@plurnk/plurnk-meta 1.21.1 → 1.22.0

package/README.md

@@ -1,6 +1,6 @@
 # @plurnk/plurnk-meta
 
-The plurnk metaproject layer, published — what the family shares that no single member owns.
+What plurnk's packages share that no single package owns: plugin discovery and trust, and the teaching material the service gives models.
 
 **Membership primitives** (`import Meta from "@plurnk/plurnk-meta"`): the shared implementation of exact family identity, trust, attribution normalization, enumeration, and root resolution consumed by the four family-owned scanners (schemes, mimetypes, providers, execs). [SPEC.md](./SPEC.md) owns the complete contract ({§plugin-discovery}).
 
@@ -14,6 +14,4 @@ The plurnk metaproject layer, published — what the family shares that no singl
 
 **The teaching corpus**: authored policy, an optional Recap, the Plurnk skill entry, and built-in scheme references resolved from this installed package. Meta owns the source bytes and membership; core owns admission, resource composition, and projection. See [`CORPUS.md`](./CORPUS.md) and {§teaching-corpus}.
 
-[`DOGFOOD.md`](./DOGFOOD.md) defines the whole-product, outside-client acceptance gate for daily-driver and release readiness.
-
 Third-party plugin authors: your package is discovered under any scope through one string `plurnk.kind`, enumerated by these primitives, and gated before import by the operator's trust knob — no registration with us required. An admitted package may declare always-on `plurnk.attribution` tags and its loaded plugin object may decide per provider attempt whether to return additional tags from `attributions(context)` ({§plugin-attribution}). MIT.

package/docs/worker.md

@@ -9,10 +9,10 @@ an entry rather than controlling that worker.
 
 | Address | Meaning | Model access |
 | --- | --- | --- |
-| `worker://alice` | Named worker | WORK/FORK create; SEND messages; READ collects; KILL terminates. |
-| `message://alice/ab3d5678` | Retained message | READ/FIND/COPY inspect; SEND replies. Neither EDIT nor KILL changes its source. |
-| `ops://alice/1` | What that loop said, or how it ended | READ/FIND/COPY inspect; source is immutable. |
-| `worker://alice/example.md` | Named scratch entry | Read and write from any worker in the workspace. |
+| `worker://exampleWorkerName` | Named worker | WORK/FORK create; SEND messages; READ collects; KILL terminates. |
+| `message://exampleWorkerName/ab3d5678` | Retained message | READ/FIND/COPY inspect; SEND replies. Neither EDIT nor KILL changes its source. |
+| `ops://exampleWorkerName/1` | What that loop said, or how it ended | READ/FIND/COPY inspect; source is immutable. |
+| `worker://exampleWorkerName/example.md` | Named scratch entry | Read and write from any worker in the workspace. |
 | `worker:///example.md` | Shared commons entry | Read and write. |
 
 The packet's `## Worker` block, below the log, names your worker, its parent
@@ -67,11 +67,11 @@ is empty when an already-delivered answer stands. While live work remains, that
 KILL joins it without cancelling it and delivers no final answer.
 
 Each child task's conclusion wakes its waiting parent and arrives once, as an
-`_plurnk` READ of `ops://alice/1` carrying what the child said: the
+`_plurnk` READ of `ops://exampleWorkerName/1` carrying what the child said: the
 execution outcome, not another message, and not delivered again as a reply. A
 failure retains its status and Problem. READ that address with a scope to
 inspect more of the exact result, even after the child starts another task.
-Bare `READ (worker://alice)` collects the current result instead, naming its
+Bare `READ (worker://exampleWorkerName)` collects the current result instead, naming its
 exact source in `resource`; while the child is running it returns `425`. A
 result does not imply that every task in that worker has finished.
 
@@ -123,11 +123,11 @@ The worker name is required. Any worker in the workspace can READ these
 sources; `log:///` remains your own context. READ brings the selected source
 into your log, and KILL of that READ curates only its log projection, never the
 original evidence. The current turn's reasoning exists when its OPs execute, so
-`READ (reasoning://alice/1/3) <1,-1>` from `worker://alice` on loop 1,
+`READ (reasoning://exampleWorkerName/1/3) <1,-1>` from `worker://exampleWorkerName` on loop 1,
 turn 3 retains it; a turn that produced no reasoning reads empty, and a turn
 that has not happened returns a missing-source result. READ never requests
 inference. NOTE retains working memory without changing the loop state; only
 NOTE also executes from exposed reasoning, while quoted examples and other
 reasoned operations remain data. Notes are ordinary log items whose read-only
-sources stay searchable (`FIND (note://alice/**) /parser/`) and READable
+sources stay searchable (`FIND (note://exampleWorkerName/**) /parser/`) and READable
 after curation or a FORK.

