skillpacks 0.1.1 → 0.1.2

package/base/claude/codebase-status/archive/v0.9/SKILL.md

@@ -0,0 +1,112 @@
+---
+name: codebase-status
+description: Report what a repository is, what recent conversation history says about it, current application status, and outstanding work
+type: analysis
+version: v0.9
+argument-hint: "[optional repo path, focus, or --no-history]"
+---
+
+# Codebase Status
+
+Invoke as `/codebase-status`.
+
+Use this skill when the user asks what a repo or application is, where work was left, what is outstanding, whether it is ready/stable, or wants a detailed status report that combines codebase evidence with local Claude/Codex conversation history.
+
+This is read-only status synthesis. It does not replace `/roadmap`: `roadmap` maintains the task pipeline and priority queue; `codebase-status` explains the actual current state of the repo and relevant prior conversations.
+
+## Process
+
+1. Resolve target repo:
+   - Default to the current working directory.
+   - If the user provides a path, use that repo.
+   - Confirm it is a git worktree when possible; if not, continue with filesystem evidence and report that git evidence is unavailable.
+2. Read project orientation:
+   - `README.md`, `AGENTS.md`, `CLAUDE.md`, `.agents/project.json`, package/workspace files, app entrypoints, and high-signal docs.
+   - Identify project type, main apps/packages, runtime, tests, and likely user-facing or developer-facing purpose.
+3. Read planning and status docs when present:
+   - `tasks/roadmap.md`, `tasks/todo.md`, `tasks/manual-todo.md`, `tasks/record-todo.md`, `tasks/recurring-todo.md`, `tasks/history.md`, `tasks/lessons.md`, `tasks/phases/`, `specs/`, `spec.md`, and `research/`.
+   - Distinguish checked-off work, active work, manual blockers, advisory work, stale docs, and missing docs.
+4. Read routing evidence for product, research, or spec repos:
+   - When `research/`, `specs/`, `spec.md`, product docs, or prototype artifacts exist, read `docs/pack-workflow-matrix.md` and `docs/skill-next-step-contracts.md` if present.
+   - Identify the last completed relevant research/product skill from artifacts, task history, or conversation history. Before recommending another research/product skill, consult that skill's active `SKILL.md` `## Next Steps` contract and any `## Next Steps` section in its output artifact.
+   - Treat those route contracts as stronger evidence than a generic status impression. If docs disagree, report the contradiction and prefer the newest active skill contract or route matrix with file evidence.
+   - For pack-local recommendations, check `.agents/project.json` `enabled_packs` first. If the target skill's pack is not enabled, recommend `npx skillpacks install <pack-name>` from the project shell, before the skill.
+5. Inspect git evidence:
+   - `git status --short`
+   - current branch and upstream status
+   - last 20 commits
+   - unpushed commits when an upstream exists
+   - changed files in the working tree
+6. Inspect codebase health signals:
+   - Project structure and primary modules.
+   - Test/build/lint scripts from package manifests or equivalent tool files.
+   - Obvious TODO/FIXME markers, failing-test notes, disabled tests, and local quality scripts.
+   - Do not run expensive commands by default. Run cheap read-only inspections freely; ask or state assumptions before long builds, network installs, destructive commands, or mutation.
+7. Find related conversation history unless `--no-history` is passed:
+   - Read full available local prompt history, not a sample:
+     - `~/.claude/history.jsonl`
+     - `~/.codex/history.jsonl`
+     - `~/.codex/sessions/**/*.jsonl` for Codex session metadata only when needed to map session IDs to cwd.
+   - Filter to records whose project/cwd matches the target repo or whose text mentions the absolute path, repo directory name, package/app name, or repository URL.
+   - Exclude system, developer, tool output, injected skill payloads, and base instruction text from examples.
+   - Use line-by-line parsing for scale and deduplicate Codex prompt records when compact and rich histories overlap.
+   - Extract recent user goals, repeated requests, unresolved questions, blockers, and prior recommendations.
+8. Synthesize status:
+   - What this repo/app is.
+   - What has happened recently.
+   - Current implementation status by major area.
+   - Outstanding work, grouped as blocking, next execution, advisory, manual/human-only, and uncertain.
+   - Mismatches between conversation history, task docs, git history, and code reality.
+   - Confidence level for each major conclusion: high, medium, or low, with evidence.
+9. Recommend next route (check whether each skill's pack is installed; if not, recommend `npx skillpacks install <pack>` from the project shell, first):
+   - Use phase-aware routing before naming a command:
+     - If `tasks/todo.md` or active phase docs contain actionable implementation work, recommend `/exec`. Do not route back to research merely because advisory gaps exist.
+     - If finished work is dirty, unpushed, unvalidated, or needs packaging/review before handoff, recommend `/ship`.
+     - If user-facing product research/prototype artifacts are missing and no implementation/shipping queue is active, follow the canonical AFPS route from the routing evidence: `customer-discovery -> competitive-analysis -> journey-map -> positioning -> user-flow-map -> ux-variations [specific-user-flow] -> ui-interview [specific-ux-variation] -> prototype -> uat --variant-evaluation -> consolidate-variations -> research-roadmap --post-prototype -> spec-interview -> research-roadmap --post-spec -> roadmap`. Use `/ui-interview --requirements-only` and `/ux-variations --layout-mode` only when the user explicitly needs a fixed content/data/action contract and layout-only alternatives.
+     - If `research/icp.md` and `research/competitive-analysis.md` exist but `research/journey-map.md` is missing, check `customer-lifecycle` availability. If it is not enabled, recommend `npx skillpacks install customer-lifecycle` from the project shell, before `/journey-map`; if enabled, recommend `/journey-map`.
+     - Treat `value-prop-canvas` and `lean-canvas` as optional risk-driven detours only. Before recommending either, check `business-research` availability; if it is not enabled, recommend `npx skillpacks install business-research` from the project shell, before `/value-prop-canvas` or `/lean-canvas`. Recommend `/value-prop-canvas` only for contested solution-fit evidence, and `/lean-canvas` only for material business-model risk; do not make either a default blocker before `journey-map`, `positioning`, `user-flow-map`, or layout-mode UX variations.
+     - If no research, spec, task, implementation, validation, dirty, or unpushed work remains, recommend `/brainstorm` to discover new AFPS work.
+   - `/exec` when the next task is already clear and executable. _(exec-loop pack)_
+   - `/roadmap` when specs exist but sequencing or task queue health needs updating. _(agent-work-admin pack)_
+   - `/feature-interview` when the next idea/direction is not yet represented in specs or tasks. _(product-design pack)_
+   - `/reconcile-dev-docs audit` or `/reconcile-dev-docs fix tasks` when task docs contradict git/code reality. _(docs-health pack)_
+   - `/spec-drift` when specs and code appear out of sync. _(agent-work-admin pack)_
+   - `/guide` only for human-only external blockers. _(guided-walkthrough pack)_
+   - `/brainstorm` when the repo is complete/current and the next step is discovering a new phase. _(product-design pack)_
+
+## Output
+
+Produce a structured report with:
+
+- **Overview:** repo path, project purpose, project type, primary apps/packages, branch, dirty/unpushed status.
+- **History Signal:** number of Claude/Codex prompt records scanned, number matched to this repo, date range, and 3-8 concise real examples.
+- **Recent Work:** last commits and task/history evidence.
+- **Current Status:** major areas and whether each appears complete, in progress, blocked, stale, or unknown.
+- **Outstanding Work:** prioritized list with evidence and recommended owner/skill.
+- **Risks And Drift:** contradictions, stale docs, missing tests, manual blockers, or uncertain claims.
+- **Recommended Next Step:** one concrete next work item and one command.
+
+End with exactly:
+
+```md
+**Next work:** <specific task, blocker, discovery task, or "none">
+**Recommended next command:** <one command or route>
+```
+
+## Constraints
+
+- Read-only by default. Do not modify files, create commits, run installs, or execute long test suites unless the user explicitly asks.
+- Do not include private conversation history unrelated to the target repo.
+- Quote only short prompt excerpts. Prefer paraphrase plus timestamp/source/project.
+- Do not treat `tasks/record-todo.md` or `tasks/recurring-todo.md` as execution queues unless an item has clearly become concrete work.
+- Do not put agent-executable work in `tasks/manual-todo.md` or route it to `/guide`.
+- Do not recommend `/roadmap` if the finding is only "tell me the status"; use `/roadmap` only when the task pipeline itself needs queue maintenance or roadmap extension.
+- When recommending a pack skill, first check `.agents/project.json` for `enabled_packs`. If the target skill's pack is not enabled, include `npx skillpacks install <pack-name>` from the project shell, as a prerequisite in the recommendation.
+- Do not create or modify GitHub Actions workflows.
+
+## Default Shipping Contract
+
+- **Default next-step routing:** when reporting completion, include either `Recommended next skill: <command>` or the two-line pair `**Next work:** <specific task or "none">` and `**Recommended next command:** <one command or route>` so the next caller has a concrete handoff.
+- Normally read-only and should not create or modify tracked repository files.
+- If the user explicitly asks this skill to create or modify tracked files, finish by committing and pushing all intended changes to the repository primary branch (`main` when present, otherwise `master`) before stopping.
+- This contract does not override stricter safety rules about secrets, destructive history changes, release publication/tag confirmation, production deploy confirmation, paid actions, or public visibility changes.

