Usability pass: tool registry, TUI polish, saved keys and settings

Tools
- Table-driven tool registry with permission classes and default/minimal
  profiles; new tools: edit_file, list_dir, glob, grep, delete_file,
  move_file, todo_write, web_fetch, ask_user (plus read_file paging).
- Built-in Myers diff; write tools return unified diffs.
- Persistent permission rules (.agent/permissions.json), per-turn undo
  snapshots (.agent/undo), AGENTS.md project notes, auto-compact at 80%
  of the model context, cancellable run_shell with process-group kill.

TUI
- Table-driven slash commands: /provider /key /effort /status /tools
  /permissions /yolo /undo /retry /diff /export /copy /rename /delete
  /init /max-steps /todo, with / and @ completion.
- Permission modal with diff and prefix rules, ask_user modal, todo
  panel, header with model/effort/context, bracketed paste, line editing,
  wide-char cursor, NO_COLOR and ASCII fallback; Esc never quits.

Startup and configuration
- API keys are entered in the TUI (hidden) and saved per provider in
  <config>/credentials (0600); env variables are only a fallback.
- Last provider, per-provider model and reasoning effort are remembered
  in <config>/settings.json; --effort/-e and /effort send
  reasoning_effort (low|medium|high|max) to models that accept it.
- DeepSeek preset uses deepseek-flash / deepseek-v4-pro.
- Session flags --new/--continue/--resume/--list-sessions, --yolo,
  --confirm.

Compatibility and tests
- ncurses detection, macOS/Homebrew support, make install, XDG config
  dir, POSIX cleanups; Gitea CI (gcc/clang/asan/musl/macOS).
- Unit tests for diff, fs tools, registry, undo/perms and config;
  e2e fixtures multi_file_refactor and ask_or_todo; vendored cJSON.

scripts/run_matrix.sh

@@ -0,0 +1,63 @@
+#!/usr/bin/env bash
+# Experiment matrix: protocol x model x fixture x N runs, then summarize.
+#
+# Usage:
+#   scripts/run_matrix.sh [-p "native-eos native-inc ctl ctl-heredoc xml"]
+#       [-m "deepseek:deepseek-flash openai:gpt-4.1-mini"]
+#       [-f "mini_calc makefile_tab big_file test_fix_test"]
+#       [-n 3] [-o out/matrix-<stamp>] [-r] [-t minimal|default]
+#
+#   -m  provider:model pairs (model may be empty: "deepseek:")
+#   -r  also record raw SSE streams per run (for replay experiments)
+#   -t  tool profile (default minimal, as in the paper runs)
+#
+# Each run is scripts/e2e_fixture.sh; results land in <out>/<run>/ and the
+# summary in <out>/results.csv and <out>/summary.md.
+set -uo pipefail
+
+ROOT="$(cd "$(dirname "$0")/.." && pwd)"
+PROTOCOLS="native-eos native-inc ctl ctl-heredoc xml"
+MODELS="deepseek:deepseek-flash"
+FIXTURES="mini_calc makefile_tab big_file test_fix_test"
+N=3
+OUT="$ROOT/out/matrix-$(date -u +%Y%m%dT%H%M%SZ)"
+RECORD=()
+TOOLS=minimal
+
+while getopts "p:m:f:n:o:rt:h" opt; do
+  case "$opt" in
+    p) PROTOCOLS="$OPTARG" ;;
+    m) MODELS="$OPTARG" ;;
+    f) FIXTURES="$OPTARG" ;;
+    n) N="$OPTARG" ;;
+    o) OUT="$OPTARG" ;;
+    r) RECORD=(--record) ;;
+    t) TOOLS="$OPTARG" ;;
+    h|*) sed -n '2,16p' "$0"; exit 0 ;;
+  esac
+done
+
+mkdir -p "$OUT"
+if [[ -z "${CST_AGENT_BIN:-}" ]]; then
+  make -s -C "$ROOT" cst-agent >/dev/null || { echo "build failed" >&2; exit 65; }
+fi
+
+total=0
+for pm in $MODELS; do
+  for proto in $PROTOCOLS; do
+    for fx in $FIXTURES; do
+      for ((i = 1; i <= N; i++)); do
+        total=$((total + 1))
+        provider="${pm%%:*}"
+        model="${pm#*:}"
+        margs=(--provider "$provider")
+        [[ -n "$model" ]] && margs+=(--model "$model")
+        printf '[%d] %-10s %-12s %-14s #%d  ' "$total" "$provider" "$proto" "$fx" "$i"
+        "$ROOT/scripts/e2e_fixture.sh" "$fx" --protocol "$proto" --tools "$TOOLS" "${margs[@]}" \
+          --out "$OUT" "${RECORD[@]}" | sed 's/ -> .*//'
+      done
+    done
+  done
+done
+
+python3 "$ROOT/scripts/summarize.py" "$OUT"