@jarel/myskills 0.1.0-beta.2 → 0.1.0-beta.20

package/README.md

@@ -25,21 +25,61 @@ Responsibilities:
 - install/export/update/rollback packages
 - submit drafts
 - support maintainer/admin workflows through role-gated API calls
+- enroll a personal Codex workspace and execute approved updates with an explicit companion command
+- inspect an explicitly selected local Codex profile with a read-only metadata observation or health report
+- enroll global and project Codex and Claude skill directories as separate read-only inventory scopes
+- create a dry-run-only work/team bootstrap plan from explicit approved roots and skill selectors
 
 CLI tokens should be stored in the platform secret store where possible.
 
 ## Current Slice
 
+This document describes the `0.1.0-beta.20` source candidate, including resumable browser login, named
+CLI configuration profiles and global/project inventory scopes. Source,
+GitHub releases, npm publication, and hosted deployment are separate states;
+see [release verification](../../docs/RELEASE.md) for their checks. Historical
+[beta.7 delivery evidence](../../docs/BETA7_RELEASE_DELIVERY.md) records that
+release only. The API and web expose their deployed version and commit at
+`/version.json`.
+
+### Beta.3 breaking security changes
+
+Beta.3 tightens the API boundary for team and sharing mutations. Team
+creation, team-owner invitation/member lifecycle changes, and
+sharing expansions to team or organization scope require an interactive
+MFA-verified session. Privileged sharing reads/writes and the deprecated
+`skills edit --visibility` alias use the same session/MFA boundary. API tokens
+remain useful for scoped reads, but they cannot perform these mutations.
+
+Migration before upgrading to beta.3: enroll TOTP through the API's
+`POST /v1/auth/mfa/totp/enroll` and `POST /v1/auth/mfa/totp/confirm` routes
+with password reauthentication, save the one-time recovery codes, then run
+`myskills login` and complete the `POST /v1/auth/mfa/verify` challenge. Move
+mutation automation to an explicitly managed session; keep API-token
+automation read-only. Invitation acceptance remains session-only where the API
+route permits it.
+
+For install guidance, use the generated `myskills install ...` or
+`myskills export ... --output ...` command. The corresponding MCP tool no longer
+returns `apiBundleEndpoint` or a bundle URL, so clients must not construct one;
+the CLI's authenticated export/install path remains the replacement flow.
+Generated commands do not embed an API URL or bearer token; configure the CLI
+with `myskills config set api-url ...` or `MYSKILLS_API_URL`, then authenticate
+separately.
+
 Implemented commands:
 
 ```text
 myskills version
 myskills --version
+myskills init <name> [--output <dir>] [--title <text>] [--summary <text>] [--license <text>] [--json]
 myskills validate --path <file-directory-or-zip>
 myskills scan --path <file-directory-or-zip>
-myskills search [query] [--api-url <url>]
-myskills info <skill-slug> [--api-url <url>]
-myskills login [--api-url <url>] [--method <password|api-key>] [--email <email>]
+myskills package --path <directory> --output <file.zip> [--json]
+myskills search [query] [--limit <1-100>] [--cursor <cursor>] [--api-url <url>]
+myskills info <skill-slug> [--version <exact-version>] [--api-url <url>]
+myskills login [--api-url <url>] [--method <password|api-key|browser>] [--email <email>]
+myskills login --method browser [--api-url <url>] [--scopes <comma-separated-scopes>] [--no-resume]
 myskills login --api-key [--api-url <url>]
 myskills logout [--api-url <url>] [--token <token>]
 myskills whoami [--api-url <url>] [--token <token>]
@@ -49,13 +89,14 @@ myskills config get api-url
 myskills config set api-url <url>
 myskills config reset api-url
 myskills config list
+myskills bootstrap codex --dry-run --profile work --context <file> [--work-source-root <dir>] [--shared-source-root <dir>] --live-root <dir> --include-slug <slug> [--include-slug <slug>] --output <report.json>
 myskills submit --path <file-directory-or-zip> [--api-url <url>] [--token <token>]
-myskills review submissions [--api-url <url>] [--token <token>]
+myskills review submissions [--limit <1-100>] [--cursor <cursor>] [--api-url <url>] [--token <token>]
 myskills review bundle <submission-id> [--platform <name>] [--output <file>] [--api-url <url>] [--token <token>]
 myskills review action <submission-id> --action <approve|request-changes|reject|publish> [--artifact-sha256 <hash>] [--reason <text>] [--api-url <url>] [--token <token>]
 myskills submissions list [--api-url <url>] [--token <token>]
 myskills submissions withdraw <submission-id> [--reason <text>] [--api-url <url>] [--token <token>]
-myskills skills edit <skill-slug> [--title <text>] [--summary <text>] [--visibility <scope>] [--tag <tag>] [--reason <text>] [--api-url <url>] [--token <token>]