package/{global → base}/claude/fork-idea-branch/ALIGNMENT-PAGE.md

@@ -57,7 +57,9 @@ The tier is set per skill via `visual_tier` in SKILL.md frontmatter (or detected
 
 **Report-only research gates.** For report-only or pre-approval research skills, the first `review` alignment page must explicitly contain proposed research scope, source plan, assumptions/confidence, recommended output path, proposed file changes, and approval gates before synthesized research begins and before any working packet, canonical research, spec, or task file is created or updated.
 
-**Staged research workflow.** Heavy report-first research uses a scope-first three-stage approval workflow. Stage 1 is minimal scope discovery only: inspect enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions/confidence, proposed working-packet and canonical output paths, and research-scope approval gates. Stop for final compiled response YAML approval of the research scope. Do not synthesize findings, rank candidates, recommend a path, or create working packets, canonical research, spec, or task files in Stage 1. Stage 2 starts only after final compiled response YAML approves the research scope with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Perform the synthesized research, run required source/code checks, and write only a non-canonical working packet. Flat mode uses `research/_working/preliminary-fork-idea-branch-research.md`; product-path mode uses `research/{slug}/_working/preliminary-fork-idea-branch-research.md`. Existing raw evidence or search-log outputs may remain as supporting evidence when the skill contract already requires them, but synthesized deliverables stay in the working packet. Update the same `review` alignment page so it renders the full preliminary packet, evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Partial compiled responses revise the working packet and review page, then remain in Stage 2. Stage 3 consumes final compiled response YAML for artifact approval only when it has `approval_status: ready-for-agent-review` and no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths, and convert the page to `confirmed` with the approval record preserved.
+**Staged research workflow.** Heavy report-first research uses a scope-first three-stage approval workflow. Stage 1 is minimal scope discovery only: inspect enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions/confidence, proposed working-packet and canonical output paths, research-scope approval gates, and a `Stage 2 Preview / Expected Review Format` section that shows the future Stage 2 review shape before research starts. Stop for final compiled response YAML approval of the research scope. Do not synthesize findings, rank candidates, recommend a path, or create working packets, canonical research, spec, or task files in Stage 1. Stage 2 starts only after final compiled response YAML approves the research scope with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Perform the synthesized research, run required source/code checks, and write only a non-canonical working packet. Flat mode uses `research/_working/preliminary-fork-idea-branch-research.md`; product-path mode uses `research/{slug}/_working/preliminary-fork-idea-branch-research.md`. Existing raw evidence or search-log outputs may remain as supporting evidence when the skill contract already requires them, but synthesized deliverables stay in the working packet. Update the same `review` alignment page so it renders the complete working-packet substance as structured HTML review UI: purpose-built sections, tables, matrices, gates, cards, and tier-appropriate charts or diagrams that preserve every packet section, finding, caveat, and decision detail without summary loss. Raw Markdown packet text may appear only as a supplemental source view after the rendered review UI; do not make a `Full Preliminary Packet` or `Full Working Packet` raw Markdown dump, giant `<pre><code>` block, link-only view, or source-only view the primary review surface. Include the evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Partial compiled responses revise the working packet and review page, then remain in Stage 2. Stage 3 consumes final compiled response YAML for artifact approval only when it has `approval_status: ready-for-agent-review` and no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths, and convert the page to `confirmed` with the approval record preserved.
+
+**Stage 2 review-page template.** Stage 1 scope-review pages must include a `Stage 2 Preview / Expected Review Format` section that previews the exact section sequence, table shapes, evidence treatment, visual density, and approval gates the user will see after research; make it specific enough for the user to flag preferred content formats, grouping, labels, visuals, and evidence density before the agent does heavy research. After scope approval and research, the Stage 2 review page must include these sections as structured HTML, not as raw Markdown dumps: `Research Scope Approved` status block with approved scope, source plan, output paths, and unresolved caveats; `Executive Findings` or a skill-specific top-findings section with confidence labels; `Evidence Matrix` mapping claim, evidence/source/path, inference, confidence, assumptions, and decision impact; `Working Packet Review` with the complete working-packet substance rendered as purpose-built HTML sections, lists, tables, matrices, cards, and tier-appropriate charts or diagrams; `Alternatives / Rejected or Lower-Confidence Findings`; `Source Coverage & Gaps`; `Assumptions / Confidence Register`; `Proposed Canonical Artifacts & File Changes`; a required `User Format Preferences` gate asking whether the layout, tables, grouping, visuals, labels, and evidence density match expectations; and a required `Final Artifact Approval` gate for Stage 3 readiness. Raw Markdown packet text, search logs, or source notes may appear only as supplemental source views after the rendered review UI and never replace the Stage 2 review sections.
 
