skillpacks 0.1.0

package/packs/gitops/codex/sync/archive/v0.1/SKILL.md

@@ -0,0 +1,102 @@
+---
+name: sync
+description: Pull latest changes from remote and report status
+type: shipping
+version: v0.1
+---
+
+# Sync
+
+Invoke as `$sync`.
+
+Pull the latest changes from the remote repository and report status.
+
+## Workflow
+
+1. Check current state:
+   - Run `git status` to check for uncommitted changes.
+   - If there are uncommitted changes, stash them first, pull, then pop the stash. Warn the user about the stash.
+2. Pull from remote:
+   - Run `git pull --rebase origin <current-branch>`.
+   - If rebase conflicts occur, abort the rebase, try `git pull --no-rebase` instead, and report any merge conflicts for the user to resolve.
+3. Check for outstanding work:
+   - Check if `tasks/roadmap.md` exists for the full plan, and `tasks/todo.md` for the current phase.
+   - If `tasks/todo.md` exists, read it and look for unchecked items (`- [ ]`).
+   - If there are incomplete items, summarise: which phase is current, what the next step is, and how many steps/phases remain.
+   - If `tasks/manual-todo.md` exists, count unchecked manual tasks and include in the summary.
+   - If `tasks/record-todo.md` or `tasks/recurring-todo.md` exists, count unchecked advisory items and include those counts separately. Do not treat them as active plan steps.
+   - If all items are checked, report that the plan is complete.
+   - If neither file exists, note that there is no active plan.
+4. Check provisioning version:
+   - If `CLAUDE.md` or `AGENTS.md` contains `<!-- provision-agentic-config vX.Y -->`, extract the version.
+   - Read `~/.codex/skills/provision-agentic-config/SKILL.md` (or `~/.claude/skills/provision-agentic-config/SKILL.md` as fallback) and extract the `version:` field from the YAML frontmatter.
+   - If the installed skill version is newer than the provisioned version in either file, warn: `⚠ CLAUDE.md provisioned with vX.Y but provision-agentic-config is at vX.Y — consider re-running $provision-agentic-config`
+   - If the version comment is missing from `CLAUDE.md` or `AGENTS.md`, note: `ℹ No provision version found in CLAUDE.md/AGENTS.md — run $provision-agentic-config to add version tracking`
+   - If neither skill file exists, skip this check silently.
+5. Report status:
+   - Branch name
+   - Commits pulled (if any) — show short log of new commits
+   - Whether stashed changes were re-applied
+   - Any conflicts that need manual resolution
+   - Current `git status`
+   - **Outstanding work** — summary from step 3 (next step, current phase, remaining work, pending manual tasks) or "No active plan"
+   - **Advisory tasks** — pending record/recurring counts, if those files exist
+6. Post-sync actions:
+   a) Check if `sync.md` exists at the project root.
+   b) **If `sync.md` exists** — parse and execute it:
+      - Read `sync.md` and identify sections by H2 headings.
+      - **Dependencies** (aliases: "Deps", "Dependency Management"): Execute shell commands in fenced code blocks. Report output briefly.
+      - **Conflict Resolution** (aliases: "Conflicts"): If the pull introduced merge conflicts (from step 2), apply the guidance in this section. If no conflicts, skip silently.
+      - **Custom** (aliases: "Project-Specific", "Scripts", "Setup"): Execute shell commands in fenced code blocks in order. Report output briefly.
+      - **Notifications** (aliases: "Awareness", "Alerts", "Watch"): For each bullet, check if the mentioned file/directory was modified in pulled commits (`git diff --name-only` against pre-pull HEAD). If any match, print a prominent alert.
+      - Unrecognised headings: skip with a note.
+      - Report a summary of all post-sync actions taken.
+   c) **If `sync.md` does not exist** — suggest creating one:
+      - Analyse the project to detect: package manager (look for lockfiles), common scripts (package.json, Makefile, Justfile), config templates (.env.example, docker-compose.yml).
+      - Present a suggested `sync.md` following the format below.
+      - Ask: "Would you like me to create this `sync.md`?"
+      - **Only create the file if the user approves.**
+
+## sync.md format
+
+The `sync.md` file lives at the project root. H2 sections are categories. Shell commands go in fenced code blocks; prose guidance goes in bullet points.
+
+```markdown
+# Post-Sync Actions
+
+## Dependencies
+
+```sh
+npm install
+```
+
+## Conflict Resolution
+
+- Always accept theirs for `package-lock.json`
+
+## Custom
+
+```sh
+npm run codegen
+```
+
+## Notifications
+
+- `.env.example` — check for new environment variables
+- `CLAUDE.md` — review if project conventions were updated
+```
+
+## Constraints
+
+- Do not force-push or rewrite history.
