skillpacks 0.1.0 → 0.1.2

package/AGENTS.md

@@ -1,13 +1,12 @@
-<!-- provision-agentic-config v0.5 -->
+<!-- provision-agentic-config v0.11 -->
 ## Workflow Orchestration
 
 ### 1. Plan Mode Default
 - Enter plan mode for ANY non-trivial task (3+ steps or architectural decisions)
-- If something goes sideways, STOP and re-plan immediately — don't keep pushing
+- If something goes sideways, STOP and re-plan immediately - don't keep pushing
 - Verification is mandatory, but routine no-op verification runs inside the active execution/shipping step. Enter plan mode for non-trivial remediation or new work discovered by verification, not for validation that already has clear commands and no expected source changes.
 - Write detailed specs upfront to reduce ambiguity
 - In Codex: use `update_plan` in Default mode and `request_user_input` only when already in Plan mode
-- Do not assume a Claude-style clear-context-on-accept flow or related JSON setting exists
 
 ### 2. Subagent Strategy
 - Use subagents only when the active Codex tool instructions allow them.
@@ -31,25 +30,19 @@
 ### 5. Demand Elegance (Balanced)
 - For non-trivial changes: pause and ask "is there a more elegant way?"
 - If a fix feels hacky: "Knowing everything I know now, implement the elegant solution"
-- Skip this for simple, obvious fixes — don't over-engineer
+- Skip this for simple, obvious fixes - don't over-engineer
 - Challenge your own work before presenting it
 
 ### 6. Autonomous Bug Fixing
 - When given a bug report: just fix it. Don't ask for hand-holding
-- Point at logs, errors, failing tests — then resolve them
+- Point at logs, errors, failing tests - then resolve them
 - Zero context switching required from the user
 - Go fix failing tests without being told how
 
 ### Missing Skill Fallback
-- When a skill invocation fails because the skill is not found, run `scripts/pack.sh which <skill-name>` to check if the skill exists in an available pack.
-- If found in an uninstalled pack, recommend `$pack install <skill>` for just that skill or `$pack install <pack>` for the full pack, and note the post-install reload path: Claude Code `/reload-skills` first, `/clear` can pick up the refreshed registry, restart if the top-level `.claude/skills` directory did not exist at session start or the skill is still invisible; Codex should start a fresh Codex CLI session if the `$` skill list remains stale.
-- If found in an installed pack, suggest the same reload path to pick up the local skill roots.
-- If not found in any pack, suggest `$skills` or `$skills search <keyword>`.
-
-### Project Pack Command Resolution
 - If a user invokes a command-like skill such as `$benchmark-test-skill design-system` and the leading command is not in the injected session skill list, search project-local packs before falling back to the trailing argument as the active skill.
-- Check `packs/*/codex/<command>/SKILL.md` and pack metadata such as `packs/*/PACK.md`; project-local pack skills may exist in this repository even when they are not visible in the active session list.
-- In this repository, `$benchmark-test-skill` lives under `packs/agentic-skills-bench/codex/benchmark-test-skill/SKILL.md`, and `design-system` is its target skill argument.
+- Check `packs/*/codex/*/SKILL.md` and pack metadata such as `packs/*/PACK.md`; project-local pack skills may exist in this repository even when they are not visible in the active session list.
+- For any missing skill, run `scripts/pack.sh which <skill-name>` to locate the providing pack. If found in an uninstalled pack, recommend `npx skillpacks install <pack-or-skill>` from the project shell for either the skill or the full pack, and note the post-install reload path: Claude Code `/reload-skills` first, `/clear` can pick up the refreshed registry, restart if the top-level `.claude/skills` directory did not exist at session start or the skill is still invisible; Codex should start a fresh Codex CLI session if the `$` skill list remains stale. If found in an installed pack, suggest the same reload path. If not found in any pack, suggest `$skills` or `$skills search <keyword>`.
 