 **Fork-specific gates.** Render the fork rationale, per-branch scope (label, slug, ICP hypothesis, value wedge hypothesis), reuse decisions per branch (fresh start vs. carry-forward with artifact types), archive confirmation for the current path, and proposed file changes as gates. The reuse decision gate must default to "Fresh start (recommended)" and explain contamination risk before presenting carry-forward options. Apply the shared artifact-destination/proposed-file-changes de-duplication rule: combine them into one visual gate when they ask only the same path-destination question, and render separate gates when destination approval differs from downstream mutation scope, timing, or allowed files.
 

package/{global → base}/claude/fork-idea-branch/SKILL.md

@@ -10,7 +10,7 @@ argument-hint: "[path-a-label] [path-b-label] [--from <skill>] [--reuse <artifac
 
 Use this skill when a single research path should split into multiple parallel product paths — for example, two compelling ICPs with conflicting pain points that need separate products, or positioning evidence showing the concept serves fundamentally different markets. This skill archives the current research, scaffolds new product paths, and routes each back to `/idea-scope-brief` for a fresh start.
 
-This is a global helper, not part of any pack's linear workflow. It can be invoked directly by the user or suggested by research skills when alignment page answers indicate divergent paths.
+This is a base helper, not part of any pack's linear workflow. It can be invoked directly by the user or suggested by research skills when alignment page answers indicate divergent paths.
 
 This skill is distinct from `/product-line fork` (which adds a path to the portfolio). This skill splits and restarts research from scratch.
 

package/{global/codex → base/claude}/idea-scope-brief/ALIGNMENT-PAGE.md

@@ -9,6 +9,8 @@ When this skill produces durable deliverables (research, specs, plans, reports,
 
 The tier is set per skill via `visual_tier` in SKILL.md frontmatter (or detected by the upgrade script). The bundled ALIGNMENT-PAGE.md receives tier-appropriate rendering guidance from the generator.
 
+**Context intake.** This skill declares `context_intake: deep` (Deep interview). Gather repo/project context before asking, present an assumptions manifest when the contract requires it, ask focused questions, and confirm coverage before building review artifacts.
+
 **Alignment lifecycle.** Alignment pages have three lifecycle states. `review` is the draft/pre-approval page: it contains the complete deliverable preview, evidence, required inline gate questions, section feedback controls, local section-feedback YAML only after a section feedback choice is selected, and one bottom `Compile Responses` control. `confirmed` is the post-approval page after the agent has consumed final compiled response YAML, made any approved user-requested edits, and written or updated the approved canonical artifacts. A confirmed page is finished and current for the completed alignment cycle, but it remains amendable when later research changes or extends the conclusions. `amended` is a future revision of a confirmed page: archive the previous confirmed page first, mark what changed, preserve the prior approval record, and highlight the new evidence, decisions, or scope that changed the earlier conclusion.
 
 **Central alignment index.** When a research-producing skill writes or updates an HTML alignment page, also maintain the repository's central alignment index. Look for `alignment/index.html` at the repository root; if the target repo documents a different central alignment-index path, use that path and report it. If the index does not exist, create it as a self-contained HTML page that works from `file://`, links to every active `alignment/*.html` page except itself, and includes the newly written page. Every index entry must include the page's alignment date in a muted meta span after the link, formatted `YYYY-MM-DD`, for example `<a href="page.html">Title</a> <span class="meta">2026-06-09</span>`. The alignment date is the review or confirmed date written inside the page itself; new pages use today's date. If the index exists, update it instead of blindly appending: preserve valid existing entries, metadata, and entry dates; remove duplicate entries for the same page path; keep links repo-relative; and include the new or updated page. When updating an existing index, add a date only to entries that are missing one, deriving it from the page's internal alignment date, or from file history when the page has no internal date. If a `new` marker is used, combine it with the date as `<span class="meta">new · YYYY-MM-DD</span>`, and remove stale `new` markers from entries that are no longer the latest addition.
@@ -57,7 +59,9 @@ The tier is set per skill via `visual_tier` in SKILL.md frontmatter (or detected
 
 **Report-only research gates.** For report-only or pre-approval research skills, the first `review` alignment page must explicitly contain proposed research scope, source plan, assumptions/confidence, recommended output path, proposed file changes, and approval gates before synthesized research begins and before any working packet, canonical research, spec, or task file is created or updated.
 