+- Do not auto-resolve merge conflicts — report them and let the user decide. However, if `sync.md` has a Conflict Resolution section, follow its guidance for the specific files/patterns it covers.
+- If stash pop fails due to conflicts, leave the stash intact and report it.
+- Post-sync commands from `sync.md` run in the project root directory.
+- If any post-sync command fails, report the error and continue with remaining actions.
+- Never auto-create `sync.md` without explicit user approval.
+- Do not execute commented-out sections (`<!-- ... -->`).
+
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/packs/gitops/codex/sync/archive/v0.2/SKILL.md

@@ -0,0 +1,114 @@
+---
+name: sync
+description: Pull latest changes from remote and report status
+type: shipping
+version: v0.2
+---
+
+# Sync
+
+Invoke as `$sync`.
+
+Pull the latest changes from the remote repository and report status.
+
+## Workflow
+
+1. Check current state:
+   - Run `git status` to check for uncommitted changes.
+   - If there are uncommitted changes, stash them first, pull, then pop the stash. Warn the user about the stash.
+2. Pull from remote:
+   - Run `git pull --rebase origin <current-branch>`.
+   - If rebase conflicts occur, abort the rebase, try `git pull --no-rebase` instead, and report any merge conflicts for the user to resolve.
+3. Check for outstanding work:
+   - Check if `tasks/roadmap.md` exists for the full plan, and `tasks/todo.md` for the current phase.
+   - If `tasks/todo.md` exists, read it and look for unchecked items (`- [ ]`).
+   - If there are incomplete items, summarise: which phase is current, what the next step is, and how many steps/phases remain.
+   - If `tasks/manual-todo.md` exists, count unchecked manual tasks and include in the summary.
+   - If `tasks/record-todo.md` or `tasks/recurring-todo.md` exists, count unchecked advisory items and include those counts separately. Do not treat them as active plan steps.
+   - If all items are checked, report that the plan is complete.
+   - If neither file exists, note that there is no active plan.
+4. Check provisioned agent config:
+   - If `CLAUDE.md` or `AGENTS.md` contains `<!-- provision-agentic-config vX.Y -->`, extract the version.
+   - Read the canonical `provision-agentic-config` skill from the first existing path in this order:
+     1. `~/.codex/skills/provision-agentic-config/SKILL.md`
+     2. `~/.claude/skills/provision-agentic-config/SKILL.md`
+     3. `global/codex/provision-agentic-config/SKILL.md` in the current repo, when present
+     4. `global/claude/provision-agentic-config/SKILL.md` in the current repo, when present
+   - Extract the `version:` field from the canonical skill's YAML frontmatter.
+   - Extract the canonical provisioned blocks from the same skill:
+     - `CLAUDE.md`: the fenced block under `Required Claude Block` or the section that says "The Claude block to insert into `./CLAUDE.md`".
+     - `AGENTS.md`: the fenced block under `Required AGENTS Block` or the section that says "The AGENTS block to insert into `./AGENTS.md`".
+   - Compare each existing project file against its corresponding canonical block after normalizing line endings and trimming only leading/trailing whitespace around the block. Do not ignore changed bullets, headings, command examples, or policy text.
+   - If the installed skill version is newer than the provisioned version in either file, warn: `⚠ CLAUDE.md provisioned with vX.Y but provision-agentic-config is at vX.Y — consider re-running $provision-agentic-config`
+   - If the version comment is missing from `CLAUDE.md` or `AGENTS.md`, note: `ℹ No provision version found in CLAUDE.md/AGENTS.md — run $provision-agentic-config to add version tracking`
+   - If a project file has the current version comment but the provisioned block content differs from the canonical block, warn: `⚠ CLAUDE.md provisioned block differs from the canonical provision-agentic-config vX.Y block — re-run $provision-agentic-config`
+   - If a canonical block cannot be extracted, fall back to the version-only check and note: `ℹ Could not extract canonical provision-agentic-config block; checked version comment only`
+   - If none of the canonical skill files exists, skip this check silently.
+5. Report status:
+   - Branch name
+   - Commits pulled (if any) — show short log of new commits
+   - Whether stashed changes were re-applied
+   - Any conflicts that need manual resolution
+   - Current `git status`
+   - **Agent config drift** — provisioning version and canonical block match/drift status for `CLAUDE.md` and `AGENTS.md`, if checked
+   - **Outstanding work** — summary from step 3 (next step, current phase, remaining work, pending manual tasks) or "No active plan"
+   - **Advisory tasks** — pending record/recurring counts, if those files exist
+6. Post-sync actions:
+   a) Check if `sync.md` exists at the project root.
+   b) **If `sync.md` exists** — parse and execute it:
+      - Read `sync.md` and identify sections by H2 headings.
+      - **Dependencies** (aliases: "Deps", "Dependency Management"): Execute shell commands in fenced code blocks. Report output briefly.