+myskills skills edit <skill-slug> [--title <text>] [--summary <text>] [--tag <tag>] [--visibility <scope>] [--reason <text>] [--api-url <url>] [--token <token>] (deprecated visibility compatibility alias; use sharing set)
 myskills skills archive|restore|delete <skill-slug> [--reason <text>] [--api-url <url>] [--token <token>]
 myskills releases list <skill-slug> [--api-url <url>] [--token <token>]
 myskills releases deprecate|unpublish|revoke|restore|delete <skill-slug>@<version> [--reason <text>] [--replacement <version>] [--api-url <url>] [--token <token>]
@@ -64,46 +105,517 @@ myskills teams create <team-name> [--name <team-name>] [--api-url <url>] [--toke
 myskills teams invite <team-id> --email <email> [--api-url <url>] [--token <token>]
 myskills teams accept <invitation-id> [--api-url <url>] [--token <token>]
 myskills sharing get <skill-slug> [--api-url <url>] [--token <token>]
-myskills sharing set <skill-slug> --visibility <scope> [--team <team-id>] [--user <email>]
+myskills sharing set <skill-slug> --visibility <scope> [--team <team-id>] [--user <email>] [--organization <organization-id>] [--organization-id <organization-id>] [--clear-organizations]
+myskills architectures patterns [--api-url <url>] [--token <token>]
+myskills architectures list [--api-url <url>] [--token <token>]
+myskills architectures show <architecture-id> [--revision <revision-id>] [--api-url <url>] [--token <token>]
+myskills architectures preview|compile <architecture-id> [--revision <revision-id>] [--profile <profile-id>] [--environment <environment-id>] [--organization-id <organization-id>|--organization <organization-id>] [--api-url <url>] [--token <token>]
+myskills architectures plan|dry-run <architecture-id> --observed <fixture.json> [--revision <revision-id>] [--profile <profile-id>] [--environment <environment-id>] [--organization-id <organization-id>|--organization <organization-id>] [--api-url <url>] [--token <token>]
+myskills architectures observe --root <absolute-dir> --profile <personal|work|shared> (--context <file> | --target-id <id> --generation <number> --architecture-id <id> --environment-id <id> --profile-id <id> --adapter-digest <sha256> --capabilities-digest <sha256>) [--json]
+myskills architectures health --root <absolute-dir> --profile <personal|work|shared> (--context <file> | --target-id <id> --generation <number> --architecture-id <id> --environment-id <id> --profile-id <id> --adapter-digest <sha256> --capabilities-digest <sha256>) [--json]
 myskills admin sharing get [--api-url <url>] [--token <token>]
-myskills admin sharing set [--public <true|false>] [--authenticated <true|false>] [--teams <true|false>] [--team-visibility <true|false>] [--user-visibility <true|false>]
+myskills admin sharing set [--public <true|false>] [--authenticated <true|false>] [--teams <true|false>] [--team-visibility <true|false>] [--user-visibility <true|false>] [--organization-visibility <true|false>]
 myskills export <skill-slug> --version <version> --platform <platform> --output <dir>
 myskills install <skill-slug> [--version <version>] [--platform <platform>] [--dir <install-root>]
 myskills list [--dir <install-root>]
 myskills update [skill-slug] [--version <version>] [--platform <platform>] [--dir <install-root>]
 myskills rollback <skill-slug> [--dir <install-root>]
+myskills codex enroll --workspace <absolute-dir> --architecture-id <id> --environment-id <id> --profile-id <id> [--name <name>]
+myskills codex observe --workspace <absolute-dir> [--upload] [--json]
+myskills scopes inventory --provider codex|claude --root <absolute-skills-dir> [--json]
+myskills scopes enroll --provider codex|claude --scope global --root <absolute-skills-dir> --architecture-id <id> --environment-id <id> --profile-id <id> [--name <name>]
+myskills scopes enroll --provider codex|claude --scope project --project <absolute-dir> --architecture-id <id> --environment-id <id> --profile-id <id> [--name <name>]
+myskills scopes observe --provider codex|claude --scope global|project [--project <absolute-dir>] [--upload] [--json]
+myskills scopes list [--provider codex|claude]
+myskills scopes resolve --provider codex|claude --path <absolute-dir>
+myskills scopes exclude|include --provider codex|claude --project <absolute-dir>
+myskills scopes unbind --provider codex|claude --scope global|project [--project <absolute-dir>]
+myskills scopes migrate plan|apply --provider codex|claude --project <absolute-dir> [--plan-digest <sha256>]
+myskills install <skill-slug> --version <version> --workspace <absolute-dir>
+myskills update [skill-slug] [--version <version>] --workspace <absolute-dir>
+myskills rollback <skill-slug> --workspace <absolute-dir>
+myskills companion run-once --workspace <absolute-dir> --holder <name>
 myskills token create --name <name> --scope <scope> [--scope <scope>]
 myskills token list
 myskills token revoke <token-id>
 ```
 
