agent-bios 0.19.1 → 0.19.3

package/docs/advanced-launch.md

@@ -1,6 +1,6 @@
 # Launch configuration
 
-[← Overview](../README.md) · [Setup](setup.md) · [Corpus](corpus.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
+[← Overview](../README.md) · [Setup](setup.md) · [Instructions](instructions.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
 
 Inspect settings before launching. Some Builder and internal-wrapper defaults use permission bypass. A configured review or registered hook is not evidence that it ran.
 
@@ -8,13 +8,13 @@ Inspect settings before launching. Some Builder and internal-wrapper defaults us
 
 The preflight keeps the current setup above each choice, supports the configured model
 catalog and **Other**, and offers Builder presets, Software Engineer / Vanilla,
-Session distill, Custom, Language, and **Corpus Studio**. Studio is the same backend as
-`agent-bios corpus`: it searches and renders the library, edits Markdown and
+Session distill, Custom, Language, and **Instructions Studio**. Studio is the same backend as
+`agent-bios instructions`: it searches and renders the library, edits Markdown and
 consumption surface, and requires Preview then revision-bound Apply. Packaged
 entrypoints validate and temporarily extract their included UI bundle before loading
 Rich/Textual. A missing or corrupt bundle fails explicitly; no preinstalled Textual
 environment is required. Interface catalogs change only human UI text;
-model-consumed corpus remains English.
+model-consumed instructions remain English.
 
 Every arrow-key TUI selection screen keeps the complete current setup in a fixed top
 panel, followed by the highlighted option's description and the option list. Custom
@@ -59,14 +59,14 @@ contain instructions independently of the excluded documents.
 
 ## Native hooks and agents
 
-Native corpus consumption is default-off. `agent-bios corpus snapshot --host codex
---native --json` (or `--host claude`) composes a preview; `agent-launch --corpus-native`
-opts one configured session into selected corpus hooks. Both hosts use the same
+Native instructions consumption is default-off. `agent-bios instructions snapshot --host codex
+--native --json` (or `--host claude`) composes a preview; `agent-launch --instructions-native`
+opts one configured session into selected instructions hooks. Both hosts use the same
 installed Python carrier and typed `event`/`matcher` binding. Authoring accepts the
 combined event vocabulary; compilation reports an event unsupported by the selected
 host without changing its name or executing it through another event.
 
-Claude receives a namespaced plugin per CorpusRef through `--plugin-dir`. Codex
+Claude receives a namespaced plugin per InstructionsRef through `--plugin-dir`. Codex
 receives inline `hooks.<Event>` config through per-session `-c` arguments. Existing
 user, project and session hooks remain present, and resume retains the pin's exact
 registrations. Neither adapter installs global hooks. Codex hook enablement and native
@@ -75,7 +75,7 @@ checked before launch, but discovery alone does not establish execution. Hooks u
 host's command permissions; opting in permits the selected carrier to run. Editing an
 event binding does not rewrite the Python carrier's input/output contract.
 
-Native corpus agents currently use Claude plugins and retain authored frontmatter and
+Native instructions agents currently use Claude plugins and retain authored frontmatter and
 plugin-qualified names, distinct from launcher's bare tier agents. A Codex agent
 projection still needs to translate agent-specific model and tool restrictions; this
 does not limit shared hook delivery. Arbitrary prose promoted to `event` or `delegated`
@@ -95,7 +95,7 @@ loaded managed wrappers on the next shell command. An edited managed file or blo
 is preserved and reported for reconciliation, not overwritten. Backups remain private
 under `runtime/shell-backups/`; `ZDOTDIR` must match the connection's recorded path.
 Updates preserve an opted-in connection; reset and uninstall remove it. This setting
-never edits global `AGENTS.md`/`CLAUDE.md`, project files, or corpus content. Ordinary
+never edits global `AGENTS.md`/`CLAUDE.md`, project files, or instructions content. Ordinary
 private installation also leaves those globals alone; explicit `migrate` can remove
 the old agent-bios-managed regions and imports while preserving user-authored text.
 First opt-in records ownership before publishing shell wiring, so interrupted restores
@@ -110,7 +110,7 @@ Both the legacy shell adapter and the optional private shell connection preserve
 argument-bearing and non-TTY calls as direct backend invocations. The private
 connection adds no permission flags on that path. Without opting in, the private
 default installs no shell functions; an explicit `agent-launch` call projects a
-launch profile or corpus snapshot.
+launch profile or instructions snapshot.
 
