: un
-> `[text](file.md)` que apunta a un archivo concreto (95% de
-> confianza) se ve sólido, mientras que un `@handle` suelto sin
-> extensión (50%, ambiguo) se ve translúcido. La opacidad cuenta
-> esa historia de un vistazo: cuanto más sólido, más confiable es
-> la inferencia.
+> `[text](file.md)` que apunta a un archivo concreto (1.00 de
+> confianza, ahora que el target existe) se ve sólido, mientras que
+> un `@handle` que no resuelve a ningún nodo se queda en 0.5
+> (ambiguo) y se ve translúcido. La opacidad cuenta esa historia de
+> un vistazo: cuanto más sólido, más confiable es la inferencia, y
+> el badge numérico arriba del conector lo dice explícito.
+>
+> 🆕 Mirá también las cards de los nodos: ahora cada nodo muestra
+> dos pequeñas chips abajo a la izquierda, ↑ y ↓, que cuentan los
+> links entrantes y salientes. `notes/todo` ahora marca 0↑ / 4↓
+> (es el hub que apunta a cuatro nodos), y los cuatro nodos
+> apuntados muestran 1↑ cada uno. Pasale el mouse a una chip y se
+> abre un tooltip con el desglose por kind (`mentions`, `invokes`,
+> `references`). Es la misma info que el grafo, resumida en la
+> card por si trabajás desde la list view.
 >
 > Confirm. If a connector is missing, refresh the browser and tell
 > me.
@@ -1094,21 +1106,39 @@ Mark `6-live-ignore: done`.
 > 🔀 **Multi-provider**: this demo ran with the
 > `<provider>` provider (base dir `<provider_dir>`). Skill-map
 > walks three other built-in conventions with identical mechanics:
-> Gemini lives under `.gemini/` (kinds: agent + skill), OpenAI
-> Codex lives under `.codex/agents/*.toml` (TOML sub-agents), and
-> the open agent-skills standard lives under `.agents/skills/`
-> (kind: skill). Drop a `.md` (or `.toml` for Codex) in any of
-> those and the same watcher picks it up, the same connectors
-> light up, the same rules run.
+> the open agent-skills standard (also used by Google's Antigravity
+> CLI, which retired Gemini CLI in May 2026) lives under
+> `.agents/skills/` (kind: skill), OpenAI Codex lives under
+> `.codex/agents/*.toml` (TOML sub-agents), and the `antigravity`
+> lens is metadata-only (no kinds of its own; selecting it just
+> tags the project as Antigravity-flavoured). Drop a `.md` (or
+> `.toml` for Codex) in any of those and the same watcher picks it
+> up, the same connectors light up, the same rules run.
 >
 > 💡 **Tip avanzado (no toques si no querés rescanear)**: en
 > Settings → Project hay un dropdown **Active provider** que
-> selecciona qué lente aplica a TODO el grafo (Claude / Gemini /
-> Codex / Cursor). Cambiarlo limpia el scan persistido y rearma
-> el grafo desde cero bajo el nuevo lente, así que sirve cuando
-> tu proyecto migra de runtime. Por default skill-map autodetecta
-> el lente correcto al primer scan (en este caso, `claude`
-> porque hay `.claude/`).
+> selecciona qué lente aplica a TODO el grafo (Claude / Antigravity
+> / Codex / Cursor / agent-skills). Cambiarlo limpia el scan
+> persistido y rearma el grafo desde cero bajo el nuevo lente, así
+> que sirve cuando tu proyecto migra de runtime. Por default
+> skill-map autodetecta el lente correcto al primer scan (en este
+> caso, `claude` porque hay `.claude/`); si tu repo tiene markers
+> ambiguos (`.claude/` + `.codex/` al mismo tiempo) `sm scan` te
+> pregunta cuál usar, o aborta con exit 2 si pasás `--yes` y todavía
+> está ambiguo. El lente `antigravity` no se auto-detecta (Google
+> adoptó el open standard sin marker propio); seleccionarlo es
+> manual.
+>
+> 🛡️ **Red de seguridad**: si en algún momento desactivás el bundle
+> al que apunta el lente activo (por ejemplo `sm plugins disable
+> claude` con `activeProvider: claude`), el próximo `sm scan` te
+> avisa con un warning explícito ("the active lens points at a
+> disabled bundle") y te ofrece los dos arreglos: reactivar el
+> bundle o cambiar el lente. Lo mismo si la DB local quedó vieja
+> respecto del binario, el open de SQLite detecta el skew de
+> versiones y te dice qué correr para refrescarla. Tip: si ves uno
+> de esos warnings durante el tutorial, NO los ignores, son la
+> señal de que el grafo no refleja lo que vos esperás.