+      - **Conflict Resolution** (aliases: "Conflicts"): If the pull introduced merge conflicts (from step 2), apply the guidance in this section. If no conflicts, skip silently.
+      - **Custom** (aliases: "Project-Specific", "Scripts", "Setup"): Execute shell commands in fenced code blocks in order. Report output briefly.
+      - **Notifications** (aliases: "Awareness", "Alerts", "Watch"): For each bullet, check if the mentioned file/directory was modified in pulled commits (`git diff --name-only` against pre-pull HEAD). If any match, print a prominent alert.
+      - Unrecognised headings: skip with a note.
+      - Report a summary of all post-sync actions taken.
+   c) **If `sync.md` does not exist** — suggest creating one:
+      - Analyse the project to detect: package manager (look for lockfiles), common scripts (package.json, Makefile, Justfile), config templates (.env.example, docker-compose.yml).
+      - Present a suggested `sync.md` following the format below.
+      - Ask: "Would you like me to create this `sync.md`?"
+      - **Only create the file if the user approves.**
+
+## sync.md format
+
+The `sync.md` file lives at the project root. H2 sections are categories. Shell commands go in fenced code blocks; prose guidance goes in bullet points.
+
+```markdown
+# Post-Sync Actions
+
+## Dependencies
+
+```sh
+npm install
+```
+
+## Conflict Resolution
+
+- Always accept theirs for `package-lock.json`
+
+## Custom
+
+```sh
+npm run codegen
+```
+
+## Notifications
+
+- `.env.example` — check for new environment variables
+- `CLAUDE.md` — review if project conventions were updated
+```
+
+## Constraints
+
+- Do not force-push or rewrite history.
+- Do not auto-resolve merge conflicts — report them and let the user decide. However, if `sync.md` has a Conflict Resolution section, follow its guidance for the specific files/patterns it covers.
+- If stash pop fails due to conflicts, leave the stash intact and report it.
+- Post-sync commands from `sync.md` run in the project root directory.
+- If any post-sync command fails, report the error and continue with remaining actions.
+- Never auto-create `sync.md` without explicit user approval.
+- Do not execute commented-out sections (`<!-- ... -->`).
+
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/packs/gitops/codex/sync/archive/v0.3/SKILL.md

@@ -0,0 +1,142 @@
+---
+name: sync
+description: Pull latest changes from remote and report status
+type: shipping
+version: v0.3
+---
+
+# Sync
+
+Invoke as `$sync`.
+
+Pull the latest changes from the remote repository and report status.
+
+## Workflow
+
+1. Check current state:
+   - Run `git status` to check for uncommitted changes.
+   - If there are uncommitted changes, stash them first, pull, then pop the stash. Warn the user about the stash.
+2. Pull from remote:
+   - Run `git pull --rebase origin <current-branch>`.
+   - If rebase conflicts occur, abort the rebase, try `git pull --no-rebase` instead, and report any merge conflicts for the user to resolve.
+3. Check for outstanding work:
+   - Check if `tasks/roadmap.md` exists for the full plan, and `tasks/todo.md` for the current phase.
+   - If `tasks/todo.md` exists, read it and look for unchecked items (`- [ ]`).
+   - If there are incomplete items, summarise: which phase is current, what the next step is, and how many steps/phases remain.
+   - If `tasks/manual-todo.md` exists, count unchecked manual tasks and include in the summary.
+   - If `tasks/record-todo.md` or `tasks/recurring-todo.md` exists, count unchecked advisory items and include those counts separately. Do not treat them as active plan steps.
+   - If all items are checked, report that the plan is complete.
+   - If neither file exists, note that there is no active plan.
+4. Check provisioned agent config:
+   - If `CLAUDE.md` or `AGENTS.md` contains `<!-- provision-agentic-config vX.Y -->`, extract the version.
+   - Read the canonical `provision-agentic-config` skill from the first existing path in this order:
+     1. `~/.codex/skills/provision-agentic-config/SKILL.md`
+     2. `~/.claude/skills/provision-agentic-config/SKILL.md`
+     3. `global/codex/provision-agentic-config/SKILL.md` in the current repo, when present
+     4. `global/claude/provision-agentic-config/SKILL.md` in the current repo, when present
+   - Extract the `version:` field from the canonical skill's YAML frontmatter.
+   - Extract the canonical provisioned blocks from the same skill:
+     - `CLAUDE.md`: the fenced block under `Required Claude Block` or the section that says "The Claude block to insert into `./CLAUDE.md`".
+     - `AGENTS.md`: the fenced block under `Required AGENTS Block` or the section that says "The AGENTS block to insert into `./AGENTS.md`".
+   - Compare each existing project file against its corresponding canonical block after normalizing line endings and trimming only leading/trailing whitespace around the block. Do not ignore changed bullets, headings, command examples, or policy text.