-## Published And Candidate Channels
+### Registry and collaboration workflows
+
+The commands below describe the current repository source. Consult the
+[capabilities matrix](../../docs/CAPABILITY_PARITY.md) for coverage and
+verification status; an implemented command does not establish npm publication
+or hosted deployment. All these commands accept `--api-url`, `--token` and
+`--json`. The API enforces current roles, ownership, membership and MFA.
+Use `myskills login` with password/MFA for session-only operations; an API key
+does not replace the required session or MFA verification.
+
+```text
+myskills skills managed [--query <text>] [--limit <1-100>] [--cursor <cursor>]
+myskills submissions show <submission-id>
+myskills submissions export <submission-id> [--platform <name>] [--output <new-file>]
+myskills review show <submission-id>
+myskills teams revoke-invitation <team-id> <invitation-id>
+myskills teams set-role <team-id> <member-id> --role <owner|member>
+myskills teams remove-member <team-id> <member-id>
+myskills organizations list|pending-invitations
+myskills organizations show|members|invitations|policies|teams|update-policy <organization-id>
+myskills organizations create <name> [--slug <slug>]
+myskills organizations create --input <request.json>
+myskills organizations archive <organization-id>
+myskills organizations invite <organization-id> --email <email> [--role <owner|admin|member>]
+myskills organizations accept <invitation-id>
+myskills organizations set-role <organization-id> <member-id> --role <owner|admin|member>
+myskills organizations remove-member <organization-id> <member-id>
+myskills organizations append-policy|set-update-policy <organization-id> --input <request.json>
+myskills organizations activate-policy <organization-id> <revision-id>
+myskills organizations create-team <organization-id> <name> [--slug <slug>]
+myskills organizations create-team <organization-id> --input <request.json>
+myskills organizations adopt-team <organization-id> <team-id>
+```
+
+`submissions show` and `submissions export` inspect the author's submission;
+`review show` uses the review permission boundary. An export prints the JSON
+bundle unless `--output` selects a new file. Its parent directory must exist,
+and an existing destination is never overwritten.
+
+`search`, `skills managed` and `review submissions` return one page at a time.
+Use the response's `nextCursor` with `--cursor` and retain the same query and
+filters. `--json` preserves the response and pagination metadata. Inspect an
+exact release with `myskills info <slug> --version <version>`; the version must
+be a semantic version such as `1.2.3`, and the response is JSON release metadata.
+
+### Local package authoring
+
+`myskills init <name>` creates a minimal private Codex package in a new
+directory. The default output is `./<name>`; use `--output` to select another
+destination under an existing parent directory. The command writes `skill.json` and `SKILL.md` with
+version `0.1.0`, `UNLICENSED`, and the `codex-skill` platform target. It does
+not authenticate, call the network, or overwrite an existing file, directory,
+or symlink. Edit the generated `SKILL.md`, then run `myskills validate --path`
+and `myskills scan --path` before any separate authenticated submission.
+
+`myskills package --path ./my-skill --output ./my-skill.zip` creates a
+deterministic ZIP from one checked snapshot of a local package directory.
+It validates the root manifest, portable paths, and bounded UTF-8 text, then
+blocks credential and other blocking scan findings before creating output.
+Warnings remain visible for review. The output reports the archive SHA-256,
+byte size, and a separate `myskills submit --path ...` command. `--json` returns
+the manifest, scan findings, output path, checksum, size, and next command.
+Packaging does not contact the registry or bypass submission and review.
+
+The output must end in `.zip` and have an existing parent directory outside
+the source tree. Existing files, directories, symlinks, and symlink ancestors
+are refused. All regular source files are included; there is no ignore file.
+Keep credentials and unrelated files outside the source directory. The package
+limits are 500 files, 1 MiB of text, 1,000 filesystem entries, and a 10 MiB
+archive. Binary files,
+symlinks, and special files are rejected. Text bytes, including BOMs and line
+endings, are preserved. Entries use ordinal UTF-8 path order, fixed timestamps
+and permissions, and uncompressed ZIP storage, so source mtimes and locale do
+not change the archive checksum. Empty directories and executable bits are not
+preserved. The archive checksum identifies the ZIP bytes, not the registry's
+package-content digest.
+
+Local authoring requires macOS or Linux. The archive is created with private
+file permissions and exclusive creation; it never overwrites a destination
+that appears during the command. A disk or write failure can leave a partial
+output, which the error asks you to inspect before removal. Work in directories
+owned by your user; this does not isolate writes from a hostile process running
+as the same operating-system user.
+
+The API-backed architecture preview includes the compiled graph, escaped
+Mermaid, and a versioned diagram artifact with a plain-text accessible outline.
+Use `--json` when a caller needs the complete JSON/Mermaid/outline projection;
+the human-readable preview prints the bounded topology summary and Mermaid.
+The browser workbench additionally offers derived JSON and Mermaid downloads.
+
+### Local Codex observation
+
+`architectures observe` and `architectures health` run the Codex adapter locally
+against an explicitly supplied absolute root. The profile must be selected
+explicitly as `personal`, `work`, or `shared`. Supply either `--context` with a
+JSON object containing exactly `targetId`, `targetGeneration`,
+`architectureId`, `environmentId`, `profileId`, `adapterDigest`, and
+`capabilitiesDigest`, or pass those seven values through their corresponding
+flags.
+
+The command reads only bounded, allowlisted metadata and safe skill
+frontmatter. It does not search home directories, follow profile pointers,
+read prompt or skill bodies, emit local paths, retain credentials, call the
+network, upload observations, or modify a target. Review the JSON output before
+any separately authorized/manual upload to an API observation route. No live
+apply, rollback, installation, or other target mutation is available through
+this command.
+
+### Fixture-only sync and recovery
+
+`architectures plan` and `architectures dry-run` accept a bounded observed
+fixture and return a dry-run plan. The fixture is not an implicit target and
+the command does not apply changes. General architecture graph execution remains
+fixture-only, even though its API persistence and recovery/rollback evidence
+can be stored in Postgres. The separate Codex workspace and managed skill
+operation commands below support one explicit filesystem installation path.
+Each bounded architecture fixture run allows at most 500 steps and 2,004 append-only
+receipts: a 1,002-receipt max-step lifecycle, one full apply/verify retry, and
+two recovery/terminal receipts. Further retries require a new bounded run.
+
+### Architecture and target management
+
+Create architectures, append revisions, manage organization grants and derive
+another architecture pattern through the API:
+
+```text
+myskills architectures create --input <request.json>
+myskills architectures revisions|grants <architecture-id>
+myskills architectures revise|draft-preview|set-grants|migration-preview|migrate <architecture-id> --input <request.json>
+myskills targets list
+myskills targets show|updates|operations|update-policy|revoke <target-id>
+myskills targets observations <target-id> [--limit <positive-integer>]
+myskills targets register --input <request.json>
+myskills targets consent|observe|health|schedule|set-update-policy <target-id> --input <request.json>
+myskills operations show|cancel <operation-id>
+myskills operations batch --input <request.json>
+```
+
+Review each JSON request file before passing it to `--input`. These commands
+use the API body schema and retain its revision, artifact and idempotency
+fields. For example, a consent file contains `{"decision":"grant"}` or
+`{"decision":"deny"}`. A scheduled update contains `action`, `slug`, an exact
+`version`, optional `platform`, and a caller-chosen `idempotencyKey`. Batch
+requests wrap rows in `operations`, with each row also containing `targetId`.
+
+Use `draft-preview` before `revise` and `migration-preview` before `migrate`.
+Supply the inspected `expectedCurrentRevisionId` when editing or migrating;
+target upgrade policies use `expectedRevisionNumber`. Organization policy and
+grant writes likewise retain the API's required revision fields. On a conflict,
+read the current state and review the change again. Retrying the same migration
+or scheduled operation must retain the same idempotency key and request body.
+The CLI does not generate a replacement key or refresh revision fields for you.
+
+Target management uses a login session, and protected writes require MFA.
+`targets observe` uploads the supplied observation; `architectures observe`
+generates local metadata as described above. Scheduling returns an operation
+for the enrolled companion to execute. Read its state and receipt with
+`operations show`; a queued operation is not proof that files changed.
+
+## Published CLI And Local Builds
 
