Task: Write docs/loopwright/principles.md #11

Goal

Write docs/loopwright/principles.md: the descriptive statement of loopwright's
core opinions and its detail territory, with the classification rule and the
amendment mechanism. This is the artifact the whole RFC exists to produce, and
the three downstream RFCs that already cite it depend on it existing.

The doc must match the conventions of its two siblings in docs/loopwright/:
single # title then ## sections, sentence-case short section names, tables
as the default format for anything enumerable, hard wrap at ~78 columns, no
front matter, no date, no emoji, impersonal declarative voice, every path and
tool name backticked.

Acceptance criteria

• docs/loopwright/principles.md exists and opens with the Context that
  states near-autonomy as the destination and the core opinions as its
  preconditions, not its constraints.
• P1–P12 are present as a list with stable ids, and each carries three
  things: the assertion, a rejects: line naming one concrete change it
  refuses, and an enforcement pointer that is a file name with no line
  number (left empty where nothing enforces it).
• T1–T8 are present, each with its territory and its current default. Six
  carry a named seam; T4 and T6 carry none — the RFC omits them
  deliberately, so inventing one is a defect, not a completion. No contract is
  specified for T1, T2, T3, T7 or T8 — the seam is named, the shape is its own
  RFC's — with T5 the one exception, whose four required capabilities are
  enumerated in full because this RFC resolves them.
• The classification rule and an Amendments section are present; the
  amendment rule states that ids are stable, amendments never renumber, and
  a withdrawn opinion stays as P4 (withdrawn, RFC #N).
• P7 names the policy file only. baseline.json is described as measured
  state belonging to P3, never as policy.
• P12's assertion carries the three-valued connector contract — a value for a
  core-owned key, unconfigured, failed — with each state's verdict stated
  (unconfigured warns and never blocks; failed always blocks), not only
  "connectors never invent a key". This is normative content and belongs in
  the doc; see the note on the RFC's rationale prose below.
• Read against RFC: Stack connectors, decouple the quality gate from JS/TS #6, RFC: Isolated agent runtime with a mediated, async channel #7, RFC: One config file, YAML, sectioned by feature #8, RFC: Versioned releases with changesets, and automatic changesets in the loop #9 and RFC (draft): Interactive installer for first-time setup #10, the doc contradicts none of them
  except the RFC: Versioned releases with changesets, and automatic changesets in the loop #9 version-bump contradiction, which the RFC states is
  deliberate.

Pointers into the RFC

• The twelve opinions and the eight territories are tabulated in the RFC's
  Decision section.
• On the RFC's "Why the contested branches settled where they did" section: what
  must not travel into the doc is the history — which alternative was
  rejected, what the original RFC said before the grill, which RFC a decision
  contradicts. What must travel is any normative content that happens to
  live only in that prose, because the Decision tables are summaries: the
  three-valued connector contract for P12, the P2/P8 division of labour, and the
  fact that baseline.json is measured state and not the policy file P7 names.
  The test is: a rule a reader must obey belongs in the doc; the story of how it
  was decided belongs in this issue.
• The four agent-host capabilities (T5) and the T6/T7 relationship are spelled
  out in the RFC's Detail territory section.
• The architecture comment on issue RFC: Core principles, opinionated about process, agnostic about stack #5 has the four boundary diagrams; the seams
  it draws are core ↔ connector, engine ↔ host, loop ↔ agent host and
  host ↔ harness. The doc names seams but draws nothing.
• docs/loopwright/loop-harness.md and docs/loopwright/quality-gate.md are the
  style reference.

Blocked by

Nothing.