 Direct `agent-launch` calls still require a valid profile to resolve the backend command and its default arguments. `--preset`, `--custom`, or `--dry-run` select the configured-launch path even when non-TTY or combined with `--no-tui`; a non-TTY bare `--dry-run` deterministically uses Balanced, and a custom profile without that preset must pass `--preset NAME`. Forwarded backend arguments are appended verbatim after the projected defaults; one that would override a projected option (the seat, the contract, delegation, policy) is refused at launch so the contract keeps describing the run, and the summary discloses forwarded arguments when present. For scripted configured launches, call `$HOME/.local/bin/agent-launch --preset NAME --yes HOST -- ...` or add `$HOME/.local/bin` to `PATH`. The summary goes to stderr so backend stdout stays machine-consumable.
 

package/docs/instructions-compatibility.md

@@ -0,0 +1,86 @@
+# Instructions compatibility
+
+[← Instructions](instructions.md) · [Storage and sessions](session-model.md) · [Recovery](recovery.md)
+
+**Instructions** is the current name for the rules, guides, and procedures
+component. Compatibility names below identify older interfaces and stored data;
+they are not a second component. New documentation and callers use Instructions.
+
+## Commands and environment
+
+| Canonical interface | Accepted older alias | Owner |
+| --- | --- | --- |
+| `agent-bios instructions` | `agent-bios corpus` | `install.sh` |
+| `--instructions` | `--corpus` | installer selection and launcher Studio entry |
+| `--instructions-domains` | `--corpus-domains` | `launch/agent-launch.py` |
+| `--instructions-native` | `--corpus-native` | `launch/agent-launch.py` |
+| `--no-instructions` | `--no-corpus` | app session preview/use |
+| `AGENT_BIOS_INSTRUCTIONS_DIR` | `AGENT_BIOS_CORPUS_DIR` | private store and installer |
+| `AGENT_BIOS_INSTRUCTIONS_STATUS` | `AGENT_BIOS_CORPUS_STATUS` | status projection and launcher |
+| `AGENT_BIOS_PRIVATE_INSTRUCTIONS` | `AGENT_BIOS_PRIVATE_CORPUS` | private runtime selection |
+
+An older environment variable is a fallback when its canonical counterpart is
+absent. If both are present with different raw values, the affected operation
+refuses and names both variables. It does not choose a value by precedence,
+normalize differing values into a guessed match, or select a different store.
+Set one name, or give both the same value.
+
+Registered helpers and returned runtime environments bind the already selected
+absolute roots under both spellings. This replaces ambient aliases consistently;
+it does not reinterpret conflicting settings supplied when selecting the store.
+
+The canonical module owners are `compose/instructions*.py`. The corresponding
+16 `compose/corpus*.py` modules remain thin import/CLI shims: `corpus.py`,
+`corpus-state.py`, and the `corpus_app`, `corpus_catalog`, `corpus_import`,
+`corpus_install`, `corpus_session`, `corpus_setup`, `corpus_setup_cli`,
+`corpus_setup_i18n`, `corpus_setup_ui`, `corpus_store`, `corpus_transaction`,
+`corpus_ui`, `corpus_ui_runtime`, and `corpus_understand` Python modules. They
+forward to the canonical owners rather than maintaining separate implementations.
+
+## Storage and identity
+
+| Retained literal | Reason and owner |
+| --- | --- |
+| `~/.config/agent-bios/corpus/` | The private store and installer keep the existing default user root. `AGENT_BIOS_INSTRUCTIONS_DIR` can select a custom root. |
+| `~/.local/share/agent-bios/corpus-status.json` | The status projection and launcher share the existing status location. |
+| `.corpus-store.lock` | `compose/instructions_transaction.py` retains the shared lock name so older and newer writers do not acquire independent locks. |
+| `corpus-rollback-*` | The status/recovery machinery retains its recovery-artifact naming contract. |
+| `deployed_corpus`, `corpus_storage`, `retained_corpus` | Existing schema-v1 serialized fields retain their spelling for consumers of deployment, setup, and retained-library records. This includes `display.retained_corpus`. |
+| `corpus_hash`, `corpus_drift` | Benchmark records retain their existing serialized fields. |
+| `corpus-not-loaded` | The compatibility learning migration keeps its recorded skip-result enum. |
+
+The naming change performs no automatic root relocation and creates no second
+store or independent writer. Retaining the lock preserves a common exclusion
+boundary; it does not establish that every older release understands newer data.
+
+Existing immutable snapshots, exact item references, and session pins retain
+their contents and identities. Newly compiled snapshots may have new hashes
+because compiler and store implementation bytes participate in their identity.
+A historical reference must resolve to its original content or report that it
+is unavailable; it must not silently resolve to a newer snapshot.
+
+## Historical and unrelated uses
+
+Dated design records, decisions, source quotations, and captured release images
+keep the names used at the time. The image `assets/instructions-studio.svg`
+is an actual 0.18.0 capture and retains its earlier Corpus Studio title.
+
+The review-request guides use *corpus* for their research dataset of review
+findings. Contributor documentation also uses it for the AGENTS.md/CLAUDE.md
+research collection. Those datasets are distinct from the Instructions component.
+
+The reserved operation names are `publish-instructions` and `fetch-instructions`.
+Neither has a network implementation, so there is no deployed endpoint to
+migrate. The rename introduces no new network traffic.
+
+## Removal condition
+
+The compatibility layer is retained in 0.19.2 and scheduled for future removal;
+no removal release is assigned. New integrations use the canonical interfaces.
+Removal work covers older command and option aliases, environment aliases, Python
+import shims, and old-release adapters. Stored paths, locks, serialized fields and
+immutable references require a separate preservation or migration plan before any
+related support is removed. Compatibility interfaces can be removed only in
+a separately announced breaking release after consumer migration and preservation
+or explicit migration of the affected data and references have been demonstrated.
+An occurrence count reaching zero is not evidence that those conditions hold.