+   - If the installed skill version is newer than the provisioned version in either file, warn: `⚠ CLAUDE.md provisioned with vX.Y but provision-agentic-config is at vX.Y — consider re-running $provision-agentic-config`
+   - If the version comment is missing from `CLAUDE.md` or `AGENTS.md`, note: `ℹ No provision version found in CLAUDE.md/AGENTS.md — run $provision-agentic-config to add version tracking`
+   - If a project file has the current version comment but the provisioned block content differs from the canonical block, warn: `⚠ CLAUDE.md provisioned block differs from the canonical provision-agentic-config vX.Y block — re-run $provision-agentic-config`
+   - If a canonical block cannot be extracted, fall back to the version-only check and note: `ℹ Could not extract canonical provision-agentic-config block; checked version comment only`
+   - If none of the canonical skill files exists, skip this check silently.
+   - Always report the local canonical source path used and its `version:` field when this check runs.
+5. Resolve GitHub freshness preference:
+   - Use the user-local machine-wide preference file at `~/.agentic-skills/preferences.json`.
+   - Read `sync.github_freshness_check`, whose only allowed values are `"ask"`, `"always"`, and `"never"`.
+   - If the file or key is missing, ask the user once which default to remember:
+     - Always check GitHub during sync
+     - Never check GitHub during sync
+     - Ask each time
+   - Create `~/.agentic-skills/preferences.json` if needed and save the selected value as:
+     ```json
+     {
+       "sync": {
+         "github_freshness_check": "ask"
+       }
+     }
+     ```
+   - If the value is `"always"`, check GitHub remote freshness automatically.
+   - If the value is `"never"`, skip GitHub freshness checks and report that local canonical `provision-agentic-config` was used without a GitHub check.
+   - If the value is `"ask"`, ask before checking GitHub for this sync.
+   - Treat malformed JSON or an unsupported value as missing preference and ask again before writing one of the allowed values.
+6. Optional GitHub freshness check:
+   - Only run this check when the resolved preference or explicit user approval says to check GitHub.
+   - Compare the local `agentic-skills` checkout against `origin/HEAD` using non-mutating commands such as `git remote get-url origin`, `git rev-parse HEAD`, `git rev-parse origin/HEAD`, and, when remote freshness is explicitly enabled, `git fetch --dry-run` or an equivalent non-mutating freshness probe.
+   - Report the local checkout commit, remote URL, local `origin/HEAD` commit if available, and whether the local checkout appears behind the remote.
+   - Do not pull, fast-forward, rebase, install, or mutate the checkout from plain `$sync`.
+   - If a GitHub check shows the local checkout is stale, recommend `$init-agentic-skills update` for an explicit update.
+7. Report status:
+   - Branch name
+   - Commits pulled (if any) — show short log of new commits
+   - Whether stashed changes were re-applied
+   - Any conflicts that need manual resolution
+   - Current `git status`
+   - **Agent config drift** — provisioning version, local canonical source path/version, and canonical block match/drift status for `CLAUDE.md` and `AGENTS.md`, if checked
+   - **GitHub freshness** — preference value, whether GitHub was checked, and the local checkout/remote status; when skipped, say local canonical was used
+   - **Outstanding work** — summary from step 3 (next step, current phase, remaining work, pending manual tasks) or "No active plan"
+   - **Advisory tasks** — pending record/recurring counts, if those files exist
+8. Post-sync actions:
+   a) Check if `sync.md` exists at the project root.
+   b) **If `sync.md` exists** — parse and execute it:
+      - Read `sync.md` and identify sections by H2 headings.
+      - **Dependencies** (aliases: "Deps", "Dependency Management"): Execute shell commands in fenced code blocks. Report output briefly.
+      - **Conflict Resolution** (aliases: "Conflicts"): If the pull introduced merge conflicts (from step 2), apply the guidance in this section. If no conflicts, skip silently.
+      - **Custom** (aliases: "Project-Specific", "Scripts", "Setup"): Execute shell commands in fenced code blocks in order. Report output briefly.
+      - **Notifications** (aliases: "Awareness", "Alerts", "Watch"): For each bullet, check if the mentioned file/directory was modified in pulled commits (`git diff --name-only` against pre-pull HEAD). If any match, print a prominent alert.
+      - Unrecognised headings: skip with a note.
+      - Report a summary of all post-sync actions taken.
+   c) **If `sync.md` does not exist** — suggest creating one:
+      - Analyse the project to detect: package manager (look for lockfiles), common scripts (package.json, Makefile, Justfile), config templates (.env.example, docker-compose.yml).
+      - Present a suggested `sync.md` following the format below.
+      - Ask: "Would you like me to create this `sync.md`?"