-The currently published package remains available under npm's `alpha` tag:
+The published beta channel is installed with:
 
 ```bash
-npm install -g @jarel/myskills@alpha
+npm install -g @jarel/myskills@beta
 myskills --version
 myskills login
 ```
 
-Update the CLI with:
+The npm `beta` tag selects the published prerelease. It can differ from this
+source candidate and the deployed API/web revision. To test source changes
+before npm publication, build and run them locally:
 
 ```bash
-npm install -g @jarel/myskills@alpha
+npm ci
+npm run build
+node apps/cli/dist/index.js --version
+node apps/cli/dist/index.js login
+npm pack -w @jarel/myskills
 ```
 
-The `0.1.0-beta.2` source manifest is configured for the `beta` dist-tag, but package publication is a separate maintainer approval step. The release-verification workflow packs and installs the candidate without publishing it.
+Use the resulting tarball for a fresh isolated installation, or run the bundle
+directly. The release gate verifies the packed executable and Apache license.
+Public npm publication is a separate release step.
 
 `validate`, `scan`, and `submit` accept a manifest file, package directory, or local `.zip` package. `login` prompts for the API URL when one is not supplied; the default is the local API at `http://localhost:3001`, and custom hosted URLs can be entered manually. Successful login stores the selected API URL in local CLI config so later commands can omit `--api-url`. API URL resolution is `--api-url`, then `MYSKILLS_API_URL`, then saved config, then `http://localhost:3001`.
 
-`login` supports an email/password session flow and an API-key flow. The email/password flow handles MFA challenges with a TOTP or recovery code prompt and stores only the verified session token. The API-key flow validates the key with `/v1/me` before storing it. Token resolution is `--token`, then `MYSKILLS_TOKEN`, then the stored login token. The default token store uses the platform credential store through `@napi-rs/keyring` and falls back to `tokens.json` with user-only file permissions when keyring storage is unavailable or `MYSKILLS_TOKEN_STORE=file`/`MYSKILLS_TOKEN_FILE` is set. `auth status` validates the current token without printing it. `logout` revokes stored session tokens and clears the local entry; stored API tokens are removed locally and must be revoked with `token revoke`.
+`login` supports email/password, API-key and browser/device flows. Password
+login handles MFA challenges and stores the verified session token. API-key
+login validates the key with `/v1/me`. Token resolution is `--token`, then
+`MYSKILLS_TOKEN`, then the stored token. The default store uses the platform
+credential store through `@napi-rs/keyring`. Credential-store failures are
+reported; a failed write does not silently store the credential in a file.
+An existing legacy file credential can be read when the keyring entry is
+confirmed absent. Explicit `MYSKILLS_TOKEN_STORE=file` or `MYSKILLS_TOKEN_FILE`
+selects file storage with user-only permissions. A successful keyring write
+clears the obsolete file entry.
+
+`login --method browser` displays the consent URL, user code and deadline. Repeat
+the command with the same API URL, configuration profile and `--scopes` choice to
+resume after an interruption, including after browser approval. The pending
+device secret stays in a separate OS keyring entry; it never enters config,
+token files or terminal output. A process lock serializes polling and logout
+for that API/profile. Its nonsecret owner metadata lives under the actual OS
+account home, independent of `HOME` and configuration-directory overrides.
+Only confirmed dead owners are reclaimed; ambiguous locks require operator
+recovery. Denial, expiry during polling, redemption and logout clear the pending
+entry. After process loss, expired keyring state is removed on the next login
+or logout for that API/profile; there is no background cleanup process. Logout
+discards a pending-only login locally; it does not cancel consent on the API.
+
+Resume keeps the original server expiry and a conservative deadline measured
+before the start request. Polling honors the saved interval, including
+`slow_down`, and bounds each request to the remaining deadline. It never renews
+consent, widens scopes, extends MFA assurance or changes token lifetime.
+Changing `--scopes` while a login is pending requires logout first.
+
+Keyring access is required for automatic resume even when final token storage
+was explicitly set to a file. A host without keyring access, including many
+cloud runners, can explicitly choose `--no-resume` for an in-memory request.
+That option cannot replace a readable existing pending login: logout first.
+If keyring access is unavailable, the CLI warns that old pending state could
+not be checked; restore access and run logout before using resume again.
+There is no plaintext fallback for pending state.
+
+Expiry, interruption, denial, token-save failure and config-save failure have
+distinct JSON error codes. If the server redeemed approval but the response was
+lost, or credential saving was not confirmed, the one-time code cannot recover
+the token. Check `auth status` with the same API/profile and, when needed,
+revoke the CLI browser login token in account settings before starting again.
+A config-save failure explicitly reports that the credential was saved; pass
+`--api-url` until the configuration store is repaired. Cleanup failures remain
+visible, and keyring-unavailable logout still attempts the selected credential's
+normal local deletion or remote session revocation.
+
+`auth status` validates the token without printing it. `logout` revokes stored
+sessions and clears local credentials. If a malformed keyring entry prevents
+revocation, logout attempts local deletion and reports that remote revocation
+is unconfirmed. It reports deletion failures. Stored API tokens are removed
+locally and must be revoked with `token revoke`. Malformed file-store JSON
+requires repair or removal of that file; other stored accounts are not silently
+discarded.
 
 `config get api-url`, `config set api-url <url>`, `config reset api-url`, and `config list` manage the saved API URL. `doctor` checks the CLI version, Node version, resolved API URL, `/health`, auth status, token-store backend, install-directory writability, and `/v1/capabilities`. If the CLI is pointed at the web app instead of the API, or a newer command is sent to an older server, command errors include concrete next steps and `--json` returns structured error codes.
 
