@plurnk/plurnk-meta 1.5.0 → 1.6.0

package/CORPUS.md

@@ -14,7 +14,7 @@ Authored defaults published by `@plurnk/plurnk-meta` and consumed by
 | `docs/questions.md`             | Read only when operator questions are enabled, then materialized as a doc.   |
 
 Core materializes eligible pull docs at `worker://plurnk/docs/<name>.md` and
-exposes them through the turn-0 `FIND(worker://plurnk/docs/**)` catalog. Merely
+exposes them through the turn-0 `## FIND0 [+init,+docs] (worker://plurnk/docs/**)` catalog. Merely
 placing a file in `docs/` does not register a scheme or make speculative
 teaching current. Every listed source is a required package member; a missing
 or failed read surfaces at the admission boundary rather than silently reducing

package/PLURNK_PERSONALITY.md

@@ -1,9 +1,7 @@
-- You answer inquiries and exploratory questions without changing code unless the user also asks for action.
-- You distill reusable knowledge into searchable, folksonomically tagged knowledgebase entries.
-- You follow modern standards and conventions unless a product property requires otherwise.
-- You verify unstable facts and answer from primary sources when possible.
-- You prove deterministically verifiable work before concluding.
+YOU SHOULD answer inquiries and exploratory questions without changing state unless the user also asks for action.
+YOU SHOULD follow modern standards and conventions unless a product property requires otherwise.
+YOU SHOULD verify unstable facts and answer from primary sources when possible.
+YOU SHOULD prove deterministically verifiable work before concluding.
 
-YOU MUST evaluate the prompt for completion conditions and track multiple conditions in a Markdown checklist.
-YOU MUST evaluate the prompt for decomposability, delegating to WORKers iff the work is naturally divisible.
-YOU MUST react and respond to the latest prompt listed under Active User Prompts.
+YOU MUST identify and track the prompt's completion conditions.
+YOU MUST evaluate the prompt for decomposability, delegating distinct subtasks to WORKers or BARE requests.

package/docs/log.md

@@ -5,7 +5,7 @@ Every operation is recorded at `log:///<loop>/<turn>/<seq>`. The row identifies
 ## Log and catalog
 
 - **Log** - operations and results in order.
-- **Catalog** - `FIND(scheme:///**)` lists the resources a scheme currently holds.
+- **Catalog** - `## FIND0 (scheme:///**)` lists the resources a scheme currently holds.
 
 ## Visibility
 
@@ -19,7 +19,9 @@ OPEN and FOLD change packet visibility, not history. Applying the current state
 
 KILL permanently erases a log row.
 
-The Budget section reports only the packet ceiling, usage, percentage, and free capacity. Each
-log row carries its own token weight. If a packet exceeds its ceiling, the engine folds eligible
-open rows from the newest turn boundary; it never selects older history by relevance. Continued
-overflow follows the reported recovery or hard-413 contract.
+The Budget section reports the packet ceiling, usage, percentage, and free capacity; negative
+free capacity adds one curation alarm. Each log row carries its own token weight. If a packet
+exceeds the gauge, the engine records a nonterminal 413 Problem, folds eligible open rows from
+the newest turn boundary, and tags them `overflow`; it never selects older history by relevance.
+Remaining ruler debt may proceed when the request fits the hard context envelope; otherwise a
+separate terminal 413 stops admission before generation.

package/docs/questions.md

@@ -1,5 +1,8 @@
-# Operator questions — `SEND[300]`
+# Operator questions — SEND signal 300
 
-Enabled here: you may ask the operator when a decision is genuinely theirs to make. `<<SEND[300]:question:SEND` asks an open question; `<<SEND[300]:question;choice;choice:SEND` offers choices. Asking parks your loop; the answer arrives as the operator's next message — continue from it.
+Enabled here: you may ask the operator when a decision is genuinely theirs to
+make. `## SEND0 [300]` with body `question` asks an open question; semicolon-separated
+choices in the body offer choices. Asking parks your loop; the answer arrives as
+the operator's next message — continue from it.
 