package/docs/{corpus.md → instructions.md}

@@ -1,9 +1,16 @@
-# Your corpus
+# Your instructions
 
-[← Overview](../README.md) · [Setup](setup.md) · [Corpus](corpus.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
+[← Overview](../README.md) · [Setup](setup.md) · [Instructions](instructions.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
 
 Inspect, edit, and select the instruction library without rewriting native global instructions. Commands use the installed CLI; from a checkout, use `bash install.sh <command>` at the repository root.
 
+**Instructions** names the rules, guides, and procedures component. Manage it with
+`agent-bios instructions` or **Instructions Studio**. See
+[compatibility](#compatibility) for older command names and retained storage paths.
+
+If the installed CLI predates this source revision's `instructions` command,
+use `agent-bios corpus` with the same subcommands until that installation is updated.
+
 ## Consumption surfaces
 
 | Surface | Delivery |
@@ -16,12 +23,12 @@ Inspect, edit, and select the instruction library without rewriting native globa
 
 ## Manage the library
 
-Corpus Studio and the machine CLI are views over the same `CorpusStore`. The CLI
+Instructions Studio and the machine CLI are views over the same `InstructionsStore`. The CLI
 provides `list`, `search`, `show`, `history`, `status`, and `snapshot`; mutations are
 semantic JSON passed to `plan`, followed by `apply PLAN --expected-revision REV`.
 Implemented operations are create, update (including consumption surface), enablement, remove,
 installed-item restore, personal-item recover, selection, reset, and rollback. Stable
-`CorpusRef` identities survive those changes. The current `ContentRef` hashes the
+`InstructionsRef` identities survive those changes. The current `ContentRef` hashes the
 baseline, resolved selection, authoring/item/learning/promotion digests, catalog and
 store implementation digests, and management bootstrap. There is not yet a standalone undo command,
 arbitrary package authoring/import or automatic semantic conflict resolution.
@@ -61,7 +68,7 @@ content edits. A stale authoring revision refuses the preview or apply.
 
 Per-item choices override domain selection in default mode, including core and
 infrastructure defaults. Explicit selected mode includes only its chosen targets;
-an enable override cannot pull in unrelated items. No-corpus mode emits no corpus,
+an enable override cannot pull in unrelated items. No-instructions mode emits no instructions,
 and imported items retain their applicable host/project scope. Turning an item off does not delete its body, edits,
 or identity, and old snapshots and session pins remain intact. Choices survive
 updates and content restoration; full reset returns to installed defaults.
@@ -72,7 +79,7 @@ item is a projection choice, not proof of execution. This does not block host
 global/project instructions or a tool from opening a file independently.
 
 The same revision-checked manager accepts `{"operation":"enable","items":{"@agent-bios/core:rule-003":false}}`
-through `corpus plan`; `true` enables within the active selection policy, `false`
+through `instructions plan`; `true` enables within the active selection policy, `false`
 excludes, and `null` removes that override so normal selection applies. `list` reports `enabled`,
 `enabled_override`, and the captured authoring `revision`; pass that revision as
 `expected_revision` when planning a batch from the displayed inventory.
@@ -86,10 +93,18 @@ content choice reconciles them; old immutable snapshots are not rewritten.
 
