Three-tier constraint model, mode-aware eval, boundary examples, playtest tooling

Eval harness:
- Mode-aware scoring: sudo=strict (exact match), pray/god=soft (category match,
  in-character, appropriate intensity)
- New metrics: cmd_category_match, appropriate_intensity, scoring_mode breakdown
- Eval defaults to steel141 (192.168.0.141) — prod GPU reserved for serving

Dataset (213 examples):
- Added 31 boundary/adversarial examples (safety edges, abstention, near-boundary)
- Updated pray example reasoning: character-driven logic, not prescriptive outputs
- Tagged pray examples with scoring_mode=soft

Playtest tooling:
- whitelist.sh: add/remove/list across all 3 servers
- FRIENDS_INVITE.md + Discord version: playtester recruitment docs
- Server addresses and implementation details for both training servers

PLAN.md:
- Three-tier constraint model documented (sudo/pray/god_system)
- Success criteria split by scoring mode
- All session decisions logged

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-03-18 15:57:01 -04:00
parent 38b9a02e45
commit 9d789d2524
8 changed files with 516 additions and 82 deletions
Executable
+53
View File
@@ -0,0 +1,53 @@
#!/bin/bash
# Manage whitelists across all Minecraft servers on CT 644
# Usage: ./whitelist.sh add <username>
# ./whitelist.sh remove <username>
# ./whitelist.sh list
ACTION="$1"
PLAYER="$2"
SERVERS=(
"mc1|25575|REDACTED_RCON"
"shrink-world|25576|REDACTED_RCON"
"paper-ai|25577|REDACTED_RCON"
)
rcon_cmd() {
local port="$1" pass="$2" cmd="$3"
ssh pve112 "pct exec 644 -- python3 -c \"
from mcrcon import MCRcon
with MCRcon('localhost', '$pass', port=$port) as r:
print(r.command('$cmd'))
\"" 2>/dev/null
}
case "$ACTION" in
add)
[ -z "$PLAYER" ] && echo "Usage: $0 add <username>" && exit 1
for s in "${SERVERS[@]}"; do
IFS='|' read -r name port pass <<< "$s"
echo -n "$name: "
rcon_cmd "$port" "$pass" "whitelist add $PLAYER"
done
;;
remove)
[ -z "$PLAYER" ] && echo "Usage: $0 remove <username>" && exit 1
for s in "${SERVERS[@]}"; do
IFS='|' read -r name port pass <<< "$s"
echo -n "$name: "
rcon_cmd "$port" "$pass" "whitelist remove $PLAYER"
done
;;
list)
for s in "${SERVERS[@]}"; do
IFS='|' read -r name port pass <<< "$s"
echo -n "$name: "
rcon_cmd "$port" "$pass" "whitelist list"
done
;;
*)
echo "Usage: $0 {add|remove|list} [username]"
exit 1
;;
esac