 ### Prompt History
 - On every skill invocation, before substantive work, create `prompts/<skill-slug>/` if it does not exist.
@@ -89,7 +82,7 @@
 - **No Laziness**: Find root causes. No temporary fixes. Senior developer standards.
 - **Minimal Impact**: Changes should only touch what's necessary. Avoid introducing bugs.
 - **Direct-To-Primary Git Flow**: Default to committing and pushing sequential work on the repository primary branch (`main` when present, otherwise `master`). Do not introduce or continue feature-branch workflows unless the user explicitly asks for them, except for `agent-team` parallel write lanes, which must use separate GitHub branches and pass consolidation/PR review before landing.
-- **Always Ship Mutations**: If a task creates or modifies tracked files, finish by committing and pushing all intended changes before stopping unless the user explicitly says not to. Codex `$exec` ships by default (validates, commits, pushes, plans next) — use `$ship` only to package existing work or unpushed commits.
+- **Always Ship Mutations**: If a task creates or modifies tracked files, finish by committing and pushing all intended changes before stopping unless the user explicitly says not to. Do not leave a dirty tracked tree or unpushed commits behind.
 - **No GitHub Actions**: Do not create, modify, or suggest GitHub Actions workflows unless the user explicitly asks for GitHub Actions. This project does not use GitHub Actions for CI/CD by default.
 
 ## Windows/WSL File Opening

package/CLAUDE.md

@@ -1,13 +1,12 @@
-<!-- provision-agentic-config v0.5 -->
+<!-- provision-agentic-config v0.11 -->
 ## Workflow Orchestration
 
 ### 1. Plan Mode Default
 - Enter plan mode for ANY non-trivial task (3+ steps or architectural decisions)
-- If something goes sideways, STOP and re-plan immediately — don't keep pushing
+- If something goes sideways, STOP and re-plan immediately - don't keep pushing
 - Verification is mandatory, but routine no-op verification runs inside the active execution/shipping step. Enter plan mode for non-trivial remediation or new work discovered by verification, not for validation that already has clear commands and no expected source changes.
 - Write detailed specs upfront to reduce ambiguity
 - In Codex: use `update_plan` in Default mode and `request_user_input` only when already in Plan mode
-- Do not assume a Claude-style clear-context-on-accept flow or related JSON setting exists
 
 ### 2. Subagent Strategy
 - Use subagents liberally to keep main context window clean
@@ -20,7 +19,6 @@
 - After ANY correction from the user: update `tasks/lessons.md` with the pattern
 - Write rules for yourself that prevent the same mistake
 - Ruthlessly iterate on these lessons until mistake rate drops
-- For misunderstood-intent corrections, capture: what the user wanted, what the agent interpreted, and which signal was missed
 - Review lessons at session start for relevant project
 
 ### 4. Verification Before Done
@@ -32,18 +30,18 @@
 ### 5. Demand Elegance (Balanced)
 - For non-trivial changes: pause and ask "is there a more elegant way?"
 - If a fix feels hacky: "Knowing everything I know now, implement the elegant solution"
-- Skip this for simple, obvious fixes — don't over-engineer
+- Skip this for simple, obvious fixes - don't over-engineer
 - Challenge your own work before presenting it
 
 ### 6. Autonomous Bug Fixing
 - When given a bug report: just fix it. Don't ask for hand-holding
-- Point at logs, errors, failing tests — then resolve them
+- Point at logs, errors, failing tests - then resolve them
 - Zero context switching required from the user
 - Go fix failing tests without being told how
 
 ### Missing Skill Fallback
 - When a skill invocation fails because the skill is not found, run `scripts/pack.sh which <skill-name>` to check if the skill exists in an available pack.
-- If found in an uninstalled pack, recommend `/pack install <skill>` for just that skill or `/pack install <pack>` for the full pack, and note the post-install reload path: Claude Code `/reload-skills` first, `/clear` can pick up the refreshed registry, restart if the top-level `.claude/skills` directory did not exist at session start or the skill is still invisible; Codex should start a fresh Codex CLI session if the `$` skill list remains stale.
+- If found in an uninstalled pack, recommend `npx skillpacks install <pack-or-skill>` from the project shell for either the skill or the full pack, and note the post-install reload path: Claude Code `/reload-skills` first, `/clear` can pick up the refreshed registry, restart if the top-level `.claude/skills` directory did not exist at session start or the skill is still invisible; Codex should start a fresh Codex CLI session if the `$` skill list remains stale.
 - If found in an installed pack, suggest the same reload path to pick up the local skill roots.
 - If not found in any pack, suggest `/skills` or `/skills search <keyword>`.
 