-**Staged research workflow.** Heavy report-first research uses a scope-first three-stage approval workflow. Stage 1 is minimal scope discovery only: inspect enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions/confidence, proposed working-packet and canonical output paths, and research-scope approval gates. Stop for final compiled response YAML approval of the research scope. Do not synthesize findings, rank candidates, recommend a path, or create working packets, canonical research, spec, or task files in Stage 1. Stage 2 starts only after final compiled response YAML approves the research scope with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Perform the synthesized research, run required source/code checks, and write only a non-canonical working packet. Flat mode uses `research/_working/preliminary-idea-scope-brief-research.md`; product-path mode uses `research/{slug}/_working/preliminary-idea-scope-brief-research.md`. Existing raw evidence or search-log outputs may remain as supporting evidence when the skill contract already requires them, but synthesized deliverables stay in the working packet. Update the same `review` alignment page so it renders the full preliminary packet, evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Partial compiled responses revise the working packet and review page, then remain in Stage 2. Stage 3 consumes final compiled response YAML for artifact approval only when it has `approval_status: ready-for-agent-review` and no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths, and convert the page to `confirmed` with the approval record preserved.
+**Staged research workflow.** Heavy report-first research uses a scope-first three-stage approval workflow. Stage 1 is minimal scope discovery only: inspect enough repository, user, and source context to propose research scope, source plan, assumptions, output paths, and approval questions. Build the `review` alignment page before synthesized research. The page must render the proposed scope, available source categories, known context, assumptions/confidence, proposed working-packet and canonical output paths, research-scope approval gates, and a `Stage 2 Preview / Expected Review Format` section that shows the future Stage 2 review shape before research starts. Stop for final compiled response YAML approval of the research scope. Do not synthesize findings, rank candidates, recommend a path, or create working packets, canonical research, spec, or task files in Stage 1. Stage 2 starts only after final compiled response YAML approves the research scope with no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Perform the synthesized research, run required source/code checks, and write only a non-canonical working packet. Flat mode uses `research/_working/preliminary-idea-scope-brief-research.md`; product-path mode uses `research/{slug}/_working/preliminary-idea-scope-brief-research.md`. Existing raw evidence or search-log outputs may remain as supporting evidence when the skill contract already requires them, but synthesized deliverables stay in the working packet. Update the same `review` alignment page so it renders the complete working-packet substance as structured HTML review UI: purpose-built sections, tables, matrices, gates, cards, and tier-appropriate charts or diagrams that preserve every packet section, finding, caveat, and decision detail without summary loss. Raw Markdown packet text may appear only as a supplemental source view after the rendered review UI; do not make a `Full Preliminary Packet` or `Full Working Packet` raw Markdown dump, giant `<pre><code>` block, link-only view, or source-only view the primary review surface. Include the evidence matrix, assumptions/confidence register, source coverage gaps, proposed canonical file changes, and artifact approval gates. Partial compiled responses revise the working packet and review page, then remain in Stage 2. Stage 3 consumes final compiled response YAML for artifact approval only when it has `approval_status: ready-for-agent-review` and no unresolved `needs-clarification`, unresolved `down` feedback, or other unresolved negative feedback. Apply approved edits first, archive the working packet to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-working-path>`, remove the active working packet, write the approved canonical artifacts to the unchanged output paths, and convert the page to `confirmed` with the approval record preserved.
+
+**Stage 2 review-page template.** Stage 1 scope-review pages must include a `Stage 2 Preview / Expected Review Format` section that previews the exact section sequence, table shapes, evidence treatment, visual density, and approval gates the user will see after research; make it specific enough for the user to flag preferred content formats, grouping, labels, visuals, and evidence density before the agent does heavy research. After scope approval and research, the Stage 2 review page must include these sections as structured HTML, not as raw Markdown dumps: `Research Scope Approved` status block with approved scope, source plan, output paths, and unresolved caveats; `Executive Findings` or a skill-specific top-findings section with confidence labels; `Evidence Matrix` mapping claim, evidence/source/path, inference, confidence, assumptions, and decision impact; `Working Packet Review` with the complete working-packet substance rendered as purpose-built HTML sections, lists, tables, matrices, cards, and tier-appropriate charts or diagrams; `Alternatives / Rejected or Lower-Confidence Findings`; `Source Coverage & Gaps`; `Assumptions / Confidence Register`; `Proposed Canonical Artifacts & File Changes`; a required `User Format Preferences` gate asking whether the layout, tables, grouping, visuals, labels, and evidence density match expectations; and a required `Final Artifact Approval` gate for Stage 3 readiness. Raw Markdown packet text, search logs, or source notes may appear only as supplemental source views after the rendered review UI and never replace the Stage 2 review sections.
 
 **Idea-specific gates.** Render the Idea Assumptions Manifest as a first-class assumptions/confidence gate inside the alignment page before proposed deliverables. The idea identity, slug, scope/non-goals, ICP readiness, and output-location/change-scope decisions must each be reviewable gates. Apply the shared artifact-destination/proposed-file-changes de-duplication rule: combine them into one visual gate when they ask only the same path-destination question, and render separate gates when destination approval differs from downstream mutation scope, timing, or allowed files.
 

package/{global → base}/claude/idea-scope-brief/CHANGELOG.md

@@ -1,5 +1,9 @@
 # idea-scope-brief Changelog
 
+## v0.16 - 2026-06-12
+
+- Standardized active pack and skill install guidance on `npx skillpacks install <pack-or-skill>` instead of agent-native `/pack install` or `$pack install` recommendations.
+
 ## v0.15 - 2026-06-11
 
 - Updated downstream business research pack checks from `business-discovery` to `business-research` after the pack rename.

