One binary: the lint and the trainer are commands of portal

portal serve runs a site; portal lint is what iris was; portal plan,
build, report and the rest are what portal-trainer was. One source at
one version, so the three never disagree about what content means.
Both histories are merged in (subtree). The release attaches the bare
binary for a content repo's CI. The trainer's plans, intakes, engines
and reference prompt come along; its runs stay in the old repo.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GT37Z1Xtfd9pUuQtMTg6Yt

engines/claude.sh

@@ -0,0 +1,22 @@
+#!/bin/sh
+# LLM backend for portal-trainer on Claude Code's own login (a Max or
+# Pro subscription): prompt on stdin, text on stdout. No API key
+# needed: an ANTHROPIC_API_KEY in the environment would outrank the
+# login, so it is dropped here. Log the CLI in once with
+# `claude auth login`; `claude auth status` shows what it uses.
+# MODEL is an alias or id (sonnet, opus, claude-sonnet-5); EFFORT is
+# low, medium or high. Not --bare: that mode skips the stored login.
+#
+# The generator's SEED and TEMPERATURE hints cannot be honoured;
+# FORMAT=json is turned into an instruction.
+set -eu
+unset ANTHROPIC_API_KEY ANTHROPIC_AUTH_TOKEN CLAUDECODE CLAUDE_CODE_ENTRYPOINT
+prompt=$(cat)
+if [ "${FORMAT:-}" = json ]; then
+  prompt="$prompt
+
+Reply with the JSON object only, no prose, no code fences."
+fi
+printf '%s' "$prompt" | claude -p --model "${MODEL:-sonnet}" --effort "${EFFORT:-low}" \
+  --output-format text --permission-mode dontAsk --no-session-persistence \
+  --tools ""