@@ -51,7 +49,6 @@
 - If a user invokes a command-like skill such as `/benchmark-test-skill design-system` and the leading command is not in the injected session skill list, search project-local packs before falling back to the trailing argument as the active skill.
 - Check `packs/*/claude/<command>/SKILL.md` and pack metadata such as `packs/*/PACK.md`; project-local pack skills may exist in this repository even when they are not visible in the active session list.
 - In this repository, `/benchmark-test-skill` lives under `packs/agentic-skills-bench/claude/benchmark-test-skill/SKILL.md`, and `design-system` is its target skill argument.
-- For any other missing skill, run `scripts/pack.sh which <skill-name>` to locate the providing pack. If found in an uninstalled pack, use the same `/pack install <skill>` or `/pack install <pack>` recommendation and reload guidance above. If found in an installed pack, suggest the same reload path. If not found in any pack, suggest `/skills` or `/skills search <keyword>`.
 
 ### Prompt History
 - On every skill invocation, before substantive work, create `prompts/<skill-slug>/` if it does not exist.
@@ -85,7 +82,7 @@
 - **No Laziness**: Find root causes. No temporary fixes. Senior developer standards.
 - **Minimal Impact**: Changes should only touch what's necessary. Avoid introducing bugs.
 - **Direct-To-Primary Git Flow**: Default to committing and pushing sequential work on the repository primary branch (`main` when present, otherwise `master`). Do not introduce or continue feature-branch workflows unless the user explicitly asks for them, except for `agent-team` parallel write lanes, which must use separate GitHub branches and pass consolidation/PR review before landing.
-- **Always Ship Mutations**: If a task creates or modifies tracked files, finish by committing and pushing all intended changes before stopping unless the user explicitly says not to. Exception (Claude only): `/exec` hands a dirty tracked tree to `/ship`. After shipping, if `tasks/todo.md` has remaining steps, run `/ship` to handle planning and the approval cycle.
+- **Always Ship Mutations**: If a task creates or modifies tracked files, finish by committing and pushing all intended changes before stopping unless the user explicitly says not to. Do not leave a dirty tracked tree or unpushed commits behind.
 - **No GitHub Actions**: Do not create, modify, or suggest GitHub Actions workflows unless the user explicitly asks for GitHub Actions. This project does not use GitHub Actions for CI/CD by default.
 
 ## Windows/WSL File Opening
@@ -127,7 +124,7 @@ The alignment-page convention is authored in `docs/alignment-page-convention.md`
 
 **Visual rendering tiers.** Each alignment page has a rendering tier declared via `visual_tier` in SKILL.md frontmatter: `document` (default — text/tables only), `visual` (adds inline Canvas/SVG charts using `scripts/alignment-chart-snippets.js` as reference), or `prototype` (adds interactive components). The upgrade script injects tier-appropriate guidance into each skill's `ALIGNMENT-PAGE.md`.
 
-**Interview depth.** Skills declare `interview_depth` in SKILL.md frontmatter: `full` (4-phase terminal interview before alignment page), `light` (1-3 context questions + findings validation), or `none` (direct to research). The interview convention is defined in `docs/interview-convention.md`. All interviewing happens in the terminal before the alignment page is built.
+**Interview depth.** Skills declare `interview_depth` in SKILL.md frontmatter: `full` (4-phase terminal interview before alignment page), `light` (1-3 context questions + findings validation), or `none` (direct to research). The interview convention is defined in `docs/interview-convention.md`. All interviewing happens in the terminal before the alignment page is built. Confirmation manifests/checklists render inline as turn-final message text per the Manifest Visibility Rule in `docs/interview-convention.md`.
 
 **Glossary write-forward.** Skills with `type: research` or `type: analysis` automatically receive a glossary additions gate in their `ALIGNMENT-PAGE.md`. The gate renders proposed terms with per-term approve/edit/reject/flag controls. Only approved terms are appended to the target glossary during the confirmed-page write step.
 