package/base/claude/idea-scope-brief/SKILL.md

@@ -0,0 +1,157 @@
+---
+name: idea-scope-brief
+description: Shape a rough product or project idea into a scoped brief before customer discovery, market research, specifications, UX, UI, or implementation planning
+type: planning
+version: v0.16
+argument-hint: "[optional rough idea, product thought, or product-path scope]"
+context_intake: deep
+---
+
+# Idea Scope Brief
+
+Use this skill when the user has a half-formed idea and needs it cleaned up enough to enter the normal research and planning workflow. This skill is intentionally pre-customer-discovery: it clarifies the concept, problem hypothesis, beneficiary hypothesis, value wedge, constraints, non-goals, and unknowns, but does not select a validated target-customer segment, analyze competitors, define UX/UI, choose architecture, or write implementation specs.
+
+## Process
+
+### 0. Product-Path Scope Resolution
+
+Resolve research scope by product path before using code or app structure as a hint:
+
+1. If `$ARGUMENTS` names a non-archived `research/{slug}/` directory or a product-path ID whose `scope_path` points there, use that path. Treat `{slug}` as the product/app name, not the customer, audience, or segment label.
+2. If `$ARGUMENTS` names only `research/_archive/{slug}/` or a manifest entry with `status: archived` or legacy `status: abandoned`, stop and warn that the path is archived; do not write or update scoped outputs there.
+3. Read `research/.progress.yaml` when present. Normalize legacy `active_path` to `active_paths` on read and write back `active_paths` on manifest updates. Treat legacy `abandoned` as `archived`; exclude `archived`, `abandoned`, `deferred`, `revisit_candidate`, `promoted`, and any `scope_path` under `research/_archive/` from active target selection.
+4. If active product paths exist in the manifest, use those paths. If multiple active paths exist, ask which one to target unless this skill explicitly supports cross-path output.
+5. If no active manifest target exists, list non-archived product directories under `research/`, excluding `research/_archive/` and dot directories. Auto-select only when exactly one exists; ask when multiple exist.
+6. If no product directories exist, use flat `research/` single-product mode.
+7. Detect monorepo/app/package structure only as a secondary hint. Suggest creating a missing `research/{slug}/` product path when code clearly exposes an app, but do not require code or monorepo detection before using `research/{slug}/`.
+
+When product path `{slug}` is active, read and write research under `research/{slug}/`, specs under `specs/{slug}/`, and treat top-level `research/*.md` files as flat-mode documents or cross-path summaries.
+
+1. **Resolve context**
+   - Read `.agents/project.json` if it exists.
+   - Read README, CLAUDE.md, AGENTS.md, existing `research/`, `specs/`, and task docs when present.
+   - Determine whether the current directory is already a bootstrapped project. Treat it as bootstrapped when it has meaningful `README.md` plus `AGENTS.md` or `CLAUDE.md`; treat it as unbootstrapped when those are missing, placeholder-only, or the user is describing an idea outside any project repo.
+   - If `$ARGUMENTS` contains a rough idea, use it as the starting draft.
+   - If `$ARGUMENTS` names a non-archived `research/{slug}/` product path, use that path. If it names only `research/_archive/{slug}/`, stop and warn that the path is archived.
+   - Determine the concept identity and a normalized concept slug as soon as either is known from `$ARGUMENTS`, repo context, or the interview. Normalize by lowercasing, removing URL suffix noise, replacing non-alphanumeric runs with `-`, trimming leading/trailing `-`, and dropping only project-wide brand prefixes when the remaining word is the actual scoped concept (for example, `poketo.work` -> `work`; `Poketo Core` -> `poketo-core`).
+   - If existing research or the prompt suggests multiple related concepts may exist, prefer slugged output paths over generic filenames. Reserve generic `idea-brief.md` only for a single unambiguous project-level concept.
+   - If no rough idea is available from arguments or repo context, ask the user for the idea in plain language.
+   - Read `research/.progress.yaml` when present. Normalize `active_path` (singular legacy) to `active_paths` (plural list) when reading; treat legacy `abandoned` as `archived` and exclude archived/deferred/revisit/promoted paths plus `research/_archive/` scopes from active target selection. Treat `active_paths` as the current product/app/customer focuses and `product_paths[]` as parked, archived, or promoted product-path state, not git branch state.
+   - When the prompt, repo context, interview, or pivot history surfaces multiple related concepts, apps, product lines, or future pivots, update or propose updates to `research/.progress.yaml` with product-path entries instead of merging them into one generic concept. Use fields: `id`, `label`, `scope_path`, `status`, `source_skill`, `reason`, `archive_reason`, `archived_at`, `promoted_at`, `evidence_refs`, `revisit_trigger`, `next_skill`, `pipeline_stage`, and `last_touched`. Set `pipeline_stage: idea-scope-brief` on entries created by this skill.
+   - Keep the central concept in `active_paths` when it is the current focus. Record related or future concepts as `status: deferred` or `status: revisit_candidate` with a concrete revisit trigger and a likely next skill such as `/customer-discovery <path/audience>`; if `business-research` is not enabled, recommend `npx skillpacks install business-research` from the project shell, before `/customer-discovery`.
+   - When 3+ product paths exist in the manifest, recommend `/product-line review` to the user for portfolio management; if `business-ops` is not enabled, recommend `npx skillpacks install business-ops` from the project shell, before `/product-line`.
+
+2. **Keep the boundary clear**
+   - Do not run customer discovery, competitive analysis, journey mapping, UX variation, UI interview, roadmap, or implementation planning inside this skill.
+   - Do not validate the market with broad web research. Use light repo/context inspection only; downstream research skills own evidence gathering.
+   - Treat every user claim as a hypothesis unless supported by existing project files.
+
+3. **Surface an Idea Assumptions Manifest**
+   - Before deep questioning, present what you think the concept is.
+   - Tag assumptions as `[from prompt]`, `[from repo]`, `[from research]`, or `[inferred]`.
+   - Cover:
+     - concept summary
+     - problem hypothesis
+     - target beneficiary or user hypothesis
+     - product/category guess
+     - value wedge
+     - constraints
+     - non-goals
+     - riskiest unknowns
+   - Ask the user to confirm, correct, or flag assumptions before writing.
+   - Deliver the manifest inline as the final message text of its own turn; ask the confirmation question in the next turn. AskUserQuestion option previews may mirror the manifest as a supplement but are never the sole channel. Never emit it only as mid-turn text in a turn that ends with a tool call — harness rendering does not guarantee mid-turn text is shown. A confirmation question must never reference content the user has not been shown.
+
+### Market Structure Handoff
+
+During the Idea Assumptions Manifest, if the concept appears marketplace/platform/B2B2C/multi-sided, add a compact `Market Structure Handoff` note:
+
+- Name the apparent sides and the expected value exchange between them.
+- Mark those sides and exchanges as hypotheses, not validated customer segments; do not decide which side is the customer, buyer, or primary target segment here.
+- Keep the source tag for each side as `[from prompt]`, `[from repo]`, or `[inferred]` unless the user provides a correction.
+- If the concept appears single-sided, omit the handoff or state that no marketplace/platform/B2B2C/multi-sided handoff is apparent.
+
+4. **Interview until idea-ready**
+   - Ask 1 to 3 focused questions per turn.
+   - Resolve only concept-level ambiguity:
+     - what problem exists
+     - who might care
+     - what outcome changes for them
+     - what makes the idea different enough to investigate
+     - what must stay out of scope
+     - what constraints are real now