+      - **Only create the file if the user approves.**
+
+## sync.md format
+
+The `sync.md` file lives at the project root. H2 sections are categories. Shell commands go in fenced code blocks; prose guidance goes in bullet points.
+
+```markdown
+# Post-Sync Actions
+
+## Dependencies
+
+```sh
+npm install
+```
+
+## Conflict Resolution
+
+- Always accept theirs for `package-lock.json`
+
+## Custom
+
+```sh
+npm run codegen
+```
+
+## Notifications
+
+- `.env.example` — check for new environment variables
+- `CLAUDE.md` — review if project conventions were updated
+```
+
+## Constraints
+
+- Do not force-push or rewrite history.
+- Do not auto-resolve merge conflicts — report them and let the user decide. However, if `sync.md` has a Conflict Resolution section, follow its guidance for the specific files/patterns it covers.
+- If stash pop fails due to conflicts, leave the stash intact and report it.
+- Post-sync commands from `sync.md` run in the project root directory.
+- If any post-sync command fails, report the error and continue with remaining actions.
+- Never auto-create `sync.md` without explicit user approval.
+- Do not execute commented-out sections (`<!-- ... -->`).
+- Plain `$sync` must not update the `agentic-skills` checkout, pull from GitHub for the local canonical source, or reinstall skills. Only `$init-agentic-skills update` / `$init-agentic-skills latest` may perform that explicit update flow after confirmation.
+
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.

package/packs/guided-walkthrough/PACK.md

@@ -0,0 +1,22 @@
+# Guided Walkthrough Pack
+
+Guided walkthrough skills for producing human-facing step-by-step instructions. Covers general guides and UAT-specific walkthroughs.
+
+Install this pack when creating step-by-step guides or UAT walkthroughs for end users.
+
+Install in a project with:
+
+```bash
+scripts/pack.sh install guided-walkthrough
+```
+
+Then refresh local skill links if needed:
+
+```bash
+scripts/pack.sh refresh
+```
+
+## Skills
+
+- `guide`: Create a step-by-step guided walkthrough for a feature or workflow.
+- `uat-guide`: Create a UAT-specific guided walkthrough with acceptance criteria and validation steps.

package/packs/guided-walkthrough/claude/guide/CHANGELOG.md

@@ -0,0 +1,9 @@
+# Changelog
+
+## v0.0
+
+- Archived previous skill contract.
+
+## v0.1 - 2026-05-25
+
+- Added research-quality alignment requirements covering claim/evidence/inference separation, no-context-loss HTML translation, evidence matrices, confidence/assumption registers, source coverage, and research completeness gates.

package/packs/guided-walkthrough/claude/guide/SKILL.md

@@ -0,0 +1,110 @@
+---
+name: guide
+description: Click-by-click instructions for manual blockers — DNS, OAuth, signups, and other GUI-dependent tasks
+type: analysis
+version: v0.1
+argument-hint: "[optional: task description or manual-todo item text]"
+---
+
+# Guide
+
+Produce detailed, click-by-click instructions for human-only manual tasks that block automated progress — service signups, DNS configuration, OAuth credential creation with no authenticated CLI/API path, dashboard settings that cannot be scripted, production smoke checks needing a real account/device or human sign-off, and anything else that requires a third-party portal or subjective human judgment.
+
+Not for development-document bookkeeping. If the requested task is
+auditing, reconciling, checking off, moving, or classifying entries in
+`tasks/manual-todo.md`, `tasks/todo.md`, `tasks/record-todo.md`, or history
+against repo reality, stop and route to `/reconcile-dev-docs fix tasks` or a
+direct dev-doc audit instead.
+
+Also not for agent-executable work. If the task is repo edits,
+SDK wiring, generated assets, local commands, tests, audits, or CLI/API work
+with available authentication, stop and route it to `/exec`, `/ship`, or the
+specific implementation skill instead of producing click instructions.
+
+## Process
+
+1. **Find the blocker:**
+   - If `$ARGUMENTS` is non-empty, use the provided text as the task description.
+   - If `$ARGUMENTS` explicitly references `tasks/record-todo.md`, use that record item as the task description and produce instructions for satisfying its condition or collecting its evidence. Do not use record items by default.
+   - If `$ARGUMENTS` is empty, read `tasks/manual-todo.md`:
+     - Find the first unchecked item (`- [ ]`) that contains a `_(blocks: ...)_` annotation.
+     - If no blocking item exists, find the first unchecked item.
+     - If the file does not exist or has no unchecked items, ask the user what they need help with.
+
+2. **Gather project context:**
+   - Read `tasks/todo.md`, `tasks/roadmap.md`, and `CLAUDE.md` (where they exist) to understand what the manual task is for — what service, what domain, what credentials are expected and where they will be consumed.