@@ -166,4 +163,4 @@ Research-producing skills maintain a shared project glossary at `research/glossa
 
 ### Cross-Pack Routing
 
-When a skill recommends another skill from a different pack, verify the target pack is installed via `.agents/project.json` `enabled_packs`. If not installed, include `/pack install <pack-name>` as a prerequisite in the recommendation.
+When a skill recommends another skill from a different pack, verify the target pack is installed via `.agents/project.json` `enabled_packs`. If not installed, include `npx skillpacks install <pack-name>` as the prerequisite in the recommendation.

package/README.md

@@ -1,19 +1,19 @@
-# Agentic Skills
+# gSkillPacks
 
 A shared skill library for Claude Code and OpenAI Codex.
 
 The library now uses a two-layer model:
 
-- **Global core skills** are domain-neutral and safe to expose in every project.
+- **Base skills** are domain-neutral and safe to expose in every project.
 - **Project-local packs** hold domain-specific workflows such as business apps, video games, and developer tools.
 
 This keeps game research out of B2B SaaS sessions, and keeps business-product assumptions out of game and devtool projects.
 
 ## Prerequisites
 
-- **Node.js 18+** for the `skillpacks` npm CLI and package build.
-- **bash** shell (macOS, Linux, or WSL on Windows) for `./init.sh`, `scripts/pack.sh`, and the remaining shell-backed `skillpacks` commands.
-- **jq** for git-checkout `scripts/pack.sh` write commands and `skillpacks install-deck` materialization: `brew install jq` (macOS) or `apt install jq` (Debian/Ubuntu). Node-owned `skillpacks` project commands do not require `jq`.
+- **Node.js 18+** for the `gskp` npm CLI and package build.
+- **bash** shell (macOS, Linux, or WSL on Windows) for `scripts/pack.sh` and the remaining shell-backed `gskp` commands.
+- **jq** for git-checkout `scripts/pack.sh` write commands and `gskp install-deck` materialization: `brew install jq` (macOS) or `apt install jq` (Debian/Ubuntu). Node-owned `gskp` project commands do not require `jq`.
 - **Claude Code** or **OpenAI Codex** installed on your machine
 - **pnpm** (optional, for running tests): `npm install -g pnpm`
 
@@ -27,11 +27,11 @@ The source-checkout path is available today and remains supported:
 
 ```bash
 git clone <this-repo-url> ~/agentic-skills
-cd ~/agentic-skills
-./init.sh
+cd ~/my-project
+npx skillpacks init
 ```
 
-Project-local packs are installed from the checkout:
+Base skills install **project-local** into the current repository (`.claude/skills/`, `.codex/skills/`). Project-local packs are installed from the checkout:
 
 ```bash
 cd ~/my-project
@@ -42,35 +42,40 @@ cd ~/my-project
 
 ### npm CLI
 
-After the first public package is published, users can install from the current project directory without cloning this repository:
+With the published npm package, users can install from the current project directory without cloning this repository:
 
 ```bash
 cd ~/my-project
+npx skillpacks init
 npx skillpacks install devtool
 npx skillpacks install code-quality
 npx skillpacks install-deck game-afps
+npx skillpacks refresh
 npx skillpacks status
 ```
 
-`npx skillpacks` writes the same project-local files as `scripts/pack.sh`: `.agents/project.json`, `.claude/skills/*`, and `.codex/skills/*`. Package semver such as `skillpacks@0.1.0` selects the bundled skill snapshot; individual skill pins still use each skill's `version:` value through `skillpacks pin <skill> <version>`.
+`npx skillpacks init` installs the base skills for the current repository as project-local roots. `npx skillpacks install <pack-or-skill>` adds domain packs or individual pack skills. Both write project-local files: `.agents/project.json`, `.claude/skills/*`, and `.codex/skills/*`. Package semver such as `skillpacks@0.1.0` selects the bundled skill snapshot; `npx skillpacks refresh` recreates enabled base skills and packs from that snapshot. Individual skill pins still use each skill's `version:` value through `skillpacks pin <skill> <version>` or `gskp pin <skill> <version>`.
+
+The same release is also published as the scoped alias package `@glexcorp/gskp`, so `npx @glexcorp/gskp init` is equivalent to `npx skillpacks init` at the same version. Both packages install both `gskp` and `skillpacks` binaries for global or local `npm exec --package skillpacks ...` or `npm exec --package @glexcorp/gskp ...` usage. Do not use `npx skillpack ...` singular; that is an unrelated npm package with a different project format.
 
 ## Initialization
 