+   - When unsure, recommend a practical default and clearly mark it as an assumption.
+
+5. **Coverage checkpoint**
+   - Present the final concept summary, unknowns, and readiness for customer discovery.
+   - Restate the resolved concept identity, slug, and exact output paths before writing.
+   - If the conversation pivoted from the initial concept to a different central concept, write the pivoted concept to its own slugged brief and preserve the initial concept as a related or future concept in the brief and interview log. Do not merge both concepts into one generic project-level brief.
+   - Ask whether any core premise, constraint, or non-goal is wrong before writing.
+
+6. **Build pre-approval alignment preview**
+   - Before writing any canonical `research/**/idea-brief.md`, `research/**/idea-brief-interview.md`, legacy flat `research/idea-brief-{slug}.md` variant, or `research/.progress.yaml`, build `alignment/idea-scope-brief-{topic}.html` as the review artifact.
+   - The HTML page must render the Idea/Concept Assumptions Manifest, artifact destinations, proposed file changes, coverage checkpoint, and approval gates, including any Market Structure Handoff.
+   - Attempt to open the page in the browser and point the user at the repo-relative path.
+   - Treat coverage-checkpoint confirmation as non-final; it only confirms the draft scope is ready to preview. Only final compiled YAML from the alignment page authorizes canonical writes.
+   - Before compiled YAML approval, the next action is review or revision of the HTML alignment page. Do not include `Recommended next skill`, `Recommended next command`, or downstream routing language until after final compiled YAML approval has been provided and the approved artifacts below have been written or updated.
+   - When feedback-only YAML is provided, revise the alignment page and ask again; do not write canonical artifacts until final compiled YAML approval is provided.
+
+### Review-Only Product Path Approval
+
+When the user approves a product-path fork or split at the alignment page level but explicitly withholds canonical-write approval (e.g., "review only, do not write canonical files"):
+
+1. **Do not write canonical files.** Keep `research/.progress.yaml`, `research/{slug}/idea-brief.md`, and `research/{slug}/idea-brief-interview.md` unchanged.
+2. **Render fully in the alignment page.** The proposed manifest entry, idea brief sections, and interview log for the review-only path must be rendered in full in the alignment page HTML — not summarized, linked, or embedded.
+3. **Mark the page as review-only-approved.** Set `approval_status: review-only-approved` in the alignment page status block. This is distinct from `confirmed` (canonical artifacts written) and `review` (awaiting any approval).
+4. **Downstream treatment.** Downstream skills must treat a review-only-approved path as provisional: it may be referenced as concept context, but it is not a canonical product path until manifest approval is later granted via a subsequent alignment cycle. See the provisional-path evidence rule in customer-discovery and competitive-analysis contracts.
+
+## Output
+
+Before writing anything in this section, verify the alignment page has final compiled YAML approval. Do not write canonical idea briefs, interview logs, or `research/.progress.yaml` until `alignment/idea-scope-brief-{topic}.html` has been reviewed and the user has provided final compiled YAML approval. Coverage-checkpoint confirmation is not final approval and does not authorize these writes.
+
+Write:
+
+- For one unambiguous project-level concept only: `research/idea-brief.md` and `research/idea-brief-interview.md`.
+- When a product identity is known, multiple concepts exist or may exist, or a pivot occurs: prefer `research/{slug}/idea-brief.md` and `research/{slug}/idea-brief-interview.md`; preserve flat `research/idea-brief-{slug}.md` only as legacy compatibility when no product path is being introduced.
+- If `$ARGUMENTS` names a non-archived product path, use unsuffixed scoped files under `research/{slug}/`: `research/{slug}/idea-brief.md` and `research/{slug}/idea-brief-interview.md`.
+- `research/.progress.yaml` — create or update only when multiple concepts, product paths, product lines, product-path scopes, or pivots are present. Use `product_paths` terminology instead of branch terminology.
+
+The idea brief must include:
+
+- `## Summary`
+- `## Problem Hypothesis`
+- `## Beneficiary Hypothesis`
+- `## Product Category Guess`
+- `## Value Wedge`
+- `## Constraints`
+- `## Non-Goals`
+- `## Assumptions And Unknowns`
+- `## Customer Discovery Readiness`
+- `## Next Steps`
+
+The `## Customer Discovery Readiness` section must state whether the concept is ready for `/customer-discovery`, what inputs `/customer-discovery` should use, and which assumptions should be tested first. If a Market Structure Handoff exists, include the apparent sides and value exchange as explicit inputs for `/customer-discovery` to validate or refute.
+
+The `## Next Steps` section must recommend exactly one primary command:
+
+- If the concept appears to be a business app or user-facing product and the business research lane is not enabled: `npx skillpacks install business-research` from the project shell — this installs the research skills (customer discovery, competitive analysis, value prop, positioning, lean canvas) needed before any repo bootstrapping or development.
+- If `business-research` or the compatibility `business-app` alias is enabled: `/customer-discovery`
+- If the concept already has customer-discovery/market evidence but needs journey, onboarding, conversion, or retention planning: `npx skillpacks install customer-lifecycle` from the project shell.
+- If project type is unclear: `/pack recommend`
+
+Include 1-3 other options only when they are materially useful.
+
+### Alignment Page
+
+When this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page following `ALIGNMENT-PAGE.md` in this skill's directory. Output: `alignment/idea-scope-brief-{topic}.html`.
+
+## Constraints
+
+- Keep the skill short and pre-research.
+- Do not write specs, UX variants, UI specs, roadmap phases, or implementation tasks.
+- Do not recommend `/scaffold` unless the user explicitly asks to create a package/app shell before research; normal product flow scaffolds after research, prototype consolidation, spec, roadmap, and phase planning identify the first implementation target. `/scaffold` requires the monorepo pack (`npx skillpacks install monorepo` from the project shell).
+- Do not update `tasks/todo.md`.
+- New files do not need archive snapshots. Before replacing an existing idea brief, including slugged briefs, archive it to `docs/history/archive/YYYY-MM-DD/HHMMSS/<original-relative-path>`.
+- Migration: if a project already has `research/concept-brief.md`, `research/concept-brief-interview.md`, or any `research/{slug}/concept-brief*.md` / `research/concept-brief-{slug}*.md` from a prior run, rename it to the `idea-brief` equivalent before re-running. Write only the `idea-brief` names and no longer recognizes the legacy `concept-brief` filenames.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/{global → base}/claude/init-agentic-skills/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## v0.10 - 2026-06-13
+
+- Sunset the user-home (global) install path. Base skills now install **project-local** via `npx skillpacks init` only; there is no `init.sh`, `--global`, or `~/.claude/skills` base install. Rewrote the process around `npx skillpacks` (`init`/`refresh`/`status`/`doctor`) and added an `uninstall-global` mode that runs `npx skillpacks uninstall-global` to clean up legacy repo-managed installs left in `~/.claude/skills` and `~/.codex/skills`. Dropped the `update`/`latest` checkout-refresh flow and the first-run drift-preference prompts that depended on the retired launcher.
+
+## v0.9 - 2026-06-13
+
+- Renamed the source skill set from "global" to "base": canonical sources moved to `base/<tool>/<skill>`, and drift/refresh prose now says "base skills". The `--global`/`init-global` install-location flags are unchanged.
+
+## v0.8 - 2026-06-12
+
+- Standardized active pack and skill install guidance on `npx skillpacks install <pack-or-skill>` instead of agent-native `/pack install` or `$pack install` recommendations.
+
 ## v0.7 - 2026-06-11
 