+   - Search the codebase for references to the service or credentials (env var names, config keys, import paths) to identify:
+     - Exact values the user will need to fill in (callback URLs, domain names, scopes).
+     - Where outputs should be stored (`.env`, config files, secrets manager).
+
+3. **Research current instructions:**
+   - Use web search to find up-to-date instructions for the specific service or platform.
+   - Prioritize official documentation over blog posts.
+   - Service UIs change frequently — never rely solely on prior knowledge; always search.
+
+4. **Produce click-by-click guide:**
+   - Write numbered steps with:
+     - Exact UI elements to click (button text, menu paths, tab names).
+     - Fields to fill and the specific values to use, drawn from project context (e.g., "Set **Authorized redirect URI** to `http://localhost:3000/auth/callback`" not just "set a redirect URI").
+     - What to look for on each screen to confirm you're in the right place.
+     - Warnings about common gotchas or easy-to-miss settings.
+   - Group steps into logical phases when the task spans multiple screens or services.
+   - If the task involves waiting (DNS propagation, approval queues), say how long and what to do in the meantime.
+
+5. **Show where to put results:**
+   - For each output the task produces (API keys, client IDs, tokens, URLs, DNS records), specify:
+     - The exact file path where it belongs (e.g., `.env`, `config/auth.ts`).
+     - The exact variable or key name to set.
+     - Example format if it matters (e.g., "paste the full JSON key file, not just the client ID").
+   - If the destination file does not exist yet, say so and show what it should look like.
+
+6. **Offer to check off:**
+   - After presenting the guide, tell the user: "When you've completed these steps, let me know and I'll mark the item done in `tasks/manual-todo.md`."
+   - When the user confirms, check off the item (`- [x]`) in `tasks/manual-todo.md`.
+
+## Output
+
+```
+## Guide: [Task title]
+
+**Context**: [One sentence — why this task is needed and what it unblocks]
+
+### Steps
+
+1. Go to [URL].
+2. Click **[Button/Link text]** in the [location description].
+3. Fill in **[Field name]**: `[exact value from project context]`
+   ...
+
+### Where to put the results
+
+| Output | File | Key / Variable |
+|--------|------|----------------|
+| Client ID | `.env` | `OAUTH_CLIENT_ID` |
+| Client Secret | `.env` | `OAUTH_CLIENT_SECRET` |
+
+### Gotchas
+
+- [Common mistake or easy-to-miss setting]
+
+---
+
+When you've completed these steps, let me know and I'll mark the item done in `tasks/manual-todo.md`.
+```
+
+## Constraints
+
+- **Always web search** — never produce instructions from memory alone. Service UIs change; stale steps are worse than none.
+- **Project-specific values** — never give generic placeholders when the codebase contains the actual values to use.
+- **Read-only by default** — this skill does not modify code. The only file it may edit is `tasks/manual-todo.md` (to check off a completed item), and only after the user confirms completion.
+- Do not check off or modify `tasks/record-todo.md`; record items are advisory unless a separate workflow updates them.
+- **No shipping contract** — checking off a manual-todo item is a minor bookkeeping edit, not a code change. Do not auto-commit just for that. If other tracked changes are present, leave them for a proper shipping skill.
+- **One task at a time** — if multiple blockers exist, guide the user through the first one. They can run `/guide` again for the next.
+- **No task-ledger reconciliation** — do not produce a guide for requests such as "reconcile manual-todo", "audit stale manual tasks", or "check off completed todo items". Route those to `/reconcile-dev-docs fix tasks` or a direct dev-doc audit.
+- **No agent-executable work** — do not produce a guide for repo edits, local commands, SDK adoption, generated assets, test runs, Lighthouse/Playwright checks, or authenticated CLI/API operations.
+- **Don't execute the task** — produce instructions for the user to follow. Do not attempt to call external APIs, create accounts, or configure services on the user's behalf.
+
+## Default Shipping Contract
+
+Follow the shared shipping contract convention in CLAUDE.md.
+

package/packs/guided-walkthrough/claude/guide/archive/v0.0/SKILL.md

@@ -0,0 +1,130 @@
+---
+name: guide
+description: Click-by-click instructions for manual blockers — DNS, OAuth, signups, and other GUI-dependent tasks
+type: analysis
+version: v0.0
+argument-hint: "[optional: task description or manual-todo item text]"
+---
+
+# Guide
+
+Produce detailed, click-by-click instructions for human-only manual tasks that block automated progress — service signups, DNS configuration, OAuth credential creation with no authenticated CLI/API path, dashboard settings that cannot be scripted, production smoke checks needing a real account/device or human sign-off, and anything else that requires a third-party portal or subjective human judgment.