 ```bash
-./init.sh
+cd ~/my-project
+npx skillpacks init
 ```
 
-`init.sh` initializes only the global core skills:
+`npx skillpacks init` initializes base skills **project-local** for the current repository:
 
-- `global/claude/*` → `~/.claude/skills/*`
-- `global/codex/*` → `~/.codex/skills/*`
+- `base/claude/*` -> `<project>/.claude/skills/*`
+- `base/codex/*` -> `<project>/.codex/skills/*`
 
-It does **not** install `packs/*` globally.
+It records `base_skills: true` in `.agents/project.json`, so later `npx skillpacks refresh` updates them from the package version being run. It does **not** install `packs/*` — domain packs are never installed as base skills.
 
-To remove repo-managed global skill installs:
+There is no user-home (global) base install. To clean up legacy repo-managed base installs left in `~/.claude/skills` and `~/.codex/skills` by the retired init path:
 
 ```bash
-./init.sh --uninstall
+npx skillpacks uninstall-global
 ```
 
 ## Project Packs
@@ -78,37 +83,48 @@ To remove repo-managed global skill installs:
 Project packs are enabled per repository.
 
 ```bash
-scripts/pack.sh list
-scripts/pack.sh recommend
-scripts/pack.sh install business-discovery
-scripts/pack.sh install business-growth
-scripts/pack.sh install business-ops
-scripts/pack.sh install game
-scripts/pack.sh install devtool
-scripts/pack.sh install creator-foundation
-scripts/pack.sh install youtube-ops
-scripts/pack.sh install code-quality
-scripts/pack.sh install agentic-skills-bench
-scripts/pack.sh install monorepo
-scripts/pack.sh install remotion
-scripts/pack.sh install project-fleet
-scripts/pack.sh install alignment-loop
-scripts/pack.sh install design-system
-scripts/pack.sh which design-system
-scripts/pack.sh status
-scripts/pack.sh remove game
-scripts/pack.sh remove design-system
+npx skillpacks list
+npx skillpacks recommend
+npx skillpacks install business-discovery
+npx skillpacks install business-growth
+npx skillpacks install business-ops
+npx skillpacks install game
+npx skillpacks install devtool
+npx skillpacks install creator-foundation
+npx skillpacks install youtube-ops
+npx skillpacks install code-quality
+npx skillpacks install agentic-skills-bench
+npx skillpacks install monorepo
+npx skillpacks install remotion
+npx skillpacks install project-fleet
+npx skillpacks install alignment-loop
+npx skillpacks install design-system
+npx skillpacks which design-system
+npx skillpacks status
+npx skillpacks remove game
+npx skillpacks remove design-system
 ```
 
-`scripts/pack.sh install <name>` accepts either a pack name or a skill name from any pack. Pack names install every skill in the pack. Skill names install only that one project-local skill and record it under `.agents/project.json` `enabled_skills`.
+`npx skillpacks install <name>` accepts either a pack name or a skill name from any pack. Pack names install every skill in the pack. Skill names install only that one project-local skill and record it under `.agents/project.json` `enabled_skills`.
+
+For source-checkout development, `scripts/pack.sh install <pack-or-skill>` remains supported from a local clone. The npm CLI also supports `npx skillpacks refresh`, `npx skillpacks doctor`, `npx skillpacks doctor --fix`, `npx skillpacks doctor --fix --agent-docs --dry-run`, and deck installation from manifest metadata including `npx skillpacks install-deck vard`, `npx skillpacks install-deck ord`, `npx skillpacks install-deck business-afps`, `npx skillpacks install-deck devtool-afps`, and `npx skillpacks install-deck game-afps`.
+
+Alignment convention maintenance also has npm wrappers:
+
+```bash
+npx skillpacks alignment bundles --check
+npx skillpacks alignment pages audit
+npx skillpacks alignment pages inject-tts --force alignment/example.html
+npx skillpacks alignment verify