 - Added npm-aware project-shell pack install guidance alongside Claude `/pack install` guidance for project-local domain workflows.

package/base/claude/init-agentic-skills/SKILL.md

@@ -0,0 +1,41 @@
+---
+name: init-agentic-skills
+description: Make the base agentic-skills available in a project by installing them project-local with npx skillpacks init, report and fix skill-install drift, clean up legacy user-home installs, and route domain packs to the pack skill. Use when Claude needs to enable base skills for a project, repair stale installs, confirm initialization status, remove old global installs, or explain how to enable packs.
+type: ops
+version: v0.10
+argument-hint: "[init|status|doctor|refresh|uninstall-global]"
+---
+
+# Init Agentic Skills
+
+Make the base agentic-skills available by installing them **project-local** into the current project's `.claude/skills` and `.codex/skills`. Installation is distributed through the npm `skillpacks` package — there is no user-home (global) install path. Active installs are repo-managed directories that expose the canonical `SKILL.md` and exclude `archive/`; pinned installs intentionally point at `archive/<version>`. Keep domain packs project-local too; use `/pack` from the target project or `npx skillpacks install <pack-or-skill>` when packs are needed.
+
+## Process
+
+1. Parse `$ARGUMENTS`:
+   - No args, `init`, or `refresh`: run `npx skillpacks init` from the project root to install the base skills project-local (`.claude/skills` + `.codex/skills`) and enable `base_skills` in `.agents/project.json`. Use `npx skillpacks refresh` instead when the project is already initialized and you only need to recreate roots from `.agents/project.json`.
+   - `status`: run `npx skillpacks status` to report the project designation, enabled packs, and installed skills.
+   - `doctor`: run `npx skillpacks doctor` to report project skill-install drift (read-only). It compares installed `.claude/skills` and `.codex/skills` copies against their canonical sources via the `.agentic-skills-managed` marker's `source_sha`. Report `STALE` skills with the printed fix command, `unknown` skills as needing a refresh to enable tracking, `missing` skills whose source is gone, and `pinned` skills as frozen. Run `npx skillpacks doctor --fix` to clean generated skill-root drift.
+   - `uninstall-global`: run `npx skillpacks uninstall-global` only after confirming the user wants legacy repo-managed base installs removed from `~/.claude/skills` and `~/.codex/skills`. This cleans up installs created by the retired user-home init path; it removes only skillpacks-owned installs and leaves unmanaged directories untouched.
+   - `help`, `--help`, or `-h`: run `npx skillpacks --help`.
+2. Report the command run, the project skill directories, installed counts, and any warnings about non-repo-managed collisions.
+3. Explain pack access separately:
+   - Do not install `packs/*` as base skills.
+   - In the project that needs domain workflows, run `/pack` for guided setup, or `npx skillpacks install <pack-or-skill>` from the project shell for an explicit pack or individual pack skill.
+   - If a project already has `.agents/project.json`, use `/pack refresh` (or `npx skillpacks refresh`) after this base initialization to recreate local pack skill roots.
+4. If the active session still cannot see a newly installed or refreshed skill, report the skill-visibility reload path:
+   - Claude Code: run `/reload-skills` first; `/clear` starts a new empty-context conversation and 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: start a fresh Codex CLI session if the `$` skill list remains stale.
+
+## Constraints
+
+- Drive installation through `npx skillpacks` (the npm `skillpacks` CLI); do not recreate install logic by hand or reintroduce a user-home/global install path.
+- Base skills are project-local: they install into the current project's `.claude/skills` and `.codex/skills`, not into `~/.claude/skills` or `~/.codex/skills`.
+- Do not overwrite real directories or files under `.claude/skills` or `.codex/skills` that are not repo-managed.
+- Treat packs as project-local capabilities managed through `/pack` (`npx skillpacks` under the hood).
+- Stop and report exact errors if `npx skillpacks` fails.
+- Do not install `packs/*` as base skills in any mode.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/base/claude/init-agentic-skills/archive/v0.8/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: init-agentic-skills
+description: Initialize this agentic-skills checkout on a developer machine by running the repository init script, verify global managed skill installs, uninstall repo-managed installs when requested, and route project-local pack setup to the pack skill. Use when Claude needs to make all global agentic-skills skills available on the current machine, repair stale installs, confirm initialization status, or explain how to enable packs for a project.
+type: ops
+version: v0.8
+argument-hint: "[init|status|doctor|update|latest|--uninstall]"
+---
+
+# Init Agentic Skills
+
+Initialize this `agentic-skills` checkout on the current machine by installing global core skills for both Claude and Codex. Active installs are repo-managed directories that expose the canonical `SKILL.md` and exclude `archive/`; copied managed launchers resolve this checkout through `.agentic-skills-managed` provenance. Pinned installs intentionally point at `archive/<version>`. Keep domain packs project-local; use `/pack` from the target project or `npx skillpacks install <pack-or-skill>` from the project shell when packs are needed.
+
+## Process
+
+1. Parse `$ARGUMENTS`:
+   - No args, `init`, `refresh`, or `sync`: run `scripts/init-agentic-skills.sh`, then run the first-run setup prompts in step 6 when the drift preferences are still unset.
+   - `status`: run `scripts/init-agentic-skills.sh status`, then report the local checkout commit, remote URL, and whether GitHub freshness is enabled by `~/.agentic-skills/preferences.json` at `sync.github_freshness_check`.
+   - `doctor`: run `scripts/init-agentic-skills.sh doctor` to report global skill-install drift (read-only). It compares installed `~/.claude/skills` and `~/.codex/skills` copies against canonical `global/<tool>/<skill>` via the `.agentic-skills-managed` marker's `source_sha`. Report `stale` skills with the fix `/init-agentic-skills update`, `unknown` skills as needing a re-init to enable tracking, and `pinned` skills as frozen.
+   - `update` or `latest`: run `scripts/init-agentic-skills.sh update` only after confirming the user wants to check GitHub, fast-forward the local checkout, and rerun `init.sh`. This re-copy is the global "refresh" — it rewrites markers with current `source_version`/`source_sha`, clearing drift.
+   - `--uninstall`, `uninstall`, or `remove`: run `scripts/init-agentic-skills.sh --uninstall` only after confirming the user wants repo-managed global installs removed.
+   - `help`, `--help`, or `-h`: run `scripts/init-agentic-skills.sh --help`.
+2. Report the initializer command, the Claude and Codex skill directories, installed/skipped counts, and warnings about non-repo-managed collisions.
+3. Explain pack access separately:
+   - Do not install `packs/*` globally.
+   - In the project that needs domain workflows, run `/pack` for guided setup, or `npx skillpacks install <pack-or-skill>` from the project shell for an explicit pack or individual pack skill.
+   - If a project already has `.agents/project.json`, use `/pack refresh` after this global initialization to recreate local pack skill roots.
+4. If the active session still cannot see a newly installed or refreshed skill, report the skill-visibility reload path:
+   - Claude Code: run `/reload-skills` first; `/clear` starts a new empty-context conversation and 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: start a fresh Codex CLI session if the `$` skill list remains stale.
+5. For `update` / `latest` mode:
+   - Confirm before running commands that fetch, pull, or reinstall.
+   - Fetch GitHub freshness, compare the local checkout to `origin/HEAD`, and update only with a fast-forward-only operation such as `git pull --ff-only` or an equivalent `git fetch` + `git merge --ff-only`.
+   - If the update cannot fast-forward cleanly, stop and report the exact reason without rebasing, merging, stashing, or force-resetting.
+   - After a successful fast-forward, rerun `init.sh`.
+   - Warn that Claude Code should use `/reload-skills` first, then `/clear` or restart if needed, and that Codex should start a fresh Codex CLI session if the `$` skill list remains stale.
+6. First-run drift preferences (init / clone setup):