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@plurnk/plurnk-meta",
-  "version": "1.21.1",
-  "description": "The plurnk metaproject layer, published: plugin-membership primitives (the ONE trust rule, scope-agnostic symlink-aware enumeration, deployment-root resolution), the project teaching corpus (personality, Recap, scheme docs), and the home for family tooling.",
+  "version": "1.22.0",
+  "description": "Plugin discovery and trust primitives shared by plurnk plugin families, and the teaching material the plurnk service gives models: policy, Recap, the plurnk skill and scheme references.",
   "keywords": [
     "plurnk",
     "agent-skills",
@@ -18,7 +18,6 @@
     "dist/**/*",
     "README.md",
     "SPEC.md",
-    "DOGFOOD.md",
     "CORPUS.md",
     "POLICY.md",
     "recap.md",

package/DOGFOOD.md

@@ -1,94 +0,0 @@
-# Metaproject orientation-readiness gate
-
-This opt-in capstone asks a floor model to inspect an assembled open-project
-forest through the outside client and deliver an evidence-bearing orientation
-report. It is intentionally expensive: a fresh campaign semantically prepares
-the complete forest. Execute it only after the focused package, integration,
-demo, live-model, and modest-candidate paths are healthy—not as a routine
-debugging loop or release gate.
-
-Invoke it with:
-
-```sh
-PLURNK_ACCEPTANCE_PROJECT_ROOT=/path/to/open-project-forest \
-PLURNK_CLIENT_CHECKOUT=/path/to/open-client \
-  npm run readiness:metaproject -- --model <selector> --requiem --preserve
-```
-
-Both paths are explicit preconditions. The runner never guesses a sibling
-checkout or treats another organization under a shared parent directory as
-part of the open project.
-
-This gate does not replace the evidence ladder that qualifies it:
-
-1. Deterministic package and integration coverage proves owned contracts.
-2. An assembled modest-repository candidate smoke proves the production client
-   and daemon can work together.
-3. Focused `live/demo` stories are toy tasks that establish the system is sane
-   enough to test seriously.
-4. The separate bench lane runs real third-party agentic benchmarks and preserves
-   their digest, reasoning, and requiem evidence.
-5. Only after those layers are healthy does this full-forest campaign ask whether
-   PLURNK is ready to develop PLURNK itself.
-
-The canonical `plurnk-service` checkout owns this gate. Its doctrine lives in
-`plurnk-meta` because the report crosses service, contracts, plugins, the
-outside client, project membership, AG-UI, model routing, persistence, and
-forensic digestion.
-
-## Preconditions
-
-- A clean canonical `plurnk-service` checkout with its gate green.
-- The explicit outside open-client checkout built from its repository head.
-- Every default-installed optional provider resolvable from the daemon.
-- An explicitly assembled open-project forest with its root `AGENTS.md`; shared
-  parent directories containing other organizations are not valid substitutes.
-- One inexpensive capable model selector and, optionally, one local smoke-test selector.
-
-Missing preconditions are RED outcomes. The runner never silently skips a phase.
-
-## Contract
-
-One campaign proves the following through production client and AG-UI surfaces:
-
-1. **Candidate build.** The exact service and explicit outside-client checkouts build.
-2. **Clean bootstrap.** A canonical daemon starts on a new database and the
-   client creates a user-named workspace over the explicit forest.
-3. **Inspection.** The model performs successful retrievals, including READ,
-   before answering and names multiple inspected repository artifacts.
-4. **Comprehension.** The report covers service, contracts/DSL, client/AG-UI,
-   repository topology, current-work evidence, and confidence-limiting gaps.
-   If the canonical forge is unavailable inside the run, it must say that the
-   current goal is unverified rather than promote archived GitHub history.
-5. **Forensics.** The supported digest captures the database, packet, reasoning,
-   usage, and operation evidence used by the deterministic verdict.
-
-Proposal modes, reconnect, model hot-swap, and restart/resume are not asserted
-by this executable slice. Expansion of the capstone remains tracked in #3.
-
-Passing individual package tests or receiving HTTP 200 is insufficient. The user-visible client must remain usable and the requested terminal deliverable must be present.
-
-## Evidence
-
-Each preserved campaign atomically claims `benchmarks/run<N>-orientation/` and writes:
-
-- `workspace` — the generated workspace name;
-- `prompt.md` — the exact orientation request;
-- `client.json` and `client.stderr.log` — the client result and diagnostics;
-- `service.stdout.log` and `service.stderr.log`;
-- `phases.json` — commands, timings, and exit statuses without secrets;
-- `verdict.json` — lifecycle, publication, inspection, evidence, and coverage checks;
-- `plurnk.db` — the preserved database;
-- `digest/` — `digest.md`, `digest.json`, `reasoning.md`, and packet sections;
-- requiem artifacts in `digest/` when `--requiem` is requested.
-
-The database is evidence, never the diagnostic interface. Assertions consume client output, AG-UI events, and digest artifacts.
-
-## Posture
-
-- Execute in the foreground and fail hard.
-- Preserve every failing specimen; preserve passing specimens when requested.
-- Never reuse a database for a clean-bootstrap comparison.
-- A provider/configuration/install failure is a gate failure, not a skipped model test.
-- Never infer current work from an archived issue or an uninspected forge reference.
-- File every cross-family defect on the canonical forge with its owning lane label and specimen path.