+```
 
-The npm CLI also supports equivalent project-local installs after publication: `npx skillpacks install <pack-or-skill>`, `npx skillpacks remove <pack-or-skill>`, `npx skillpacks refresh`, `npx skillpacks doctor`, and deck installation from manifest metadata including `npx skillpacks install-deck vard`, `npx skillpacks install-deck ord`, `npx skillpacks install-deck business-afps`, `npx skillpacks install-deck devtool-afps`, and `npx skillpacks install-deck game-afps`.
+Use direct `node scripts/upgrade-alignment-page.mjs`, `node scripts/audit-alignment-pages.mjs`, and `node scripts/inject-tts.mjs` commands from a source checkout. Use `npx skillpacks alignment ...` from npm-installed target repos.
 
-`scripts/pack.sh list-packs` is an internal subcommand used by Codex `$exec` routing (see `global/codex/exec/SKILL.md`). It prints enabled packs from `.agents/project.json` one per line with no decoration, distinct from the human-facing `list` above; prefer `list` or `status` for interactive use.
+`scripts/pack.sh list-packs` is an internal subcommand used by Codex `$exec` routing (see `base/codex/exec/SKILL.md`). It prints enabled packs from `.agents/project.json` one per line with no decoration, distinct from the human-facing `list` above; prefer `list` or `status` for interactive use.
 
 Claude users can run `/pack` with no arguments, and Codex users can run `$pack` with no arguments. If `.agents/project.json` exists, the skill refreshes local skill roots from that committed project designation. If it is missing, the assistant inspects the repository, recommends a pack, and asks before installing.
 
-`scripts/pack.sh which <skill>` shows which pack provides a skill and whether it is installed. `scripts/pack.sh refresh` recreates local skill roots; it does not by itself force an active CLI skill registry to reload. Claude Code watches existing `.claude/skills` roots and supports `/reload-skills`; `/clear` starts a new empty-context conversation and can also pick up refreshed skills. Restart Claude Code if `.claude/skills` did not exist when the session started or the skill is still invisible. In Codex, start a fresh Codex CLI session if the `$` skill list stays stale.
+`npx skillpacks which <skill>` shows which pack provides a skill and whether it is installed. `npx skillpacks refresh` recreates local skill roots; it does not by itself force an active CLI skill registry to reload. Claude Code watches existing `.claude/skills` roots and supports `/reload-skills`; `/clear` starts a new empty-context conversation and can also pick up refreshed skills. Restart Claude Code if `.claude/skills` did not exist when the session started or the skill is still invisible. In Codex, start a fresh Codex CLI session if the `$` skill list stays stale.
 
 Pack installation creates local repo-managed skill roots in the current project. Active installs expose the canonical `SKILL.md` and exclude `archive/`:
 
@@ -153,13 +169,13 @@ For mixed monorepos, keep `project_type` as the default designation, set `enable
 }
 ```
 
-Git-checkout `scripts/pack.sh` commands preserve existing `project_scopes` and `notes` fields when `jq` is available. The npm `skillpacks` Node-owned commands preserve `project_scopes`, `notes`, `pinned_versions`, `enabled_skills`, `skill_updates`, and `agent_mode` without requiring `jq`.
+Git-checkout `scripts/pack.sh` commands preserve existing `project_scopes` and `notes` fields when `jq` is available. The npm `gskp` Node-owned commands preserve `project_scopes`, `notes`, `pinned_versions`, `enabled_skills`, `skill_updates`, and `agent_mode` without requiring `jq`.
 
 Pack commands also write `.agents/.pack.lock` owner metadata and automatically remove stale locks whose recorded process is no longer running.
 