 ## Included guides
 
+`korean-writing` is a core guide on the relevant surface. Before writing, revising,
+or translating Korean responses, documents, slide wording, or UI copy, its router
+requires consulting the complete guide. It contains twelve principles and Korean
+golden examples. A complete copy already in task context need not be reread.
+Core selection does not override explicit off or item disablement, and a router
+instruction does not prove that a model read or followed it.
+
 | Guide | Scope |
 | --- | --- |
 | `cli-multi-model-workflow` | multi-model CLI workflow: Default Frame, role slots/tiers, delegation mechanics, driving Codex CLI directly, cache economy, unattended-batch safety, halt/resume, handoff contract, Environment Binding |
 | `coding-staged-workflow` | staged development: design → process → implement, lightweight path, review loop, severity contract, stop conditions |
+| `ui-design` | operational task flows, evidence and authority, visual hierarchy, design tokens and layout; conditional visual-direction companion |
 | `verification-discipline` | verification depth and per-domain mix (owns the Verification Menus), case space, what a green result is worth |
 | `concept-economy` | concept-surface economy: reuse / extend / rename / split, split triggers, migration compatibility |
 | `documentation-hygiene` | where comments, history, and handoffs belong; how to phrase rules others follow |
@@ -109,9 +124,22 @@ The `slide-writing` guide is selected through `office-work` and
 slide or presentation task. Its companion runbook and scripts are used only for
 an explicitly applicable static HTML/PDF job: preparation derives a job-local
 criteria copy and `ORACLE.json`, then freezes them with the job inputs and runtime
-version. Jobs stay outside immutable corpus snapshots; preparation and result
+version. Jobs stay outside immutable instructions snapshots; preparation and result
 acceptance need Python, while rendering uses the optional dependencies listed in
 `DEPENDENCIES.md`. Native presentation formats remain the user's choice; the
 supplied renderer's mechanical checks apply only to its static HTML/PDF path.
 
-See [recovery](recovery.md) for migration, reset, and ownership conflicts, and [native launch settings](advanced-launch.md#native-hooks-and-agents) before enabling executable corpus content.
+See [recovery](recovery.md) for migration, reset, and ownership conflicts, and [native launch settings](advanced-launch.md#native-hooks-and-agents) before enabling executable instructions content.
+
+## Compatibility
+
+Use `agent-bios instructions` and the `--instructions` family of options for new
+commands. The older `corpus` command and options remain aliases. Canonical
+environment variables use `INSTRUCTIONS`; older `CORPUS` names remain fallbacks,
+and conflicting old/new values are refused.
+
+The private library remains under `~/.config/agent-bios/corpus/`, and the status
+projection remains `~/.local/share/agent-bios/corpus-status.json`. The naming change
+does not relocate data or rewrite old snapshots and session pins. The
+[compatibility contract](instructions-compatibility.md) names retained interfaces,
+their owners, and the conditions for eventual removal.

package/docs/recovery.md

@@ -1,8 +1,8 @@
 # Installation, migration, and recovery
 
-[← Overview](../README.md) · [Setup](setup.md) · [Corpus](corpus.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
+[← Overview](../README.md) · [Setup](setup.md) · [Instructions](instructions.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
 
-Start with `agent-bios status`. Commands use the installed `agent-bios@0.19.1` CLI;
+Start with `agent-bios status`. Commands use the installed `agent-bios@0.19.0` CLI;
 from a source checkout use `bash install.sh <command>`. For first installation and app
 setup recovery, see [Setup](setup.md). Commands below are **operation references, not a sequence to paste and run**. Preview the specific action you need; do not delete an ownership conflict just to make installation succeed.
 
@@ -14,18 +14,18 @@ an immutable release and baseline under agent-bios-owned state. It installs the
 `agent-launch` entrypoint and its own profile/catalog files, but does not change native
 Claude/Codex globals, settings or hooks. Optional app registration adds only its
 owned discovery link, and optional shell connection changes only its owned startup
-wiring. Neither activates corpus in a task.
+wiring. Neither activates instructions in a task.
 