+
+This skill is not for development-document bookkeeping. If the requested task is
+auditing, reconciling, checking off, moving, or classifying entries in
+`tasks/manual-todo.md`, `tasks/todo.md`, `tasks/record-todo.md`, or history
+against repo reality, stop and route to `/reconcile-dev-docs fix tasks` or a
+direct dev-doc audit instead.
+
+This skill is also not for agent-executable work. If the task is repo edits,
+SDK wiring, generated assets, local commands, tests, audits, or CLI/API work
+with available authentication, stop and route it to `/run`, `/ship`, or the
+specific implementation skill instead of producing click instructions.
+
+## Process
+
+1. **Find the blocker:**
+   - If `$ARGUMENTS` is non-empty, use the provided text as the task description.
+   - If `$ARGUMENTS` explicitly references `tasks/record-todo.md`, use that record item as the task description and produce instructions for satisfying its condition or collecting its evidence. Do not use record items by default.
+   - If `$ARGUMENTS` is empty, read `tasks/manual-todo.md`:
+     - Find the first unchecked item (`- [ ]`) that contains a `_(blocks: ...)_` annotation.
+     - If no blocking item exists, find the first unchecked item.
+     - If the file does not exist or has no unchecked items, ask the user what they need help with.
+
+2. **Gather project context:**
+   - Read `tasks/todo.md`, `tasks/roadmap.md`, and `CLAUDE.md` (where they exist) to understand what the manual task is for — what service, what domain, what credentials are expected and where they will be consumed.
+   - Search the codebase for references to the service or credentials (env var names, config keys, import paths) to identify:
+     - Exact values the user will need to fill in (callback URLs, domain names, scopes).
+     - Where outputs should be stored (`.env`, config files, secrets manager).
+
+3. **Research current instructions:**
+   - Use web search to find up-to-date instructions for the specific service or platform.
+   - Prioritize official documentation over blog posts.
+   - Service UIs change frequently — never rely solely on prior knowledge; always search.
+
+4. **Produce click-by-click guide:**
+   - Write numbered steps with:
+     - Exact UI elements to click (button text, menu paths, tab names).
+     - Fields to fill and the specific values to use, drawn from project context (e.g., "Set **Authorized redirect URI** to `http://localhost:3000/auth/callback`" not just "set a redirect URI").
+     - What to look for on each screen to confirm you're in the right place.
+     - Warnings about common gotchas or easy-to-miss settings.
+   - Group steps into logical phases when the task spans multiple screens or services.
+   - If the task involves waiting (DNS propagation, approval queues), say how long and what to do in the meantime.
+
+5. **Show where to put results:**
+   - For each output the task produces (API keys, client IDs, tokens, URLs, DNS records), specify:
+     - The exact file path where it belongs (e.g., `.env`, `config/auth.ts`).
+     - The exact variable or key name to set.
+     - Example format if it matters (e.g., "paste the full JSON key file, not just the client ID").
+   - If the destination file does not exist yet, say so and show what it should look like.
+
+6. **Offer to check off:**
+   - After presenting the guide, tell the user: "When you've completed these steps, let me know and I'll mark the item done in `tasks/manual-todo.md`."
+   - When the user confirms, check off the item (`- [x]`) in `tasks/manual-todo.md`.
+
+## Output Format
+
+```
+## Guide: [Task title]
+
+**Context**: [One sentence — why this task is needed and what it unblocks]
+
+### Steps
+
+1. Go to [URL].
+2. Click **[Button/Link text]** in the [location description].
+3. Fill in **[Field name]**: `[exact value from project context]`
+   ...
+
+### Where to put the results
+
+| Output | File | Key / Variable |
+|--------|------|----------------|
+| Client ID | `.env` | `OAUTH_CLIENT_ID` |
+| Client Secret | `.env` | `OAUTH_CLIENT_SECRET` |
+
+### Gotchas
+
+- [Common mistake or easy-to-miss setting]
+
+---
+
+When you've completed these steps, let me know and I'll mark the item done in `tasks/manual-todo.md`.
+```
+
+## Alignment Page
+
+When this skill produces durable deliverables (research, specs, plans, reports, prototypes, or any document output), build a full-depth HTML alignment page at `alignment/guide-{topic}.html`. Use a normalized topic slug derived from the app, feature, research subject, report subject, or output filename.
+
+**Full content requirement.** The alignment page must contain the complete content of every proposed markdown deliverable -- every section, every finding, every detail, every list item. It is a thorough interactive review document, not a summary. Render the full deliverable content in clean, readable HTML with appropriate hierarchy, styling, and navigation. If the skill writes multiple scoped deliverables in one run, build one alignment page that contains all deliverables with anchor-linked navigation. Durable tracker artifacts, such as `research/assumption-tracker.md`, remain canonical markdown outputs but must also be fully rendered into the alignment page before approval.