 `.agents/project.json` also accepts an optional `agent_mode` field (`"claude-only" | "codex-only" | "hybrid"`) that names the Phase 11 operating mode for the project. Set or clear it with `scripts/pack.sh set-mode <claude-only|codex-only|hybrid|unset>`; the value is preserved across `install`, `remove`, and `refresh`. `SKILLS_AGENT_MODE` overrides the file for the current shell, and `scripts/agent-mode.sh` resolves the effective mode (env > project.json > empty). See `docs/operating-modes.md`.
 
-If an assistant does not discover project-local skills, use the global `pack` or `research-roadmap` skill as the launcher. The pack files still stay project-local.
+If an assistant does not discover project-local skills, use the base `pack` or `research-roadmap` skill as the launcher. The pack files still stay project-local.
 
 For workflow ordering, lead-in recommendations, and overlay dependencies, see [`docs/pack-workflow-matrix.md`](docs/pack-workflow-matrix.md).
 
@@ -167,7 +183,7 @@ For workflow ordering, lead-in recommendations, and overlay dependencies, see [`
 
 ```text
 agentic-skills/
-├── global/
+├── base/
 │   ├── claude/<name>/SKILL.md
 │   └── codex/<name>/SKILL.md
 ├── packs/
@@ -193,34 +209,34 @@ agentic-skills/
 │   └── skills-showcase/
 ├── packages/
 │   └── skillpacks/
-├── init.sh
 └── docs/
 ```
 
-The repository root is private workspace metadata. The publishable `skillpacks` package lives in `packages/skillpacks/`, while the public Skills Showcase app and its data generators live in `apps/skills-showcase/`.
+The repository root is private workspace metadata. The publishable `skillpacks` package source lives in `packages/skillpacks/`; `@glexcorp/gskp` is generated as a staged package variant during release. The public Skills Showcase app and its data generators live in `apps/skills-showcase/`.
 
-## Global Core
+## Base Core
 
-The global surface is intentionally small and domain-neutral. `./init.sh` installs six skills under `global/claude/` (mirrored under `global/codex/`):
+The base surface is intentionally small and domain-neutral. `npx skillpacks init` installs 11 skills under `base/claude/`, 8 of which are mirrored under `base/codex/`:
 
 ```text
-codebase-status, idea-scope-brief, init-agentic-skills, pack,
-provision-agentic-config, skills
+afps-status, animation-design-planner, autoresearch, autoresearch-prep,
+codebase-status, fork-idea-branch, idea-scope-brief, init-agentic-skills,
+pack, provision-agentic-config, skills
 ```
 
-### Codex-only global skill
+### Claude-only base skills
 
-One global skill ships only under `global/codex/` (no `global/claude/` mirror):
+Three base skills ship only under `base/claude/` (no `base/codex/` mirror):
 
 ```text
-afps-status
+autoresearch, autoresearch-prep, fork-idea-branch
 ```
 
 `$afps-status` summarizes AFPS product-workflow progress from existing artifacts and recommends the next concrete skill command. It is read-mostly reconciliation, not a competing workflow state system.
 
 ### `delegate` moved to the `agent-bridge` pack
 
-`delegate` is **no longer global** — it lives in `packs/agent-bridge/claude/delegate`. Install it with `scripts/pack.sh install agent-bridge`. It remains **Claude-only** with no Codex mirror.
+`delegate` is **no longer a base skill** — it lives in `packs/agent-bridge/claude/delegate`. Install it with `npx skillpacks install agent-bridge`. It remains **Claude-only** with no Codex mirror.
 
 `/delegate` is the synchronous, in-session sibling of `/handoff --target=codex`: it drafts and approves a packet via `scripts/approved-plan.sh`, then invokes `codex exec "<target-skill> --execute-approved"` live from Claude. It is `hybrid`-only by design. Codex users should not expect a `$delegate` — drive hybrid delegation from the Claude side, or use `/handoff --target=codex` for the async variant. See `packs/agent-bridge/claude/delegate/SKILL.md` and `docs/operating-modes.md`.
 