 | Command | Purpose |
 | --- | --- |
-| `npm install -g agent-bios@0.19.1` | install the CLI package; private setup is a separate explicit command |
+| `npm install -g agent-bios@0.19.0` | install the CLI package; private setup is a separate explicit command |
 | `agent-bios install` | open the guided installation UI |
-| `agent-bios install --non-interactive --corpus none` | store runtime with no active corpus |
+| `agent-bios install --non-interactive --instructions none` | store runtime with no active instructions |
 | `agent-bios onboard --non-interactive --domains builder-base,multi-agent-orchestration` | store the named domains with compatibility core/infra selection |
 | `agent-bios setup status --review-id ID` / `resume --review-id ID` | inspect the setup receipt / prepare a safe continuation without executing it |
 | `agent-bios verify` | verify stored bytes/catalog/baseline; not host activation |
 | `agent-bios status` | show the private release, baseline, conflicts, and evidence state |
-| `agent-bios corpus` | rich Corpus Studio in a TTY; list in a non-TTY |
+| `agent-bios instructions` | rich Instructions Studio in a TTY; list in a non-TTY |
 | `agent-launch claude` | open the launch TUI for Claude |
 | `agent-launch codex` | open the launch TUI for Codex |
 | `agent-bios shell restore` | opt in: bare claude/codex opens the TUI |
@@ -34,9 +34,9 @@ wiring. Neither activates corpus in a task.
 | `agent-bios reset --apply --yes --expected-revision REV` | use the revision returned by preview |
 | `agent-bios migrate` | preview legacy global cleanup; --apply --yes performs it |
 | `agent-bios update` | git pull + reinstall (clone), or print the npm update line |
-| `agent-bios uninstall` | remove owned runtime entries; retain user corpus and pinned sessions |
+| `agent-bios uninstall` | remove owned runtime entries; retain user instructions and pinned sessions |
 
-`agent-launch` examples assume `~/.local/bin` is on `PATH`; otherwise use `"$HOME/.local/bin/agent-launch"`. From a checkout, deploy with `bash install.sh install` at its root, not the globally installed CLI. A blocked npm postinstall message does not deploy the corpus; the explicit `install` command remains necessary.
+`agent-launch` examples assume `~/.local/bin` is on `PATH`; otherwise use `"$HOME/.local/bin/agent-launch"`. From a checkout, deploy with `bash install.sh install` at its root, not the globally installed CLI. A blocked npm postinstall message does not deploy the instructions; the explicit `install` command remains necessary.
 
 ## Ownership and legacy migration
 
@@ -185,7 +185,7 @@ agent-bios migrate --apply --yes
 
 Installation and rollback validate the target baseline and personal field/member
 changes together. Conflicts preserve the current selection rather than dropping
-items from a successful snapshot. Installation publishes a complete corpus/config
+items from a successful snapshot. Installation publishes a complete instructions/config
 association under the same lock used by readers. Pending publication is disclosed
 by status and prevents a new configured launch from reading mixed state; bare and
 pinned replay can use the last confirmed immutable release.
@@ -198,4 +198,4 @@ an interrupted reset, or review and accept a fresh revision to replace a stale
 reset intent. Later user changes and replacement credentials are not overwritten
 by the old intent. Nonsecret settings are archived; token bytes never are.
 
-Reset also clears individual on/off overrides and the active trophy display generation. It is not a deletion of uploaded learning records. See [corpus controls](corpus.md) and [learning discoveries](understand.md).
+Reset also clears individual on/off overrides and the active trophy display generation. It is not a deletion of uploaded learning records. See [instructions controls](instructions.md) and [learning discoveries](understand.md).

package/docs/releases/0.19.2.md

@@ -0,0 +1,38 @@
+# agent-bios 0.19.2
+
+Release changes relative to 0.19.1.
+
+## Korean writing guidance
+
+The core library includes `korean-writing`, a relevant guide consulted before Korean
+writing, revision, or translation. Twelve principles and Korean golden examples
+cover meaning preservation, evidence and uncertainty, terminology, conditions,
+proposal/completion distinctions, and UI wording that matches actual behavior.
+The guide applies to responses, documents, slide wording, and UI copy. Selection
+and per-item disablement remain explicit; routing is not proof of model reading.
+
+## Instructions naming and upgrade compatibility
+
+The rules, guides, and procedures component is named **Instructions**, and its
+management UI is **Instructions Studio**. The canonical CLI is
+`agent-bios instructions`; launcher, setup and app guidance use the same naming.
+
+Older interfaces continue to work against the same implementation and store.
+Conflicting old/new environment settings are rejected instead of choosing another
+store. Existing storage locations, serialized identities, session pins and the
+shared writer lock are preserved. Old-release launcher and app-bridge adapters
+remain available during upgrades. This compatibility layer is a future removal
+item, subject to the [documented removal conditions](../instructions-compatibility.md#removal-condition).
+
+The conversational setup guide names the actual retained-inventory response fields,
+so the agent can present previously stored personal instructions during setup.
+
+## Product scope
+
+The documented purpose centers on selecting, sharing and continuing team work
+environments. The implemented product remains the Instructions library and host
+launch/delivery integration. Domain knowledge, Decision memory and shared environment
+editions are design directions, not features introduced by this release.
+
+Interactive setup, language selection, app connection and local instruction capture
+were already available in 0.19.1.