-`submit` validates and scans locally before sending package directories as normalized text entries or `.zip` packages as base64 archive uploads for server-side extraction. Authors can inspect their submitted versions with `submissions list` and withdraw unreviewed or changes-requested submissions with `submissions withdraw`. Maintainers can fetch the reviewed artifact and approval hash with `review bundle`, then approve, request changes, reject, and publish submitted versions through `review action`.
+### Older workspace scope identities
+
+Scope state now records both device and inode. Existing inode-only state stays
+readable, but observation and ordinary re-enrollment stop with
+`SCOPE_ROOT_IDENTITY_LEGACY`. After confirming the same directory, repeat its
+original `scopes enroll` command with `--accept-current-root` and the original
+configuration profile. The command verifies the account, registry, architecture,
+and existing target, then backs up local state and updates only the root identity.
+It does not register targets, grant consent, or upload inventory. Pending
+enrollments retain their recovery identity and can resume with normal enrollment
+after acknowledgment. A repeated acknowledgment is a no-op; it cannot override
+a known device/inode mismatch. Device and inode do not detect every same-device
+inode reuse. See [scope identity migration](../../docs/WORKSPACE_SCOPES.md#upgrade-an-inode-only-scope-identity)
+for the exact upgrade and recovery behavior.
+
+### Separate work and personal configuration
+
+Every command accepts `--config-profile <name>` before or after the command.
+It overrides `MYSKILLS_CONFIG_PROFILE`. Names contain 1–64 lowercase letters,
+digits, hyphens or underscores and start with a letter or digit. For example:
+
+```bash
+myskills config set api-url https://work-registry.example/api --config-profile work
+myskills login --config-profile work
+myskills whoami --config-profile work
+myskills scopes list --config-profile work
+myskills config list --config-profile work --json
+```
+
+There is no create or persistent select command. Writes create the selected
+profile's state as needed; keep the flag on each command or set
+`MYSKILLS_CONFIG_PROFILE` in that shell. `help`, `config list`, and `doctor`
+show the selected configuration. JSON uses `configProfile: null` for the
+legacy default. This selector is separate from architecture `--profile-id`
+and provider inspection `--profile`.
+
+Without a selector, existing config, credentials, and scope bindings remain at
+their current locations. A named `personal` profile starts empty; it does not
+copy or migrate the default. Each named profile has independent saved API URL,
+credentials, and scope state, including for different accounts at one API URL.
+A missing credential never falls back to another profile or the default.
+
+The base directory is `MYSKILLS_CONFIG_DIR`, then
+`$XDG_CONFIG_HOME/myskills-app`, then `~/.config/myskills-app`. Named state lives
+under `<base>/profiles/<name>/` (`config.json`, optional `tokens.json`, and
+`scopes/`). OS credential keys include the canonical profile directory and API
+URL. Aliases for the base directory are supported; linked profile directories
+are rejected so they cannot share another profile’s files. Changing the base
+directory selects separate credentials. Named profiles
+reject `MYSKILLS_CONFIG_FILE` and `MYSKILLS_TOKEN_FILE`; these retain their
+existing precedence when no profile is selected. `MYSKILLS_TOKEN_STORE=file`
+remains available for profile-local file storage.
+
+Managed workspace overlap checks use one protected index under the actual OS
+account home at `.config/myskills-app/workspace-enrollments`. `HOME`,
+`XDG_CONFIG_HOME`, registry URLs and configuration profiles cannot split this
+authority. The selected older XDG/home index is read for reservation migration;
+its files are preserved. This coordinates normal CLI processes. Preserve the
+index and workspace bindings when a conflicting root requires explicit recovery.
+
+Explicit API and token overrides still apply: `--api-url` then
+`MYSKILLS_API_URL` override the saved registry, and `--token` then
+`MYSKILLS_TOKEN` override the stored credential. Clear inherited overrides when
+you want the profile's saved registry/account. `auth status` reports the token
+source without printing its value. Selection does not move skill directories,
+change native provider profiles, or unbind existing targets.
+
+`bootstrap codex --dry-run` is a work/team-only local planner. It requires an
+explicit `work` context, a normalized HTTPS target origin, stable target
+instance/workspace/actor identifiers, explicit source and target trust
+compartments (`work`, `shared`, or `work+shared` for sources, and `work-team`
+for the target), typed `work` and/or `shared` source roots, a target root, and
+a positive `--include-slug` allowlist. The context must approve the computed
+source-root type/digests and target-root digest. Personal, consumer, public,
+and unclassified sources are rejected, and candidates are never selected by
+implicit inventory discovery. A plan accepts at most eight selected candidates;
+each candidate snapshot accepts at most 1,000 entries and 32 nested directory
+levels. A one-skill work-owned canary is supported.
+
+The context JSON must include `profile: "work"`, `targetOrigin`,
+`instanceId`, one `tenantId` or `workspaceId`, `actorId`, the canonical trust
+compartments, `approvedSourceRoots` entries of `{type, identityDigest}`, and
+`approvedTargetRootIdentityDigest`.
+
+The planner reads selected source and target content without network or
+registry access. It opens each selected tree through a bootstrap-owned secure
+snapshot reader, then hashes, parses, scans, and reports only those held bytes.
+It writes only the explicitly requested new private report (`0600`) under a
+user-controlled private parent. Reserved package manifests remain strict:
+invalid or multiple manifests fail closed, and no metadata is relocated. The
+full report contains the work binding, approved root digests, owning package
+contract identifiers, candidate content identities, execution identities, the
+exact target precondition (`absent` or `present-identical` with its expected
+artifact identity), and private snapshots. Terminal output is a redacted DTO
+with counts and digests; it does not contain paths, candidate names, inventory,
+or private identifiers.
+There is no apply, publishing, adoption, or target-write mode in this command.
+`ready` means that the held snapshot passed deterministic checks only. It is
+not approval to apply. Sensitive configuration and credential paths are
+excluded before content enters the report, and heuristic scan-clean never
+replaces explicit human content review. Any future executor must repeat the
+retained source/target identity check immediately before its first write
+(compare-and-swap) and require that review at that boundary.
+
+`validate`, `scan`, and `submit` read one bounded snapshot of the local package.
+`submit` sends those same validated and scanned text entries for directory and
+ZIP inputs. Authors can inspect and withdraw submissions; maintainers fetch the
+reviewed artifact and approval hash before approving, requesting changes,
+rejecting, or publishing it. The browser shows review reasons and scan findings.
+Corrections create another immutable version.
+
+Published artifacts remain immutable. `skills edit` changes metadata;
+`releases deprecate`, `unpublish`, `revoke`, `restore`, and `delete` change
+server-owned lifecycle state. Deprecated releases remain installable. Hidden,
+revoked, archived, or deleted releases cannot be installed or exported.
+`export` verifies artifact size and SHA-256 before writing normalized paths.
+
+`install`, `update`, and `rollback` share an exclusive install-root lock and
+verify package bytes and current eligibility before promotion. Updates retain
+a verified rollback snapshot. Interrupted transactions preserve unknown or
+edited local bytes for investigation. Each installation binds the API origin
+and registry instance ID. Legacy records without that identity are not adopted
+automatically: keep their files as a backup, review the source, and install into
+a new root. A registry change requires a separate root.
+
+New installation and snapshot records declare
+`contentDigestAlgorithm: "sha256-json-ordinal-v1"`; transaction journals use
+`targetContentDigestAlgorithm` for the candidate. This scheme normalizes paths,
+sorts them by JavaScript string ordinal order (UTF-16 code units), then hashes
+the UTF-8 JSON array of `{path, content}` entries with SHA-256. It does not
+depend on the process locale. Readers reject unknown algorithm identifiers.
 
