@plurnk/plurnk-service 0.21.0 → 0.22.0

package/migrations/0000-00-00.01_schema.sql

@@ -98,7 +98,7 @@ CREATE TABLE IF NOT EXISTS entries (
     pathname   TEXT    NOT NULL,
     params     TEXT                         CHECK (params IS NULL OR json_valid(params)),
     attributes TEXT    NOT NULL DEFAULT '{}' CHECK (json_valid(attributes)),
-    -- SPEC §14.3 — how a file member entered the curated surface. 'git' rows are
+    -- SPEC §membership — how a file member entered the curated surface. 'git' rows are
     -- reconciled against `git ls-files − ignore` each turn (registered + un-registered
     -- so entries == members); 'client'/'constraint' (model-created, add-glob) are not
     -- git's to reclaim. NULL = not a file member (other schemes don't carry origin).
@@ -116,7 +116,7 @@ CREATE TABLE IF NOT EXISTS entries (
 CREATE UNIQUE INDEX IF NOT EXISTS entries_agent_identity   ON entries (scheme, pathname)             WHERE scope = 'agent';
 CREATE UNIQUE INDEX IF NOT EXISTS entries_session_identity ON entries (session_id, scheme, pathname) WHERE scope = 'session';
 
--- The ONE engine-imposed constraint (SPEC §7.8): 100 MiB char-length cap
+-- The ONE engine-imposed constraint (SPEC §stream-constraints): 100 MiB char-length cap
 -- per channel content body. All other limits are extrinsic.
 CREATE TABLE IF NOT EXISTS entry_channels (
     entry_id INTEGER NOT NULL,
@@ -217,7 +217,7 @@ CREATE TABLE IF NOT EXISTS log_entries (
     sequence        INTEGER NOT NULL           CHECK (sequence >= 1),
     at              TEXT    NOT NULL DEFAULT (strftime('%Y-%m-%dT%H:%M:%fZ', 'now')),
     origin          TEXT    NOT NULL           CHECK (origin IN ('model', 'client', 'plurnk', 'plugin')),
-    -- §14.5 environment-delta cause: a sibling run-id or a scheme ('file');
+    -- §env-delta environment-delta cause: a sibling run-id or a scheme ('file');
     -- NULL = the owning run itself (self), rendered without a run= label.
     source          TEXT,
 
@@ -354,7 +354,7 @@ BEGIN
 END;
 
 -- INIT: subscriptions
--- Subscription registry per SPEC §7.1. Exists ONLY for cancellation
+-- Subscription registry per SPEC §subscriptions. Exists ONLY for cancellation
 -- routing (SEND[499] → lookup → scheme teardown). Closed rows persist
 -- for forensics; partial unique index enforces one active subscription
 -- per (run, entry).
@@ -385,9 +385,9 @@ CREATE INDEX IF NOT EXISTS subscriptions_scheme_active
 CREATE INDEX IF NOT EXISTS subscriptions_opened_at ON subscriptions (opened_at);
 
 -- INIT: run_watermarks
--- §14.5 environment-delta detection. Per (run, entry, channel): the content this
+-- §env-delta environment-delta detection. Per (run, entry, channel): the content this
 -- run last reconciled. First sight sets it silently (no delta); a later content
--- change materializes a delta-EDIT (the diff span, §14.6) and advances the mark.
+-- change materializes a delta-EDIT (the diff span, §edit-result-render) and advances the mark.
 -- plurnk:// derived entries (manifest/prompt) are excluded at the query, not here.
 CREATE TABLE IF NOT EXISTS run_watermarks (
     run_id   INTEGER NOT NULL,
@@ -400,7 +400,7 @@ CREATE TABLE IF NOT EXISTS run_watermarks (
 ) STRICT, WITHOUT ROWID;
 
 -- INIT: session_constraints
--- SPEC §14.3 constraint overlay — the client's supersede over git membership.
+-- SPEC §membership constraint overlay — the client's supersede over git membership.
 -- Per (session, effect, glob): `add` (members git misses, resolved by a targeted
 -- client-dictated scan), `ignore` (drop git-tracked matches), `read-only` (member
 -- for read; File.edit rejects the write). git-absent, `add` rows are the sole

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@plurnk/plurnk-service",
-  "version": "0.21.0",
+  "version": "0.22.0",
   "description": "Plurnk agent runtime — the server / engine / core. Implements the @plurnk/plurnk-grammar contract.",
   "type": "module",
   "license": "MIT",
@@ -51,8 +51,8 @@
     "@plurnk/plurnk-execs-search": "0.2.3",
     "@plurnk/plurnk-execs-sqlite": "0.1.3",
     "@plurnk/plurnk-execs-wasm": "0.1.2",
-    "@plurnk/plurnk-grammar": "0.32.0",
-    "@plurnk/plurnk-mimetypes": "0.15.8",
+    "@plurnk/plurnk-grammar": "0.33.0",
+    "@plurnk/plurnk-mimetypes": "0.15.9",
     "@plurnk/plurnk-mimetypes-embeddings": "0.1.0",
     "@plurnk/plurnk-providers": "0.2.2",
     "@plurnk/plurnk-schemes": "0.6.0",
@@ -66,21 +66,21 @@
     "@plurnk/plurnk-mimetypes-text-clojure": "0.2.2",
     "@plurnk/plurnk-mimetypes-text-cmake": "0.2.2",
     "@plurnk/plurnk-mimetypes-text-common-lisp": "0.2.2",
-    "@plurnk/plurnk-mimetypes-text-csharp": "0.2.2",
+    "@plurnk/plurnk-mimetypes-text-csharp": "0.2.3",
     "@plurnk/plurnk-mimetypes-text-datalog": "0.3.2",
     "@plurnk/plurnk-mimetypes-text-erlang": "0.2.2",
     "@plurnk/plurnk-mimetypes-text-graphql": "0.2.2",
-    "@plurnk/plurnk-mimetypes-text-mariadb": "0.3.2",
-    "@plurnk/plurnk-mimetypes-text-plsql": "0.2.2",
-    "@plurnk/plurnk-mimetypes-text-postgresql": "0.3.2",
+    "@plurnk/plurnk-mimetypes-text-mariadb": "0.3.3",
+    "@plurnk/plurnk-mimetypes-text-plsql": "0.2.3",
+    "@plurnk/plurnk-mimetypes-text-postgresql": "0.3.3",
     "@plurnk/plurnk-mimetypes-text-prolog": "0.2.2",
     "@plurnk/plurnk-mimetypes-text-protobuf": "0.3.2",
     "@plurnk/plurnk-mimetypes-text-redis": "0.3.2",
     "@plurnk/plurnk-mimetypes-text-sparql": "0.2.2",
-    "@plurnk/plurnk-mimetypes-text-sqlite": "0.3.2",
+    "@plurnk/plurnk-mimetypes-text-sqlite": "0.3.3",
     "@plurnk/plurnk-mimetypes-text-swift": "0.1.2",
-    "@plurnk/plurnk-mimetypes-text-terraform": "1.0.2",
-    "@plurnk/plurnk-mimetypes-text-tsql": "0.2.2",
+    "@plurnk/plurnk-mimetypes-text-terraform": "1.0.3",
+    "@plurnk/plurnk-mimetypes-text-tsql": "0.2.3",
     "@tree-sitter-grammars/tree-sitter-kotlin": "^1.1.0",
     "@tree-sitter-grammars/tree-sitter-lua": "^0.4.1",
     "@tree-sitter-grammars/tree-sitter-toml": "^0.7.0",

package/requirements.md

@@ -3,7 +3,7 @@ Syntax: <<OPsuffix[signal]?(target)?<Line/Result>?:body?:OPsuffix
 YOU MUST document all relevant questions and uncertainties into taxonomized, tagged, and topical unknown entries.
 YOU MUST ONLY populate known entries with source entry information, never with model training.
 YOU SHOULD prefer deterministic calculations and retrievals over model training or speculation.
-YOU SHOULD FOLD log entries that are irrelevant, resolved, or already distilled into known entries.
+YOU SHOULD manage your own context to maximize signal, as irrelevant tokens degrade reasoning.
 YOU SHOULD leverage folksonomic path names, tags, and bulk pattern operations to optimize for context relevance.
 YOU MUST use OPEN and FOLD to keep your context budget healthy, optimized, and topical, below the `tokensFree` limit.
 YOU MUST terminate the turn with either a continuing (102) or loop terminating (200) `<<SEND[102]:status update:SEND`