+   - Default policy is **track-latest** (installed skills follow canonical; drift is observable). The two prompts below are **opt-in and off by default**, so the explicit-control default stays out of the box.
+   - Only prompt when the preference keys are still unset. Check with `scripts/init-agentic-skills.sh show-prefs`; skip silently when both are already set.
+   - Ask (a): "Enable a session-start hook that warns when tracked skills are behind canonical?" Default **no**.
+     - If yes: run `scripts/init-agentic-skills.sh hook enable` (registers a `SessionStart` hook in `~/.claude/settings.json` invoking `scripts/skill-drift-hook.sh` and sets `skills.session_start_hook=true`).
+     - If no: run `scripts/init-agentic-skills.sh set-pref session_start_hook false`.
+   - Ask (b): "Auto-refresh latest skills on session start (instead of just warning)?" Default **no**.
+     - Run `scripts/init-agentic-skills.sh set-pref auto_refresh <true|false>` with the answer. Auto-refresh only takes effect when the hook is enabled.
+   - To disable later: decline the prompt, run `scripts/init-agentic-skills.sh set-pref session_start_hook false`, or run `scripts/init-agentic-skills.sh hook disable` to remove the settings entry.
+
+## Constraints
+
+- Delegate global initialization to `scripts/init-agentic-skills.sh`; do not recreate install logic by hand.
+- Do not overwrite real directories or files under `~/.claude/skills`, `~/.codex/skills`, `.claude/skills`, or `.codex/skills`.
+- Treat packs as project-local capabilities managed by `scripts/pack.sh` through `/pack`.
+- Stop and report exact errors if the launcher cannot resolve the repository root or if root `init.sh` fails.
+- Do not install `packs/*` globally in any mode.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/base/claude/init-agentic-skills/archive/v0.9/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: init-agentic-skills
+description: Initialize this agentic-skills checkout on a developer machine by running the repository init script, verify base managed skill installs, uninstall repo-managed installs when requested, and route project-local pack setup to the pack skill. Use when Claude needs to make all base agentic-skills skills available on the current machine, repair stale installs, confirm initialization status, or explain how to enable packs for a project.
+type: ops
+version: v0.9
+argument-hint: "[init|status|doctor|update|latest|--uninstall]"
+---
+
+# Init Agentic Skills
+
+Initialize this `agentic-skills` checkout on the current machine by installing base skills for both Claude and Codex. Active installs are repo-managed directories that expose the canonical `SKILL.md` and exclude `archive/`; copied managed launchers resolve this checkout through `.agentic-skills-managed` provenance. Pinned installs intentionally point at `archive/<version>`. Keep domain packs project-local; use `/pack` from the target project or `npx skillpacks install <pack-or-skill>` from the project shell when packs are needed.
+
+## Process
+
+1. Parse `$ARGUMENTS`:
+   - No args, `init`, `refresh`, or `sync`: run `scripts/init-agentic-skills.sh`, then run the first-run setup prompts in step 6 when the drift preferences are still unset.
+   - `status`: run `scripts/init-agentic-skills.sh status`, then report the local checkout commit, remote URL, and whether GitHub freshness is enabled by `~/.agentic-skills/preferences.json` at `sync.github_freshness_check`.
+   - `doctor`: run `scripts/init-agentic-skills.sh doctor` to report base skill-install drift (read-only). It compares installed `~/.claude/skills` and `~/.codex/skills` copies against canonical `base/<tool>/<skill>` via the `.agentic-skills-managed` marker's `source_sha`. Report `stale` skills with the fix `/init-agentic-skills update`, `unknown` skills as needing a re-init to enable tracking, and `pinned` skills as frozen.
+   - `update` or `latest`: run `scripts/init-agentic-skills.sh update` only after confirming the user wants to check GitHub, fast-forward the local checkout, and rerun `init.sh`. This re-copy is the base "refresh" — it rewrites markers with current `source_version`/`source_sha`, clearing drift.
+   - `--uninstall`, `uninstall`, or `remove`: run `scripts/init-agentic-skills.sh --uninstall` only after confirming the user wants repo-managed base installs removed.
+   - `help`, `--help`, or `-h`: run `scripts/init-agentic-skills.sh --help`.
+2. Report the initializer command, the Claude and Codex skill directories, installed/skipped counts, and warnings about non-repo-managed collisions.
+3. Explain pack access separately:
+   - Do not install `packs/*` as base skills.
+   - In the project that needs domain workflows, run `/pack` for guided setup, or `npx skillpacks install <pack-or-skill>` from the project shell for an explicit pack or individual pack skill.
+   - If a project already has `.agents/project.json`, use `/pack refresh` after this base initialization to recreate local pack skill roots.
+4. If the active session still cannot see a newly installed or refreshed skill, report the skill-visibility reload path:
+   - Claude Code: run `/reload-skills` first; `/clear` starts a new empty-context conversation and 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: start a fresh Codex CLI session if the `$` skill list remains stale.
+5. For `update` / `latest` mode:
+   - Confirm before running commands that fetch, pull, or reinstall.
+   - Fetch GitHub freshness, compare the local checkout to `origin/HEAD`, and update only with a fast-forward-only operation such as `git pull --ff-only` or an equivalent `git fetch` + `git merge --ff-only`.
+   - If the update cannot fast-forward cleanly, stop and report the exact reason without rebasing, merging, stashing, or force-resetting.
+   - After a successful fast-forward, rerun `init.sh`.
+   - Warn that Claude Code should use `/reload-skills` first, then `/clear` or restart if needed, and that Codex should start a fresh Codex CLI session if the `$` skill list remains stale.
+6. First-run drift preferences (init / clone setup):
+   - Default policy is **track-latest** (installed skills follow canonical; drift is observable). The two prompts below are **opt-in and off by default**, so the explicit-control default stays out of the box.
+   - Only prompt when the preference keys are still unset. Check with `scripts/init-agentic-skills.sh show-prefs`; skip silently when both are already set.
+   - Ask (a): "Enable a session-start hook that warns when tracked skills are behind canonical?" Default **no**.
+     - If yes: run `scripts/init-agentic-skills.sh hook enable` (registers a `SessionStart` hook in `~/.claude/settings.json` invoking `scripts/skill-drift-hook.sh` and sets `skills.session_start_hook=true`).
+     - If no: run `scripts/init-agentic-skills.sh set-pref session_start_hook false`.
+   - Ask (b): "Auto-refresh latest skills on session start (instead of just warning)?" Default **no**.
+     - Run `scripts/init-agentic-skills.sh set-pref auto_refresh <true|false>` with the answer. Auto-refresh only takes effect when the hook is enabled.
+   - To disable later: decline the prompt, run `scripts/init-agentic-skills.sh set-pref session_start_hook false`, or run `scripts/init-agentic-skills.sh hook disable` to remove the settings entry.
+
+## Constraints
+
+- Delegate base initialization to `scripts/init-agentic-skills.sh`; do not recreate install logic by hand.
+- Do not overwrite real directories or files under `~/.claude/skills`, `~/.codex/skills`, `.claude/skills`, or `.codex/skills`.
+- Treat packs as project-local capabilities managed by `scripts/pack.sh` through `/pack`.
+- Stop and report exact errors if the launcher cannot resolve the repository root or if root `init.sh` fails.
+- Do not install `packs/*` as base skills in any mode.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/{global → base}/claude/pack/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## v0.10 - 2026-06-13
+
+- Base-skill install reference updated from the retired `init.sh` to project-local `npx skillpacks init`.
+
+## v0.9 - 2026-06-13
+
+- Renamed the domain-neutral core skill set from "global" to "base" in the description and notes (these are the `init.sh`-initialized base skills).
+
+## v0.8 - 2026-06-12
+
+- Standardized active pack and skill install guidance on `npx skillpacks install <pack-or-skill>` instead of agent-native `/pack install` or `$pack install` recommendations.
+
 ## v0.7 - 2026-06-11
 
 - Added published `npx skillpacks install` alternatives to pack and individual-skill install guidance while preserving source-checkout `scripts/pack.sh` and Claude `/pack` routes.