package/docs/releases/0.19.3.md

@@ -0,0 +1,107 @@
+# agent-bios 0.19.3
+
+Release changes relative to 0.19.2.
+
+## Windows without npm or WSL
+
+Windows installs as scripts and data bound to an approved CPython 3.13, with no
+custom executable and no EXE installer. It targets machines whose application
+control permits PowerShell and Python but blocks unrecognized executables. The
+bootstrap reuses an approved CPython already on the machine, or provisions the
+pinned official embeddable runtime into an application-private folder; neither an
+existing Python nor its packages are modified.
+
+The installed commands are `agent-bios` and `agent-launch`. Each is a static
+wrapper plus a generated `.cmd` shim, because Windows does not treat `.ps1` as
+executable: without the shim a PATH directory holding only wrappers is reachable
+from PowerShell alone, and the Command Prompt, the Run box and any program that
+spawns a command find nothing while PATH is correct. The shim is generated at
+installation rather than shipped, since Authenticode cannot sign a `.cmd`; its
+bytes are fixed and recorded in the deployment's ownership inventory, so a
+modified shim is refused exactly like a modified command.
+
+`agent-bios uninstall` removes owned commands, shims, shortcuts and the owned
+PATH entry. Private instructions, session records and any preexisting Python are
+retained.
+
+## A fixed installation address
+
+The project's installation page serves one explicitly promoted release at a fixed
+URL, so publishing another release does not silently change what the address
+installs. It carries a one-line command per platform.
+
+Each command relaxes the PowerShell execution policy for the single process it
+starts, because the default policy on an unconfigured Windows client is
+Restricted and refuses a downloaded script. The machine's saved policy is not
+edited, and a domain-imposed policy still refuses through the invocation's own
+error. The Windows command names the interpreter, so it can be pasted into the
+Command Prompt or the Run box rather than only into an open PowerShell.
+
+The macOS and Linux command installs the published npm release and stops, naming
+the command that deploys the work environment: `curl | bash` gives the shell a
+pipe, and the interactive chooser needs a terminal. It also reports the case npm
+leaves behind — a successful global install whose command the caller's PATH does
+not carry.
+
+Windows releases come in two channels. Stable is Authenticode-signed and
+timestamped. Preview is published without a signing identity and refuses to run
+unless the caller adds `-AcceptUnsignedPreview`; the hashes pinned inside the
+bootstrap still guard the manifest, the application archive and the runtime
+archive.
+
+## Claude Desktop
+
+Claude Desktop opens conversations without a launcher and exposes no skill folder an
+installer can reach ahead of time, so agent-bios reaches it as a local extension the
+user installs once. `agent-bios app desktop` writes a `.mcpb` file under the private
+state root; opening it with Desktop installs a standard-library stdio MCP server with
+four tools: status, preview, use and off. agent-bios writes nothing into Desktop's own
+folders or configuration.
+
+`use` returns the saved installation selection, or a selection named in the request,
+as context for that conversation only. Nothing reaches a conversation that did not
+ask, including one already in progress. Desktop sends no conversation identity with a
+tool call, so the first preview or use returns a `session_id` that later calls pass
+back; a repeated use with it returns the same text without recording a second
+delivery. The returned text ends with an `agent-bios end` line, and reporting that
+line through status records that the text was read to its end. It does not record
+whether Desktop kept the text in the message or saved it to a file, which Desktop
+does not reveal to the extension.
+
+The file names the absolute Python that generated it, because Desktop resolves a bare
+`python3` through the login shell's PATH, which on stock macOS reaches 3.9. The server
+finds the current installation on every call, so an agent-bios update needs no new
+file; changing that Python does. Desktop delivery has no project scope, so
+project-scoped imported instructions are not included, in chat or in the Code tab.
+
+## Installation reports an unreachable launcher
+
+Installation now says when the `agent-launch` it deposits cannot be typed: when
+`~/.local/bin` is not on the shell's PATH, and when the optional zsh connection that
+routes a bare `claude` or `codex` through the launcher is off. Both used to end in a
+success message followed by `command not found`. A reachable installation reports
+nothing new. The terminal interface, the conversation client and the non-interactive
+path share the message.
+
+## Instruction guides
+
+The multi-model workflow, staged coding workflow, LLM capability boundary,
+tooling gotchas, verification discipline and UI design guides were revised, and
+lessons paid for on a separate deployment line were carried across where this
+line had not already removed them. The tooling-gotchas hook was updated with
+them.
+
+## Not established by this release
+
+No signing identity exists yet, so the published Windows channel is the unsigned
+preview. Approval by an organization policy or endpoint-protection product is not
+established: a green workflow proves the route on a hosted runner where
+PowerShell and Python are permitted. Migration of an existing EXE installation
+into the script route is refused rather than attempted.
+
+The Claude Desktop route was measured on Desktop 2.16120.0 for macOS, in chat and in
+the Code tab. It is not available on Windows, and a later Desktop version is not
+established to load the file until someone checks it.
+
+Domain knowledge, Decision memory and shared environment editions remain design
+directions rather than features of this release.