-Published artifacts remain immutable. `skills edit` changes mutable skill metadata only, while `releases deprecate`, `releases unpublish`, `releases revoke`, `releases restore`, and `releases delete` update server-owned lifecycle state for a specific version. Deprecated releases remain visible and installable; unpublished, revoked, archived, and deleted releases are hidden from install/export queries. `export` downloads server-authorized bundle content, verifies byte size and SHA-256 against release metadata, and writes normalized package paths under the requested output directory. `install` uses the same verified bundle path, writes into `--dir`, `MYSKILLS_INSTALL_DIR`, or the user data directory, and records local state in `.myskills-app/installed.json`; `update` preserves a rollback snapshot before replacing files, and `rollback` restores the most recent snapshot. `token create` prints the plaintext API token only once and does not overwrite the stored login session. Browser/device login, platform-specific install adapters, and archive creation are still planned.
+A missing algorithm field identifies the legacy `localeCompare` scheme.
+The CLI verifies these records with the current process's legacy comparison
+rules. It does not retry with a different algorithm. If a locale change makes
+legacy bytes unverifiable, return to the original verification locale and
+compatible Node/ICU environment before a trusted migration. Preserve the files
+and recovery copies if their identity still cannot be proved. Never add an
+algorithm field to an old hash by hand. After verification, an install or
+update records the held bytes with an ordinal digest in the new journal and
+snapshot; rollback does the same for its verified source snapshot. Untouched
+legacy history and journals retain their original verification rules.
+
+Local package intake, install, export, and rollback support macOS and Linux.
+No-follow payload reads and writes, private staging directories, and drift
+checks protect these operations. They do not isolate the workspace from a
+hostile process already running as the same OS user. Other operating systems
+can use the API's portable buffer upload path; native Windows filesystem
+installation is unsupported.
+
+### Personal Codex workspace
+
+Create a personal architecture in the browser and note its architecture,
+environment, and profile IDs. Enable MFA and sign in with a password session.
+Enroll an existing absolute workspace directory using the command above.
+Enrollment creates a user-owned target, grants consent, and stores a local
+binding. A fresh workspace cannot attach to an existing target ID; that option
+only confirms or resumes its existing local binding.
+
+Use `--workspace` for all managed writes. Skills are installed under
+`.agents/skills`, and records stay under `.myskills-app` in that workspace.
+The CLI checks Codex compatibility and valid `SKILL.md` YAML frontmatter with a
+matching name and text description. `--dir` cannot bypass an enrolled workspace's
+binding. Team-shared skills can be installed when your account can read them;
+team-owned execution targets are outside this adapter's beta scope.
+
+### Global and project skill inventories (workspace scopes)
+
+`myskills scopes` enrolls a provider's user-level skills directory, or a
+project, as a personal read-only inventory target: `codex-inventory` for Codex
+and `claude-inventory` for Claude. Pass each directory explicitly; the CLI does
+not search your home directory. Run `scopes inventory` first to review locally
+what an upload would contain. Uploads carry skill slugs, counts, and
+`SKILL.md` digests only. Absolute paths, exclusions, skill bodies, and names of
+linked or invalid entries stay local.
+
+Projects resolve to the provider's global scope unless you exclude them or
+enroll them separately. Resolution uses real paths and whole path segments, the
+deepest rule wins, and an excluded project never falls back to the global
+scope. Scopes are MySkills ownership only: they do not change how Codex or
+Claude load or inherit skills, and they never write skill files. Existing
+managed Codex workspace bindings can be adopted with a previewed
+`scopes migrate plan` and `scopes migrate apply`. See
+[Workspace Scopes](../../docs/WORKSPACE_SCOPES.md) for the rules, recovery
+steps, and local state format.
+
+`codex observe --upload` records verified filesystem state. Confirm separately
+that Codex loaded the skill. To process one browser-queued update, supply a
+separate token with `skills:read` and `targets:execute` through `MYSKILLS_TOKEN`
+and run `companion run-once`. If the workspace has a skill installed with
+`--library-entry`, the token also needs `libraries:read`: the companion rechecks
+that adoption before it changes files. Without it, the operation fails with
+`API_TOKEN_SCOPE_REQUIRED` and a message that names the scope; files are not
+changed. The CLI never widens a token's scopes. This command checks current
+authorization, consent, policy, lease, and exact release identity. It does not
+start a background daemon.
+Additional provider install adapters remain planned.
+
+To change skill visibility, use the canonical `myskills sharing set
+<skill-slug> --visibility <scope>` command. It accepts either
+`--organization <organization-id>` or `--organization-id <organization-id>` for
+the complete organization grant set. Omitting both organization options keeps
+the beta.2 compatibility behavior and preserves already-issued organization
+grants. Pass `--clear-organizations` to send `organizationIds: []` and revoke
+the complete organization grant set; this flag is mutually exclusive with both
+organization ID options.
+
+`myskills skills edit --visibility <scope>` remains a deprecated beta.2
+compatibility alias. It preserves omitted organization grants and does not
+provide complete-set organization controls; use the canonical sharing command
+to grant or clear organization access. Canonical sharing remains subject to
+the API's session and MFA security rules; the beta.2 metadata alias is also
+session-only and requires an MFA-verified session before it reads or replaces
+grants. API tokens cannot widen a skill through the alias. Neither path
+bypasses server policy. Organization policy and membership remain API-owned.
+Do not treat a successful CLI command for another scope as evidence of
+organization sharing.
+
+Use `architectures grants` and `architectures set-grants --input <request.json>`
+for the separate architecture organization-grant replacement workflow. Supply
+the architecture ID and retain the current revision in the request body.
+The read-only `architectures preview`, `compile`, `plan`, and `dry-run`
+commands already accept `--organization-id <organization-id>` (with
+`--organization <organization-id>` retained as an input alias). The server
+authorizes that exact organization projection; it is a scope filter, not an
+ownership shortcut.
+
+The beta.2 compatibility shims remain available and are planned for removal only
+at a later, separately published prerelease boundary that includes migration
+guidance and release verification. The source release does not imply a hosted
+deployment.
 