+
+**Dark-mode styling.** Use a dark color scheme by default. Base CSS variables: `--bg: #0d1117; --surface: #161b22; --border: #30363d; --text: #c9d1d9; --text-muted: #8b949e; --accent: #58a6ff; --green: #3fb950; --red: #f85149; --orange: #d29922; --purple: #bc8cff;`. Apply `background: var(--bg); color: var(--text);` on body. Use `--surface` for cards, nav, and table headers. Use `--border` for all borders. Use `--purple` for question blocks and gate headings. Use `--accent` for links and section headings. Keep headings `color: #fff` or `var(--accent)` for hierarchy. Question block backgrounds should use `#1c2333`.
+
+**Alignment gates.** Treat gates as explicit review sections inside the HTML page. A gate blocks finalization until its required inline questions are answered and compiled into YAML. Include every gate that applies to the skill output, and include these gate types whenever relevant: evidence coverage, assumptions/confidence, scope/non-goals, candidate/verdict decisions, artifact destination, proposed file changes, coverage checkpoint, and post-approval route.
+
+**Report-only research gates.** For report-only or pre-approval research skills, the alignment page must explicitly contain evidence coverage, assumptions/confidence, recommended path, proposed file changes, and approval gates before any canonical research, spec, or task file is created or updated.
+
+
+**Required inline questions.** Each gate must contain at least one required inline question placed directly under the content it governs, inside a visually distinct question block. Each question must use radio-button inputs and include two standing options after the skill-generated choices: "Other / None of the above" backed by a multi-line text box for free-form input, and "Need clarification" backed by an optional notes box where the user can explain what is unclear. When any radio option other than "Other" or "Need clarification" is selected, show an optional "Additional notes" text box beneath it so the user can qualify their choice. Generate questions based on what genuinely needs user input -- do not add filler questions. Do not create a separate bottom "Decisions & Clarifications" section.
+
+**Gate YAML contract.** At the bottom of the page, include a "Compile Answers" button that aggregates answers from all inline gate questions throughout the page, including free-text notes. The button remains disabled until every required question has a selection, shows a count of remaining unanswered questions, and scrolls to the first unanswered question if clicked early. When every question is answered, generate a structured YAML block with one item per gate answer using this stable shape: `section`, `gate_type`, `status` (`answered`, `other`, or `needs-clarification`), `answer`, optional `notes`, and optional `target_artifact` or `target_path` when the gate controls file output. After successful compilation, automatically attempt to copy the YAML to the clipboard with the Clipboard API, display copy status, and display the YAML in a read-only textarea with an explicit "Copy YAML" button. The copy button must retry clipboard copy when supported and fall back to selecting the textarea contents when clipboard access is unavailable or blocked.
+
+**Pre-approval stop.** Before user approval, the next action is review of the HTML alignment page, not downstream routing. Ask the user to review the page and provide the compiled YAML answers. Do not include `Recommended next skill`, `Recommended next command`, or downstream routing language until after compiled YAML has been provided and the approved artifacts have been written or updated.
+
+**Diff highlighting on updates.** When the agent updates an existing alignment page after receiving compiled answers, highlight what changed since the previous version. The agent chooses inline annotation or side-by-side layout per situation.
+
+**Archiving.** Before replacing an existing alignment page, archive it to `docs/history/archive/YYYY-MM-DD/HHMMSS/alignment/guide-{topic}.html`.
+
+**Browser open.** Attempt to open the resulting HTML page in the browser and report whether the open succeeded or was blocked. A blocked browser-open attempt does not make the skill fail when the files were written correctly.
+
+## Constraints
+
+- **Always web search** — never produce instructions from memory alone. Service UIs change; stale steps are worse than none.
+- **Project-specific values** — never give generic placeholders when the codebase contains the actual values to use.
+- **Read-only by default** — this skill does not modify code. The only file it may edit is `tasks/manual-todo.md` (to check off a completed item), and only after the user confirms completion.
+- Do not check off or modify `tasks/record-todo.md`; record items are advisory unless a separate workflow updates them.
+- **No shipping contract** — checking off a manual-todo item is a minor bookkeeping edit, not a code change. Do not auto-commit just for that. If other tracked changes are present, leave them for a proper shipping skill.
+- **One task at a time** — if multiple blockers exist, guide the user through the first one. They can run `/guide` again for the next.
+- **No task-ledger reconciliation** — do not produce a guide for requests such as "reconcile manual-todo", "audit stale manual tasks", or "check off completed todo items". Route those to `/reconcile-dev-docs fix tasks` or a direct dev-doc audit.
+- **No agent-executable work** — do not produce a guide for repo edits, local commands, SDK adoption, generated assets, test runs, Lighthouse/Playwright checks, or authenticated CLI/API operations.
+- **Don't execute the task** — produce instructions for the user to follow. Do not attempt to call external APIs, create accounts, or configure services on the user's behalf.