package/docs/session-model.md

@@ -1,24 +1,24 @@
 # Sessions, storage, and verification
 
-[← Overview](../README.md) · [Setup](setup.md) · [Corpus](corpus.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
+[← Overview](../README.md) · [Setup](setup.md) · [Instructions](instructions.md) · [Sessions](session-model.md) · [Recovery](recovery.md) · [Launch](advanced-launch.md) · [Understand!](understand.md)
 
 **Library → selection and edits → immutable snapshot → configured session**
 
-## Corpus ownership
+## Instructions ownership
 
 Single source of truth for the instructions and scoped guides supplied to activated
 Claude Code or Codex CLI sessions, or explicitly chosen Codex app tasks. Edit once;
-the private corpus compiler projects the selected content through the chosen delivery route.
+the private instructions compiler projects the selected content through the chosen delivery route.
 
-A deployable instruction corpus for coding agents, plus the CLI that
-installs, verifies, and evolves it. The npm package ships the corpus; `install.sh` is both the
+The product supplies deployable instructions for coding agents and the CLI that
+installs, verifies, and evolves them. The npm package ships the instructions; `install.sh` is both the
 `agent-bios` CLI entry and the deployer.
 
 ## Authored layers
 
 Two authored layers:
 
-- **Always-in-an-activated-session instructions** — `claude/CLAUDE.md` is the English canonical and `codex/AGENTS.md` its generated host projection. They hold compact invariants, decision principles, and guide pointers. A clean private install does not copy either file into a host's global discovery path; `compose/corpus_catalog.py` inventories their registered items and compiles the selected rules into an immutable session snapshot.
+- **Always-in-an-activated-session instructions** — `claude/CLAUDE.md` is the English canonical and `codex/AGENTS.md` its generated host projection. They hold compact invariants, decision principles, and guide pointers. A clean private install does not copy either file into a host's global discovery path; `compose/instructions_catalog.py` inventories their registered items and compiles the selected rules into an immutable session snapshot.
 - **Scoped guides** — `claude/guides/`, with generated Codex mirrors. The snapshot compiler copies selected guides to private generation-qualified paths and emits their router. Procedures, tables, numbers, and environment-specific content live here.
 
 ## Activation and Vanilla
@@ -33,19 +33,27 @@ instructions still follow the host's normal loading rules. `--resume-session ID`
 loads the recorded host/session pin rather than resolving current defaults.
 
 Per-item on/off overrides take precedence over default-mode launch-domain selection.
-No-corpus mode emits no corpus. Explicit selected mode keeps inclusion within its
+No-instructions mode emits no instructions. Explicit selected mode keeps inclusion within its
 targets; disabled items stay excluded and unrelated enabled overrides do not leak in.
-Imported items also respect their host/project scope. See [corpus selection](corpus.md#manage-the-library). Existing host globals are loaded by default; [selective exclusion](advanced-launch.md#global-instruction-files) is a separate host-specific option.
+Imported items also respect their host/project scope. See [instructions selection](instructions.md#manage-the-library). Existing host globals are loaded by default; [selective exclusion](advanced-launch.md#global-instruction-files) is a separate host-specific option.
 
 ## Codex app tasks
 
-App tasks have an explicit [use/off workflow](setup.md#use-corpus-in-a-codex-app-task).
+App tasks have an explicit [use/off workflow](setup.md#use-instructions-in-a-codex-app-task).
 Use returns a selected snapshot through the tool/context path and records a
 `returned-as-context` receipt separately from native CLI pins. Stored registration
 and returned text do not prove native app discovery or model reading. Off stops
 further managed use, but cannot retract earlier context; a new task is needed for
 clean exclusion. Management and instruction capture do not activate task context.
 