 Common scopes:
 
@@ -111,3 +623,196 @@ Common scopes:
 - `profile:read` for `whoami`.
 - `skills:submit` for author submissions.
 - `review:read` and `review:write` for maintainer review workflows.
+- `architectures:read` for architecture list, detail, preview, and fixture-plan reads.
+- `libraries:read` for library reads and for install, update, and companion runs of library-bound skills.
+- `libraries:write` for library changes.
+
+## Account and instance administration
+
+Account commands use the existing registration, session and MFA rules. Passwords,
+emailed verification/reset tokens, invitation tokens and authenticator codes are
+read from hidden prompts. New passwords require confirmation. These secrets are
+not accepted as command options.
+
+```text
+myskills account register --email <email> [--name <name>]
+myskills account verify-email-request|password-reset-request --email <email>
+myskills account verify-email|confirm-email-change
+myskills account password-reset|change-password
+myskills account change-email --email <new-email>
+myskills account mfa-status
+myskills account mfa-enroll --output <new-file> [--label <label>]
+myskills account mfa-confirm <factor-id> --output <new-file>
+myskills account mfa-disable
+myskills connections info|list
+myskills connections revoke <connection-id>
+myskills instance info|health|ready|version|capabilities|branding|site
+```
+
+`mfa-enroll` writes the authenticator setup secret to the new private JSON file
+selected by `--output`; it prints the factor ID and output path without the
+secret. Configure the authenticator, then use that factor ID with `mfa-confirm`.
+Confirmation prompts for the current code and writes recovery codes to another
+new private JSON file. Choose files in an existing directory outside the
+repository; existing destinations are refused. Save the recovery codes securely,
+then run `myskills login` and complete MFA to obtain a verified session.
+
+Administration commands retain the API's owner/admin role, MFA and last-owner
+safeguards. `admin users roles` replaces the user's role set with the repeated
+`--role` values. `admin audit` returns one page and supports its `nextCursor`.
+
+```text
+myskills admin branding get
+myskills admin branding set --input <branding.json>
+myskills admin site get
+myskills admin site set --landing-page enabled|disabled
+myskills admin registration get
+myskills admin registration set --mode closed|request|open
+myskills admin registration invite --email <email> [--name <name>]
+myskills admin providers list
+myskills admin providers set <provider-key> --input <provider.json>
+myskills admin users list
+myskills admin users action <user-id> --action approve|activate|disable|delete [--reason <text>]
+myskills admin users roles <user-id> --role <owner|admin|maintainer|author|user> [--role <role>] [--reason <text>]
+myskills admin tokens list
+myskills admin tokens revoke <token-id>
+myskills admin audit [--limit <1-100>] [--cursor <cursor>]
+```
+
+Review branding and provider JSON bodies before passing `--input`. Provider
+configuration accepts non-secret settings; configure provider credentials in the
+deployment secret store. `connections` manages the signed-in user's remote MCP
+connections. `instance` reads public instance information without credentials.
+
+## Libraries
+
+`myskills libraries help` lists the source, import, review, tracking, subscription
+and adoption commands. Requests use the [Libraries API contract](../../docs/plans/2026-09-26-library-api-contract.md)
+and reviewed JSON files supplied with `--input`. The [Libraries guide](../../docs/LIBRARIES.md)
+contains a complete workflow and scope requirements.
+
+Filter library entries with
+`myskills libraries entries <library-id> --kind source` (or `skill`). Filter
+candidates with `myskills libraries candidates <entry-id> --state ready-for-review`;
+other states are `blocked`, `accepted`, `ignored`, `superseded` and `expired`.
+List reads accept `--limit <1-100>` and `--cursor <cursor>`; retain the filters
+when continuing a page. Library deletion requires the current revision:
+`myskills libraries remove <library-id> --revision <current-revision>`.
+
+Collections and Groups organize existing skill entries in one Library. List with
+`libraries collections <library-id>` or `libraries groups <library-id>`; inspect
+with `libraries collection <collection-id>` or `libraries group <group-id>`.
+`collection-members` and `group-members` list ordered entries with the same
+`--limit` and `--cursor` pagination. Collection summaries include source tracking
+health and authorized pending candidate counts.
+
+Create with `libraries create-collection <library-id> --input reviewed.json`
+or `libraries create-group <library-id> --input reviewed.json`. The JSON contains
+`name`, optional `description`, `memberEntryIds` (up to 200 existing skill entry
+IDs), and an optional `clientMutationId`. Preserve the mutation ID and payload
+for retries. Edit with `edit-collection <collection-id>` or `edit-group <group-id>`
+and `--input reviewed.json`; include `expectedRevision` from the latest read.
+Supplying `memberEntryIds` replaces the complete ordered membership. An empty
+array clears it; omitting the field preserves it. Use `remove-collection` or
+`remove-group` with `--revision <current-revision>` to delete the organization
+only. Library entries, source tracking, adoptions and installations remain intact.
+Reads require `libraries:read`; changes require `libraries:write`, current
+Library curation permission and MFA for team Libraries. Groups can overlap and
+are independent of Collections.
+
+Bind an installation to an adopted version with
+`myskills install <slug> --library-entry <entry-id>`. If the adopted release
+requires user action, read its notes and add `--accept-user-action`. Updates
+resolve that exact recommendation and recheck it before changing files. A