 >
 > If you want, **we can keep going deeper**: I'll walk you through
 > the CLI verbs and flags (`list`, `graph`, `export`, `orphans`,
@@ -1228,6 +1258,17 @@ We'll plant one and watch it surface, that's the easiest way to
 internalise that it is an **issue** on a node, NOT a graph
 connector and NOT the same thing as an "orphan".
 
+> ℹ️ `broken-ref` is one of ~16 built-in rules. Others surface
+> different families: `core/reserved-name` (a file shadows a vendor
+> built-in like `/help`), `core/self-loop` (a node links to itself),
+> `core/redundant-target-reference` (two surfaces in the same body
+> point at the same target), `core/signal-collision` (two extractors
+> detected the SAME byte range with different interpretations, the
+> resolver picked one and the warning explains who lost and why).
+> Same `sm check --analyzers <id>` pattern works for any of them.
+> We will not plant fixtures for the rest, the broken-ref demo
+> covers the mechanics.
+
 Ask the tester to **append one bullet** to `notes/todo.md`:
 
 ```markdown
@@ -1399,6 +1440,28 @@ the id to `disable` / `enable`. Per-extension toggles only work on
 extension-granularity bundles like `core`; the `claude` bundle is
 bundle-granularity and only accepts the bundle id.
 
+**Multiple ids in one call**: both verbs accept any number of ids
+in a single invocation, e.g. `sm plugins disable antigravity openai
+agent-skills` or `sm plugins enable claude core/external-url-counter`.
+Batches are all-or-nothing: if any id is unknown or
+granularity-mismatched the entire call aborts before any
+`config_plugins` write, so the user never lands in a partial state.
+Repeated ids are deduped; locked plugins inside a batch are
+silently skipped (matching `--all` semantics).
+
+### Reserved names (e.g. `commands/help.md`)
+
+If the tester ever names a file after a built-in (`/help`,
+`/clear`, `/init`, `/agents`, `/model`, or one of the documented
+agent reservations like `general-purpose`), `sm check` surfaces a
+`reserved-name` warning. The vendor runtime ignores user-owned
+files that shadow its built-ins, so the warning is not a bug,
+it's skill-map telling the operator "Claude will never invoke this
+file; pick another name". Incoming links to the shadowed file
+resolve at confidence `0.1` instead of `1.0`, so the graph also
+visually de-emphasises them. Rename the file and the warning
+clears on the next scan.
+
 ### `sm sidecar annotate` vs `sm bump` vs `sm sidecar refresh`
 
 - `sm sidecar annotate` is the scaffold verb (creates a fresh
@@ -1432,7 +1495,7 @@ sm tutorial master
 
 > That drops `sm-master.md` in the cwd. Then load it from your
 > agent (e.g. `ejecutá @sm-master.md` in Claude Code, or the
-> equivalent `@`-mention in Gemini CLI) and the deep-dive starts.
+> equivalent `@`-mention in Antigravity CLI) and the deep-dive starts.
 >
 > To delete everything THIS tutorial left behind, if the cwd was a
 > dedicated dir:
@@ -1472,9 +1535,9 @@ anything**:
 
 2. If the cwd matches, read `tutorial.provider` from the yaml and
    use it to compute `<provider_dir>` (and the subset of files
-   actually present, since gemini and agent-skills skip some). Then
-   show the tester the exact list of paths you'll delete and ask
-   for an explicit typed confirmation:
+   actually present, since agent-skills / Antigravity skip some).
+   Then show the tester the exact list of paths you'll delete and
+   ask for an explicit typed confirmation:
 
    > Start over will delete these paths from `<cwd>`:
    >
@@ -1483,9 +1546,9 @@ anything**:
    > findings.md
    > .skillmapignore
    > .skill-map/
-   > <provider_dir>/agents/demo-agent.md          (claude, gemini)
+   > <provider_dir>/agents/demo-agent.md          (claude only)
    > <provider_dir>/commands/demo-command.md      (claude only)
-   > <provider_dir>/skills/demo-skill/            (all three)
+   > <provider_dir>/skills/demo-skill/            (both providers)
    > notes/todo.md
    > notes/demo-guideline.md
    > notes/private-credentials.md