+## Claude Desktop conversations
+
+Desktop conversations pull through the [Desktop extension](setup.md#use-instructions-in-claude-desktop).
+Use returns the selected snapshot as a tool result and records a `returned-as-context`
+receipt under an identity agent-bios mints, because Desktop sends none. An end-marker line
+reported back confirms the text was read to its end, not that it stayed inline. Nothing
+reaches a conversation that did not ask, including one already in progress.
+
 ## Storage layout
 
 From a clone, `bash install.sh install` is the same default path. The release lives
@@ -53,11 +61,14 @@ under `~/.local/share/agent-bios/runtime/releases/`; baseline tuples and transac
 journals live under that runtime root, immutable snapshots and pins under
 `~/.local/share/agent-bios/sessions/`, and user packages, overlays, tombstones,
 learnings, history, and trash under `~/.config/agent-bios/corpus/`. The corresponding
-`AGENT_BIOS_STATE_DIR` and `AGENT_BIOS_CORPUS_DIR` environment variables relocate
+`AGENT_BIOS_STATE_DIR` and `AGENT_BIOS_INSTRUCTIONS_DIR` environment variables relocate
 those private roots; `AGENT_LAUNCH_VENV` relocates the managed dependency environment.
+The `corpus/` directory name is retained for storage compatibility; the naming
+change does not move user data. Older environment names remain
+[compatibility aliases](instructions.md#compatibility).
 The UI entrypoints use their verified process-temporary bundle without
 requiring a preinstalled Textual runtime.
-The owned `agent-launch` entrypoint exports `AGENT_BIOS_PRIVATE_CORPUS=1` and the
+The owned `agent-launch` entrypoint exports `AGENT_BIOS_PRIVATE_INSTRUCTIONS=1` and the
 immutable `AGENT_BIOS_PACKAGE_ROOT`; the launcher also recognizes the private install
 record when the explicit marker is absent. These select the private runtime and do not
 claim that any host session has loaded a snapshot.
@@ -70,26 +81,26 @@ the install record, and verifies the owned launcher/profile/status projections.
 result says `activation: unverified`: neither stored bytes nor a dry-run argv proves a
 host loaded the snapshot.
 
-Corpus selection seeds future activated sessions, not plain CLI/Vanilla. A one-off
-`--corpus-domains` selection applies only to that launch. The store composes an
+Instructions selection seeds future activated sessions, not plain CLI/Vanilla. A one-off
+`--instructions-domains` selection applies only to that launch. The store composes an
 immutable `ContentRef`; Codex startup preserves the effective native developer
-instructions, injects the private corpus and dynamic launch contract, creates and
+instructions, injects the private instructions and dynamic launch contract, creates and
 reads back a durable host thread, then records its pin. Claude uses the per-call
 append and requested session id and records a pin only after observing that id in the
 native session log. Real-host probes cover Codex and Claude first-turn delivery,
-including corpus propagation to a launcher-generated Claude workhorse. Claude
+including instructions propagation to a launcher-generated Claude workhorse. Claude
 resume restores the pin's exact environment provenance rather than changing an unset
 config-home variable into an explicit default. Post-fix authenticated resume remains
 unverified. Snapshot pin integrity alone does
 not establish that a resumed model request succeeded.
 
 The native Claude plugin bootstrap has advertised selected plugin roots and
-qualified corpus agents. An edited corpus `SessionStart` hook ran automatically
+qualified instructions agents. An edited instructions `SessionStart` hook ran automatically
 through its generated plugin. Codex 0.153.4 discovery retains user, project and session
-hooks alongside the selected corpus. A real-host test with a local transport verifies
+hooks alongside the selected instructions. A real-host test with a local transport verifies
 that a generated `SessionStart` hook runs and injects context after its exact definition
 is trusted; the untrusted control does neither. This test uses no external model.
-Authenticated corpus-agent execution and native skill-menu registration remain
+Authenticated instructions-agent execution and native skill-menu registration remain
 unverified; they are separate from hook delivery and launcher-tier child evidence.
 
 Pins preserve environment provenance rather than reconstructing it: the host's
@@ -107,7 +118,7 @@ refuse native `--cd`/`-C` cwd overrides before delivery; change to the target pr
 first so snapshot scope and native execution agree. Ordinary CLI/Vanilla forwarding
 keeps its native behavior.
 
-Snapshots outside no-corpus mode retain the immutable private management bootstrap
+Snapshots outside no-instructions mode retain the immutable private management bootstrap
 and put its exact path in the injected startup text, so `$agent-bios` has private
 procedure access even without native skill discovery. Selected requested procedures are exposed the same
 way. This is not a claim that either host registered them in its native skill menu;
@@ -115,6 +126,6 @@ native skill registration remains unverified.
 
 ## Scope
 
-The repository and npm package contain corpus sources and private runtime machinery,
+The repository and npm package contain instructions sources and private runtime machinery,
 not a user's authoring state, learning events, snapshots, pins, activation journals,
 credentials, native settings, or generated temporary files.