+deleted or inaccessible entry never falls back to the newest registry release.
+Local drift still stops replacement. Use
+`myskills libraries unbind-local <slug> --dir <root>` to detach that local
+binding explicitly while retaining files and rollback history.
+
+`updates` and `update` report each skill separately. If an entry is deleted,
+access is lost, the adopted release is revoked, or the entry has no adoption,
+that skill shows `curation-unavailable` with the reason (`library.state` and
+`library.reason` in `--json`). Its files and binding stay unchanged, other skills
+are still evaluated and updated, and `update` exits 1. `install` fails closed
+with `LIBRARY_CURATION_UNAVAILABLE`. A malformed or mismatched resolution stops
+the command with `LIBRARY_RESOLUTION_INVALID`.
+
+If a library adopts a version older than the installed one, the skill shows
+`library-adopts-older` (`adoption-older-than-installed` in `--json`). MySkills
+never downgrades automatically. When the last rollback snapshot is exactly the
+adopted release, run `myskills rollback <slug>`; otherwise install the adopted
+version into a new root with `--dir <new-root>`. `libraries unbind-local` keeps
+the installed version instead.
+
+`libraries review-bundle <submission-id>` prints only after the SHA-256 of the
+response body matches `x-myskills-artifact-sha256`. The JSON output labels its
+`payload` as `parsed-for-inspection`. Use `--output <new-file>` to keep the
+exact verified bytes; an existing file is never replaced. Supply
+`--artifact-sha256 <digest-from-review-requests>` to also verify the immutable
+artifact requested for review. Without it, verification covers the response
+header only; `expectedDigestVerified` reports this distinction in JSON output.
+
+## Skill improvement
+
+Prepare local skill reviews with `myskills improve`, inspect the report, export a draft, and submit it through normal review. Registry plans pin source/reviewer versions and apply user, team or organization policy. Execution requires the exact local plan digest and explicit cloud consent. See [the feature guide](../../docs/SKILL_IMPROVEMENT.md) for commands, JSON bodies, evaluation suites and current adapter limits. The first runner is Claude Code 2.1.283 or newer. Use an exact model ID and an existing Claude sign-in. Codex improvement execution remains disabled pending its isolation gate; final release verification is recorded in the feature guide.
+
+Inspect a saved registry plan and its runs with
+`myskills improve show-plan --id <plan-id> --json`. This reads the plan without
+starting local execution. `myskills improve compatibility --release <slug>@<version>`
+supports anonymous reads of public releases and uses available credentials for
+authorized private releases.
+
+Share evidence from either a local job or an authorized completed registry run:
+
+```text
+myskills improve share --job <dir> --disclosure summary --subject baseline|candidate --release <slug>@<version>
+myskills improve share --run <run-id> --disclosure summary|selected-evidence --subject baseline|candidate --release <slug>@<version>
+myskills improve share --run <run-id> --disclosure summary --proposals-file <proposals.json>
+```
+
+`--job` retains the original registry endpoint, plan and confirmed report digest
+checks and supports summary disclosure only. `--run` reads the completed run and
+uses its recorded report digest; API tokens need `improvements:read` and
+`improvements:report`. Both forms accept a reviewed proposals file in place of
+`--subject` and `--release`:
+
+```json
+{"proposals":[{"subject":"baseline","slug":"example-skill","version":"1.0.0"}]}
+```
+
+The file accepts zero to four proposals, each with only `subject`, `slug` and an
+exact `version`. Choose exactly one of `--job` or `--run`, and one proposal input
+form. Repeating the same run, report digest, disclosure and proposals reuses the
+same evidence request identity. Evidence sharing does not run an evaluation or
+publish a skill release.
+
+### Skill bundles
+
+Browse related skills without installing them:
+
+```bash
+myskills bundles list --view grouped --query engineering
+myskills bundles show <bundle-id>
+myskills bundles members <bundle-id> --limit 25
+myskills bundles memberships <skill-slug>
+myskills bundles sources
+```
+
+Create or edit with `myskills bundles create --input reviewed-bundle.json` or
+`myskills bundles edit <bundle-id> --input reviewed-bundle.json`. The request
+contains `kind`, `name`, `purpose`, `owner`, `visibility` and `memberSlugs`.
+Source groups also require `sourceEntryId`; edits require `expectedRevision`.
+The API validates current ownership, author privileges and member access.
+
+Save with `myskills bundles save <bundle-id> --input save-reference.json`, where
+the file contains `libraryId` and `expectedRevision`. This saves one reference;
+it does not adopt, install or follow any skill. Use `--cursor` with list/member
+reads. Refresh from the first page if the authorized catalog changes.
+
+### GitHub connection controls
+
+Use `myskills account github status` to inspect the connected account and `myskills account github disconnect` to revoke it. Administrators can use `myskills admin github get` and `myskills admin github test` for safe configuration status and verification. These commands preserve API scope, account, role and MFA checks.
+
+`myskills account github connect` and `myskills admin github configure` return instructions for the trusted browser flow. They do not open a browser, enter credentials, grant consent or claim completion. Keep GitHub App secrets and private keys in the trusted Admin form. Inspect the safe status command after completing that flow.
+
+Use `myskills skills managed <slug>` to inspect an authorized archived or unpublished skill from the managed inventory. The public `info` command retains the public release contract.