feat(instructions): fit the delivery fold — 2k server map, floor Systems reflex, write-time systems_hint

Claude Code injects only the first ~2,048 chars of an MCP server's
instructions and silently cuts the rest mid-word (#2562, observed live):
_INSTRUCTIONS was 20,002 chars, so ~90% — including all Systems tagging
guidance — never reached any session. Rearchitect delivery around what
each surface actually delivers:

- _INSTRUCTIONS becomes a 1,997-char purpose-sorted map, with a header
  comment stating the budget and where detail belongs instead.
- Tool docstrings keep the per-tool HOW (audit: nearly all displaced
  topics were already duplicated there); backfill the four gaps —
  enter_project session scoping + project bootstrap, create_rule
  entity-vs-rule test, create_design_system not-a-rulebook,
  create_system two-records test.
- The plugin static context (the delivery floor) gains the
  tag-to-Systems reflex and a surfaces-layering statement; plugin
  0.1.25 -> 0.1.26 so the executing cache refreshes (#2209).
- create_task / create_note / create_snippet return a systems_hint when
  a record is created untagged in a project that has Systems — in-band
  at the exact write it applies to, fail-open like the dedup gate.
- Guards: _INSTRUCTIONS length budget, floor-states-the-reflex, and a
  displaced-topics sweep asserting every cut topic still lives on a
  delivered surface.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

tests/test_instruction_surfaces_agree.py

@@ -85,6 +85,95 @@ def test_every_session_start_surface_states_the_pull():
     )
 
 
+def _instructions_text() -> str:
+    """The _INSTRUCTIONS literal from server.py, as the client would see it."""
+    import re
+    src = (ROOT / "src" / "scribe" / "mcp" / "server.py").read_text()
+    match = re.search(r'_INSTRUCTIONS = """(.*?)"""', src, re.S)
+    assert match, "server.py no longer defines _INSTRUCTIONS as a triple-quoted literal"
+    return match.group(1).strip()
+
+
+# Claude Code injects only the first ~2,048 characters of an MCP server's
+# instructions and silently cuts the rest mid-word (#2562: observed live —
+# the previous 20k-char version delivered ~10% of itself, and none of the
+# Systems tagging guidance ever reached a session). 2,000 leaves margin.
+INSTRUCTIONS_BUDGET = 2000
+
+
+def test_instructions_fit_the_fold():
+    text = _instructions_text()
+    assert len(text) <= INSTRUCTIONS_BUDGET, (
+        f"_INSTRUCTIONS is {len(text)} chars; the client injects only ~2,048 "
+        f"and silently cuts the rest (#2562). This block is a MAP — move the "
+        f"detail to the tool's docstring (delivered at reach-for time), the "
+        f"plugin static context (always delivered), or a skill; see the "
+        f"comment above _INSTRUCTIONS."
+    )
+
+
+def test_floor_states_the_systems_reflex():
+    """Write-time tagging guidance must live on the surface that always arrives.
+
+    #2562's behavioral finding: with the guidance only in tool descriptions,
+    sessions filed records untagged. The static context is the delivery floor,
+    so the tag-as-you-write reflex has to be stated there.
+    """
+    floor = (ROOT / "plugin" / "hooks" / "scribe_static_context.md").read_text()
+    for needle in ("system_ids", "create_system"):
+        assert needle in floor, (
+            f"plugin/hooks/scribe_static_context.md no longer mentions "
+            f"{needle} — the Systems tagging reflex must be stated on the "
+            f"floor, not only in tool descriptions (#2562)."
+        )
+
+
+# Topics displaced from _INSTRUCTIONS when it was cut to fit the fold. Each
+# must remain stated on at least one DELIVERED surface: a tool docstring
+# (arrives with the tool schema), the plugin static context (always arrives),
+# or a bundled skill (arrives on trigger match). Keyed by a phrase distinctive
+# enough that its disappearance means the guidance is gone, not reworded —
+# update the phrase alongside a deliberate rewording.
+DISPLACED_TOPICS = {
+    "supersedes": "supersedes",
+    "trash is recoverable": "deleted_batch_id",
+    "duplicate gate": "duplicate",
+    "systems tag-as-you-write": "system_ids",
+    "reference note vs dev-log": "reference note",
+    "work-logs over body rewrites": "add_task_log",
+    "rule homes / altitude": "create_project_rule",
+    "rules vs other entities": "standing instruction",
+    "shared records are suggestions": "shared",
+    "processes run verbatim": "verbatim",
+    "snippet reuse reflex": "when_to_use",
+    "compaction at seams": "compact",
+    "plans are milestones": "start_planning",
+    "scope to the entered project": "cross-project",
+    "project bootstrap needs confirmation": "never guessing a project",
+}
+
+
+def test_displaced_topics_live_on_a_delivered_surface():
+    corpus = ""
+    for p in (ROOT / "src" / "scribe" / "mcp" / "tools").glob("*.py"):
+        corpus += p.read_text()
+    corpus += (ROOT / "plugin" / "hooks" / "scribe_static_context.md").read_text()
+    for p in (ROOT / "plugin" / "skills").rglob("SKILL.md"):
+        corpus += p.read_text()
+    corpus = corpus.lower()
+    missing = [
+        f"{topic} (phrase: {phrase!r})"
+        for topic, phrase in DISPLACED_TOPICS.items()
+        if phrase.lower() not in corpus
+    ]
+    assert not missing, (
+        f"guidance displaced from _INSTRUCTIONS has fallen off every delivered "
+        f"surface (tool docstrings / static context / skills): {missing}. It "
+        f"was cut from _INSTRUCTIONS deliberately (#2562) on the premise it "
+        f"lives elsewhere — restore it somewhere that delivers."
+    )
+
+
 def test_no_surface_names_the_push_without_stating_the_pull():
     """The exact shape #2497 took.