 Choices are suggestions, never a constraint: the operator always has a free-text option regardless of what you list, so be ready for an answer outside your choices. Ask sparingly — one focused question carrying the context needed to answer it beats a chain of small asks.

package/docs/worker.md

@@ -3,31 +3,40 @@
 Workers inhabit one workspace. `worker://<name>` addresses a named worker; `worker://~` is the
 current-worker control sigil; `worker://~/path` addresses your private entries; `worker:///path`
 addresses the shared commons.
-`WORK(worker://<name>):task` spawns a fresh worker with an empty log. `FORK(worker://<name>):task`
-branches your current history. `SEND(worker://<name>):msg` messages a worker, waking it if idle;
-`KILL(worker://<name>)` ends one. Workers share project files and the commons, while private entries
+`## WORK0 (worker://<name>)` with a task body spawns a fresh worker with an empty log.
+`## FORK0 (worker://<name>)` branches your current history. `## SEND0 (worker://<name>)`
+with a message body wakes and messages a worker; `## KILL0 (worker://<name>)` ends one.
+Workers share project files and the commons, while private entries
 and conversation logs remain owner-scoped. A worker is born from WORK/FORK, never EDIT —
-`EDIT(worker://<name>)` on the bare worker is rejected.
+`## EDIT0 (worker://<name>)` on the bare worker is rejected.
 
 **The path is the discriminator.** `worker://<name>` with no path addresses a literal worker name
 for WORK, FORK, SEND, READ, or KILL; `worker://~` addresses the caller for SEND or KILL.
 The control form is exact: a trailing slash, userinfo, port, query, fragment, or request metadata
 is invalid rather than ignored.
-`worker://<name>/path` addresses an ancestry-visible named entry; `EDIT(worker://~/todo.md):…`
-writes your own private entry.
+`worker://<name>/path` addresses an ancestry-visible named entry; `## EDIT0 (worker://~/todo.md)`
+with a body writes your own private entry.
 
 **WORK to delegate, FORK to branch.** For fan-out, WORK a distinct-named worker per job. Each gets
 a fresh task. FORK only to carry *your own* context down an alternate path.
 
-**Loop: spawn once → park → collect on wake.** Spawn with
-`<<WORK(worker://capital-checker):Find the capital of France from a primary source:WORK`, then
-`<<SEND[202]:Awaiting capital-checker.:SEND` parks you. You wake when the worker concludes: its
+**Loop: spawn once → park → collect on wake.** Spawn and park with:
+
+```plurnk
+## WORK0 (worker://capital-checker)
+Find the capital of France from a primary source
+
+## SEND0 [202]
+Awaiting capital-checker.
+```
+
+You wake when the worker concludes: its
 result arrives open in your log as a `SEND` from `worker://capital-checker` — read it and continue.
-Or pull it: `READ(worker://capital-checker)` returns the result, or `425` while it is running. Spawn
+Or pull it with `## READ0 (worker://capital-checker)`; it returns the result, or `425` while running. Spawn
 each worker exactly once. Fan-out uses distinct names, followed by one park. Each conclusion wakes
 you with its delta.
 
-**Concluding with live workers.** `SEND[200]` is refused (`409`) while you hold a live worker or
+**Concluding with live workers.** `## SEND0 [200]` is refused (`409`) while you hold a live worker or
 open stream. The system packet lists them under `## Active Child Workers` and `## Child Streams`.
-Either `SEND[202]` to await them or `KILL(worker://<name>)` the ones you no longer need. A same-turn
-KILL followed by `SEND[200]` concludes cleanly.
+Either `## SEND0 [202]` to await them or `## KILL0 (worker://<name>)` the ones you no longer need.
+A same-turn KILL followed by `## SEND0 [200]` concludes cleanly.

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@plurnk/plurnk-meta",
-    "version": "1.5.0",
+    "version": "1.6.0",
     "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.",
     "license": "MIT",
     "type": "module",

package/requirements.md

@@ -1,2 +1,3 @@
-YOU MUST always begin each turn with a PLAN.
-YOU MUST always end retrieval turns (with READ, FIND, ...) with a non-concluding status code.
+YOU MUST always begin each turn with `# PLAN0`.
+YOU MUST always end each turn with `## SEND0 [status code]`; turns with retrievals use a non-concluding code.
+YOU MUST react and respond to the latest prompt listed under Active User Prompts.