…rryover

L1 (lws+json container + conneg) is complete & merge-ready: la3d/lws-container,
8 commits over la3d/main (0.0.210), PR LA3D/JavaScriptSolidServer#1, 993/993 green.
Updates the START-HERE block: L1 done; merge-model note (la3d/main stays pristine,
use a la3d/lws integration branch); L2 = linkset + Storage Description (track spec
PRs #183/#180); deferred carryover (--no-lws, HEAD parity, pagination, 304/ETag,
unit-test gaps); then L3 (SHACL admission) and L4 (OKF projection rewritten).

Co-Authored-By: Claude Opus 4.8 <[email protected]>

…rce-error content

Review fixes #1,JavaScriptSolidServer#7,JavaScriptSolidServer#9,JavaScriptSolidServer#12:
- #1 put_typed_resource: declaring the describedby shape is now transactional —
  snapshot the target .meta, MERGE the shape in (don't clobber other keys),
  then roll the .meta back if admission rejects. A rejected write leaves no
  durable side effect and no dangling shape pointer.
- JavaScriptSolidServer#7 call_remote_pod: deep-strip hidden/bidi chars from the federated
  remote_result (least-trusted source) before it enters the model's context.
- JavaScriptSolidServer#9 ResourceError now carries the same isError + content[] teaching shape as a
  tool error, so resources/read failures surface model-readable content, not a
  bare message.
- JavaScriptSolidServer#12 errors.structuredError reuses the toolText envelope.
- Fix a stale mcp.test.js assertion that expected a raw (pre-envelope) body.

Co-Authored-By: Claude Opus 4.8 <[email protected]>

Correctness/security: transactional put_typed_resource .meta (#1),
type/describedby + federated-content sanitize (#2/JavaScriptSolidServer#7), container ACL-path
resolution (JavaScriptSolidServer#3), malformed-percent invalid-params (JavaScriptSolidServer#4), bounded read +
truncation signal (JavaScriptSolidServer#5/JavaScriptSolidServer#6), skill read-error not masked (JavaScriptSolidServer#8), resource-error
teaching content (JavaScriptSolidServer#9). Maintainability: single surface registry (JavaScriptSolidServer#11),
helper reuse (JavaScriptSolidServer#12), restored live-gate coverage (JavaScriptSolidServer#10, lws-pod repo).

All fork changes TDD'd in test/mcp-v2-review-fixes.test.js; mcp/lws/wac/acl/
storage/container/handler suites green per-file.

…ce gate — no 200 where GET teaches 406

- negotiateHeadFileContentType's legacy JSON-LD branch (~line 1190) was
  still gated on bare isRdfContentType(storedContentType), which includes
  plain application/json — under --lws a stored .json HEAD with
  Accept: text/turtle or Accept: application/ld+json returned a lying 200
  where GET correctly 406-teaches (reviewer-reproduced, final whole-round
  review Important #1)
- fix: mirror GET's ternary (lwsEnabled ? isRdfSourceType : isRdfContentType)
  so under --lws plain JSON falls past the legacy branch to the F3 HEAD
  406-teaching mirror; stored JSON-LD → JSON-LD target still passes;
  --lws-off stays byte-identical (ternary's else arm is the old predicate)
- test/lws-nonrdf-teaching.test.js: two new HEAD/.json cases (Accept:
  text/turtle and Accept: application/ld+json), each asserting HEAD status
  equals the paired GET's 406 and content-type application/problem+json
- verified: targeted 4-file run 33/33 pass; full suite 1437 tests, 1436
  pass, 0 fail, 1 pre-existing skip

Co-Authored-By: Claude Opus 4.8 <[email protected]>

…text-free docs (task-6 review #1)

- applyN3Patch detects whether the target document carries an @context and
  passes the verdict into insertTriple: no @context -> the predicate key is
  the full IRI (compactPredicate skipped). On the JavaScriptSolidServer#7 Turtle-family path
  (stored bytes -> expanded JSON-LD, no @context) the compacted keys
  ("rdf:type", "dc:title") re-parsed as literal scheme-IRIs <rdf:type> —
  silent corruption of every common-vocabulary insert.
- Legacy JSON-LD-stored docs (which DO have an @context resolving those
  prefixes) keep the existing compaction behavior byte-identically.
- deleteTriple needs NO symmetric fix: its match already tries the raw key
  first (key === predicate; expanded docs carry full-IRI keys), verified by
  repro and pinned by the new DELETE test. patch.where is parsed but never
  applied, so no compaction assumption there.
- New tests pin both paths TRIPLE-LEVEL via the n3 Parser on an .nt-stored
  doc (N-Triples has no prefix mechanism, so a Turtle round-trip can't
  launder the corrupt CURIE back to the right IRI and mask the bug).

Co-Authored-By: Claude Opus 4.8 <[email protected]>