@@ -233,7 +249,7 @@ For SaaS, marketplaces, productivity tools, business applications, and enterpris
 Install only the current lane:
 
 ```text
-business-discovery: icp, enterprise-icp, competitive-analysis, customer-feedback,
+business-discovery: customer-discovery, enterprise-icp, competitive-analysis, customer-feedback,
 value-prop-canvas, lean-canvas, positioning
 
 customer-lifecycle: journey-map, onboarding-map, conversion-map,
@@ -247,7 +263,7 @@ burn-rate, runway-model, investor-update, reconcile-research,
 mvp-gap, scale-audit, platform-strategy
 ```
 
-`scripts/pack.sh install business-app` remains a compatibility alias that installs all four, but narrow installs are preferred for context size.
+`npx skillpacks install business-app` remains a compatibility alias that installs all four, but narrow installs are preferred for context size.
 
 
 ### Game
@@ -291,7 +307,7 @@ youtube-portfolio, youtube-peer-benchmark, youtube-search-positioning,
 youtube-cadence-diagnosis
 ```
 
-`scripts/pack.sh install creator-media` remains a compatibility alias that installs `creator-foundation` and `youtube-ops`. Use `remotion` for `youtube-format-research`, `video-script`, and `video-build`.
+`npx skillpacks install creator-media` remains a compatibility alias that installs `creator-foundation` and `youtube-ops`. Use `remotion` for `youtube-format-research`, `video-script`, and `video-build`.
 
 LinkedIn-first evidence work uses the `creator-foundation` matrix/schema/dossier path:
 
@@ -348,7 +364,7 @@ extract-shared-types, quality-sweep
 For pnpm workspace monorepos that may use Turborepo and need package-aware execution, guardrails, and shipping.
 
 ```bash
-scripts/pack.sh install monorepo
+npx skillpacks install monorepo
 ```
 
 ```text
@@ -381,16 +397,16 @@ PoketoWork kanban packs are hibernated while Poketo.work is being rebuilt. Their
 
 ## Moved Skills
 
-Former global business/product skills now live in narrower project packs. `business-app` remains a compatibility alias for all four business packs.
+Former base business/product skills now live in narrower project packs. `business-app` remains a compatibility alias for all four business packs.
 
 ```bash
-scripts/pack.sh install business-discovery
-scripts/pack.sh install customer-lifecycle
-scripts/pack.sh install business-growth
-scripts/pack.sh install business-ops
+npx skillpacks install business-discovery
+npx skillpacks install customer-lifecycle
+npx skillpacks install business-growth
+npx skillpacks install business-ops
 ```
 
-Creator-media and YouTube work is similarly split between `creator-foundation`, `youtube-ops`, and `remotion`. Fleet/portfolio work moved from global core into `project-fleet`.
+Creator-media and YouTube work is similarly split between `creator-foundation`, `youtube-ops`, and `remotion`. Fleet/portfolio work moved from base core into `project-fleet`.
 
 ## Version Pinning
 
@@ -398,14 +414,14 @@ Skills can be pinned to an archived version. When a skill's version is bumped, t
 
 ```bash
 # Archive current version before bumping
-bash scripts/skill-archive.sh global/claude/ship
+bash scripts/skill-archive.sh base/claude/codebase-status
 
 # Pin a pack skill to an archived version
 scripts/pack.sh pin devtool-adoption v0.0
 scripts/pack.sh unpin devtool-adoption
 
-# Pin a global skill during initialization
-./init.sh --pin ship=v0.0
+# Pin a base skill to an archived version
+npx skillpacks pin ship v0.0
 
 # Audit archive integrity
 bash scripts/skill-archive-audit.sh
@@ -425,7 +441,7 @@ See [`docs/skill-versioning.md`](docs/skill-versioning.md) for the full versioni
 pnpm --dir tests test
 ```
 
-`skill-deps.sh`, `skill-pack-routing-audit.sh`, and `skill-versions.sh` scan `global/` and `packs/`; `skill-mirror-parity-audit.sh` checks mirrored pack skill pairs under `packs/`.
+`skill-deps.sh`, `skill-pack-routing-audit.sh`, and `skill-versions.sh` scan `base/` and `packs/`; `skill-mirror-parity-audit.sh` checks mirrored pack skill pairs under `packs/`.
 
 Live agent behavior tests are opt-in because they invoke authenticated CLIs and may spend model budget: