@plurnk/plurnk-meta 2.1.3 → 2.1.5

package/CORPUS.md

@@ -11,7 +11,7 @@ Authored defaults published by `@plurnk/plurnk-meta` and consumed by
 | `POLICY.md`                     | Read before the first-run seed of the user-owned XDG config `AGENTS.md`.     |
 | `recap.md`                       | Read for an optional Recap rendered last when its content is nonempty.       |
 | `docs/*.md`                     | Built-in reference pages (`env.md`, `members.md`, `ops.md`, `reasoning.md`, `session.md`, `workspace.md`) read when registered pull docs are materialized. |
-| `skills/plurnk/**`              | Standard skill entry (`SKILL.md`) and reference chapters (`references/configuration.md`, `references/extensibility.md`); Core composes its package-owned and generated resources (`copy-move.md`, `models.md`, `.env.defaults`). |
+| `skills/plurnk/**`              | Standard skill entry (`SKILL.md`) and reference chapters (`references/configuration.md`, `references/extensibility.md`); Core composes its package-owned and generated resources (`copy-move.md`, `models.md`, `plurnk.service`, `plurnk.nginx.conf`, `.env.defaults`). |
 
 Core materializes eligible scheme and runtime pull docs at
 `plurnk:///<name>.md` or `<scheme>:///<name>.md` and exposes them through turn-0 FIND surveys. Merely

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@plurnk/plurnk-meta",
-  "version": "2.1.3",
+  "version": "2.1.5",
   "description": "Extension discovery and trust primitives shared by every plurnk kind, and the teaching material the plurnk service gives models: policy, Recap, the plurnk skill and scheme references.",
   "keywords": [
     "plurnk",
@@ -68,5 +68,5 @@
     "url": "git+https://github.com/plurnk/plurnk-service.git",
     "directory": "plurnk-meta"
   },
-  "gitHead": "db65532b51408a2014f7ff53a288d834ff6694f2"
+  "gitHead": "609126ca4725442150214f4fcb3774f6fade89d9"
 }

package/skills/plurnk/SKILL.md

@@ -1,24 +1,28 @@
 ---
 name: plurnk
-description: Plurnk harness configurations, extensions, and definitions.
+description: Configure, customize, extend, and integrate Plurnk — settings (.env), models, operating policy (AGENTS.md), personas, skills, MCP servers, schedules, hooks, plugins, extensions, ACP clients, and systemd/nginx templates.
 ---
 
 # Plurnk
 
 | Need | Reference |
 | --- | --- |
-| Configuration cascade, persistence, permissions, skills | [Configuration](references/configuration.md) |
+| Configuration cascade, persistence, permissions, skills, personas | [Configuration](references/configuration.md) |
+| Daemon status, effective configuration, and validation | `plurnk-service config`, `plurnk-service config show`, `plurnk-service config check`, `GET /health` |
+| Systemd user service unit and nginx reverse-proxy templates | [plurnk.service](plurnk.service), [plurnk.nginx.conf](plurnk.nginx.conf) |
 | COPY/MOVE operands, source regions, destination placement | [Copy and move](references/copy-move.md) |
 | Model routes, reasoning, output and context limits, cost | [Models](references/models.md) |
 | Every installed package's settings and defaults | [Complete .env.defaults](.env.defaults) |
+| Operating policy (`AGENTS.md` / `AGENTS.local.md`) and session instructions | `$XDG_CONFIG_HOME/plurnk/AGENTS.md`, `<project>/AGENTS.md` (`.local.md`), `session://<session>/AGENTS.md` |
+| Personas (`persona://` and `plurnk_personas`) | [Personas](mcp:///plurnk_personas.md), `persona://<name>/PERSONA.md` |
+| Agent Skills, MCP servers, environment variables, and schedules | `mcp:///plurnk_skills.md`, `mcp:///plurnk_mcp.md`, `mcp:///plurnk_env.md`, `mcp:///plurnk_schedule.md` |
+| Which project files you can see, and why a file is missing | [Membership](mcp:///plurnk_members.md) |
 | Plugins, extensions, clients and hooks: what each reaches and how it loads | [Extending Plurnk](references/extensibility.md) |
 | Each installed package's contract and type declarations | `packages/<package name>/SPEC.md`, `dist/*.d.ts` |
 | Session hierarchy, messaging, and delegation | [Session reference](plurnk:///session.md) |
-| Which project files you can see, and why a file is missing | [Membership](mcp:///plurnk_members.md) |
-| Agent Skills, MCP servers, environment variables, and schedules | `mcp:///plurnk_skills.md`, `mcp:///plurnk_mcp.md`, `mcp:///plurnk_env.md`, `mcp:///plurnk_schedule.md` |
 | Running code: the shell and each interpreter's body, stdin, arguments, exit status | `plurnk:///sh.md`, `plurnk:///js.md`, `plurnk:///python3.md`, … |
 | Enabled MCP server invocations and schemas | `mcp:///` |
 | Built-in executors and scheme references | `plurnk:///` |
 
-READ the relevant reference; FIND can select paths and search their bodies.
-Defaults describe available settings, not the operator's effective configuration.
+READ the relevant reference; FIND selects paths and searches their bodies or symbols across `skill://plurnk/**` (for example `FIND (skill://plurnk/packages/*/SPEC.md) ~ /§tag/` or `FIND (skill://plurnk/packages/**/dist/*.d.ts) & /Symbol/`), and `KILL (log:///...)` retires read slices once applied.
+Defaults describe available settings; `plurnk-service config` and `plurnk-service config show` inspect the operator's effective configuration.

package/skills/plurnk/references/extensibility.md

@@ -12,12 +12,17 @@ The surfaces run from least to most code and trust, and each is for what the one
 before it cannot do:
 
 - **Configuration** when what the integration needs already exists: a setting in
-  the cascade, or a definition in `mcp.json` or a skill root.
+  the cascade, or a definition in `mcp.json`, a skill root, or a persona root.
+- **Operating policy** (`AGENTS.md` / `AGENTS.local.md`) when standing instructions
+  apply to every turn in a user environment, project, or session (`session://<session>/AGENTS.md`).
+- **A persona** (`persona://<name>/PERSONA.md`, managed via `plurnk_personas` or `EDIT`)
+  when a reusable operating identity, optional model/effort route, and instruction body
+  can be selected per session or copied to `session://<session>/AGENTS.md`.
 - **A skill** when the model needs knowledge, or a script it runs with its existing
   runtimes (`sh`, `js`).
 - **An MCP server** when X has an API the model calls as tools. A server X already
   publishes needs only its definition.
-- **A plugin** when skills and MCP servers install together, or travel to other agents.
+- **A plugin** when skills, personas, and MCP servers install together, or travel to other agents.
 - **A client** when X drives Plurnk from outside through ACP.
 - **A hook** when X hears about Plurnk's lifecycle events and answers nothing.
 - **An extension** only when the model needs something new inside the daemon that
@@ -31,24 +36,25 @@ before it cannot do:
 
 | Surface | Shape | The model sees | Code in the daemon |
 | --- | --- | --- | --- |
+| Persona | `<name>.md` or `<name>/PERSONA.md` (or inline definition) | `persona://<name>/PERSONA.md` | none |
 | Skill | a directory holding `SKILL.md` | `skill://<name>/` | none |
 | MCP server | one MCP definition | a runtime named for the server, its tools as invocation targets, and its resources as an address scheme | none: a process or a URL |
-| Plugin | a directory holding `plugin.json`, `skills/` and `mcp.json` | its skills and MCP servers | none, unless it declares an extension |
+| Plugin | a directory holding `plugin.json`, `skills/`, `personas/`, and `mcp.json` | its skills, personas, and MCP servers | none, unless it declares an extension |
 | Extension | a package declaring one kind | a runtime, an address scheme, a media-type reader, a model provider, a page materializer, or what a module contributes | trusted, in the daemon |
 
-Skills and MCP servers are definitions. The `plurnk_skills` and
-`plurnk_mcp` runtimes (`add`, `replace`, `enable`, `disable`, and `remove` tools) manage them in the current workspace,
-and `mcp:///plurnk_skills.md` and `mcp:///plurnk_mcp.md` show each
+Personas, skills, and MCP servers are definitions. The `plurnk_personas`, `plurnk_skills`, and
+`plurnk_mcp` runtimes (`add`, `replace`, `enable`, `disable`, and `remove` tools; personas also accept `EDIT` and `KILL` on `persona://`) manage them in the current workspace,
+and `mcp:///plurnk_personas.md`, `mcp:///plurnk_skills.md`, and `mcp:///plurnk_mcp.md` show each
 definition's form. A definition needs no restart; an extension loads only when
 the daemon starts.
 
 ## X uses Plurnk
 
-| Surface | Shape | Code in the daemon |
-| --- | --- | --- |
-| Application | an ACP v2 client: WebSocket or Streamable HTTP JSON-RPC 2.0 at `/acp` on the daemon's listener (`PLURNK_PORT`) | none |
-| Command | the executable in `PLURNK_HOOKS_COMMAND`, run once per event named in `PLURNK_HOOKS_EVENTS` with one JSON line on stdin; its output and exit status change nothing | none |
-| Protocol server | a module that declares mounts and serves its routes on the daemon's listener | trusted, in the daemon |
+| Surface | Shape | Code in the daemon | Contract |
+| --- | --- | --- | --- |
+| Application | an ACP v2 client: WebSocket or Streamable HTTP JSON-RPC 2.0 at `/acp` on the daemon's listener (`PLURNK_PORT`), authenticated via OAuth 2.1 (`/.well-known/*`, `/oauth/*`) when `PLURNK_AUTH=1` | none | `skill://plurnk/packages/@plurnk/plurnk-acp/SPEC.md`, `skill://plurnk/packages/@plurnk/plurnk-auth/SPEC.md` |
+| Command | the executable in `PLURNK_HOOKS_COMMAND`, run once per event named in `PLURNK_HOOKS_EVENTS` with one JSON line on stdin; its output and exit status change nothing | none | `skill://plurnk/packages/@plurnk/plurnk-hooks/SPEC.md` |
+| Protocol server | a module that declares mounts and serves its routes on the daemon's listener | trusted, in the daemon | `skill://plurnk/packages/@plurnk/plurnk-modules/SPEC.md` |
 
 ## Extensions