Seth/minecraft-ai-god-paper-fork
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix sudo follow-up status checks and intent repair gaps

Browse Source
This commit is contained in:
Claude Code
2026-03-17 20:26:29 -04:00
parent 40b4da345a
commit 30aa8388e3
2 changed files with 110 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
SESSION.md
+3
View File
@@ -131,6 +131,9 @@ This section captures decisions and context accumulated across conversations wit
- **Lookup and benevolent TP context refinements (2026-03-17):** lookup mode now has local fallback answering for contextual gameplay questions (e.g., invisibility vs mobs) when retrieval returns no hits, and prompts/gateway guidance now explicitly discourage teleport use in helpful responses unless movement is explicitly requested.
- **Post-retest fixes (2026-03-17):** added execute-tail syntax repair so command fixers apply inside `execute ... run ...` payloads (fixes old bow NBT + fill fire variants emitted under execute wrappers), and added TNT quantity expansion from prompt count for summon-heavy intents (bounded by `sudo_tnt_max_commands`, default 80) when model output under-produces summons.
- **Retest outcome (2026-03-17 late):** bow repair now works in live runs (`give strongest bow` successfully converted old NBT to `minecraft:bow[enchantments={...}]` and delivered item). Remaining issues: fire-spread requests still often execute as no-op/invalid hybrid fill chains (`execute ... run fill ... fire ...` with mixed legacy args), and TNT intent can still collapse into single-command failure then destructive fallback (large `fill ... air` + few TNT) instead of honoring requested count semantics.
- **Late-night retest results (2026-03-18):** strongest-bow path is confirmed fixed; lookup follow-up (`did that command do what I asked?`) still intermittently returns no result; fire-spread prompt still no-ops/invalid despite syntax repair; TNT quantity prompts (`spawn 80/20 tnt`) still execute a single summon in some runs, indicating quantity expansion is not consistently applied before execution.
- **Follow-up fixes after late-night retest (2026-03-18):** added per-player `last_sudo_feedback` memory and deterministic lookup fallback answer for `did that command do what I asked?`; expanded execute-tail repair coverage and fire fill metadata repair (`... fire 0 replace <block>`); disabled destructive air-fill fallback for TNT intents; added fire-intent fallback retry and robust TNT expansion by parsing summon tail from nested execute wrappers.
- **Project governance for bug intake (2026-03-17):** `bug_log` reports from all players are useful advisory input, but direction/prioritization authority remains with project owner (`slingshooter08`); non-owner reports should be weighted accordingly.
- **God voice update (2026-03-17):** Increased default God persona emphasis on irony, dark humor, and sarcastic one-liners in both command and message system prompts (vanilla + Paper variants) while keeping command strictness unchanged.
- **Bug-log triage (2026-03-17):** `bug_log` entry confirmed an unintended-feeling movement reward in prayer flow (`execute as slingshooter08 run tp slingshooter08 ~ ~10 ~`) during a build-oriented prayer; prioritize pray-path teleport safety guards and intent alignment.
- **Bug follow-up (2026-03-17):** second `bug_log` entry reported God feeling "too nice" after greedy follow-up prayer; prompt context updated to bias repeated greedy demands toward corrective responses (rebuke/debuff/symbolic punishment) instead of extra rewards.
mc_aigod_paper.py
+107 -22
View File
@@ -9,6 +9,7 @@ Config: /etc/mc_aigod.json
import json, os, random, re, socket, struct, threading, time, logging
from collections import deque
from datetime import datetime
from typing import Any, Dict
import shutil
from urllib.parse import parse_qs, unquote, urljoin, urlparse
import requests
@@ -80,6 +81,9 @@ sudo_history: deque = deque() # entries: (ts, player, prompt, translated_
SUDO_FAILURE_SIZE = 20
sudo_failures: deque = deque() # entries: (ts, player, command, error)
# Last sudo execution feedback by player for follow-up questions.
last_sudo_feedback: Dict[str, Dict[str, Any]] = {}
_memory_lock = threading.Lock()
# Gateway client session mapping (player+mode -> session_id)
@@ -1012,6 +1016,21 @@ def get_sudo_failures_block(player: str = "") -> str:
return "\n=== RECENT FAILED SUDO PATTERNS ===\n" + "\n".join(lines) + "\n"
def set_last_sudo_feedback(player: str, prompt: str, results_seen: list, ineffective: bool):
with _memory_lock:
last_sudo_feedback[player] = {
"ts": time.time(),
"prompt": (prompt or "")[:240],
"ineffective": bool(ineffective),
"results": [(c[:220], (r or "")[:240]) for c, r in results_seen[:12]],
}
def get_last_sudo_feedback(player: str) -> Dict[str, Any]:
with _memory_lock:
return dict(last_sudo_feedback.get(player, {}))
def _bug_log_path(config) -> str:
return config.get("bug_log_path", "/var/log/mc_aigod_paper_bug.log")
@@ -2306,16 +2325,36 @@ def fix_weather_command(cmd: str) -> str:
def fix_fill_fire_command(cmd: str) -> str:
"""Fix legacy fill syntax like `... fire 0 replace air` for 1.21."""
raw = (cmd or "").strip()
m = re.match(r'^(fill\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+)(minecraft:)?(fire|soul_fire)\s+0\s+replace\s+air$', raw, flags=re.IGNORECASE)
m = re.match(
r'^(fill\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+)(minecraft:)?(fire|soul_fire)\s+0\s+replace\s+(\S+)$',
raw,
flags=re.IGNORECASE,
)
if not m:
return raw
prefix, ns, block = m.groups()
prefix, ns, block, repl = m.groups()
block_id = f"minecraft:{block.lower()}" if not ns else f"{ns.lower()}{block.lower()}"
fixed = f"{prefix}{block_id} replace air"
fixed = f"{prefix}{block_id} replace {repl.lower()}"
log.warning(f"Fixed fill fire syntax: '{cmd}' -> '{fixed}'")
return fixed
def _split_execute_tail(cmd: str):
"""Return (wrappers, tail) for execute chains."""
tail = (cmd or "").strip()
wrappers = []
for _ in range(6):
if not tail.startswith("execute "):
break
marker = " run "
idx = tail.find(marker)
if idx < 0:
break
wrappers.append(tail[: idx + len(marker)])
tail = tail[idx + len(marker):].strip()
return wrappers, tail
def fix_bow_enchant_syntax(cmd: str) -> str:
"""Rewrite old bow Enchantments NBT to 1.21 component format."""
raw = (cmd or "").strip()
@@ -2418,6 +2457,11 @@ def _is_destructive_intent(prompt: str) -> bool:
return any(k in p for k in keys)
def _is_fire_intent(prompt: str) -> bool:
p = (prompt or "").lower()
return any(k in p for k in ("fire", "ignite", "burn", "flame"))
def _normalize_sudo_command_shape(cmd: str, player: str) -> str:
c = (cmd or "").strip()
if not c:
@@ -2565,6 +2609,20 @@ def _repair_failed_sudo_commands(player: str, results_seen: list, config) -> lis
if len(out) >= max_retry:
return out[:max_retry]
# Fire fill repair: remove legacy metadata and simplify execution anchor.
if "incorrect argument" in r and "fill" in c and "fire" in c:
w, tail = _split_execute_tail(c)
tail = fix_fill_fire_command(tail)
out.append(f"execute at {player} run {tail}")
if len(out) >= max_retry:
return out[:max_retry]
# Empty result for large fire fill: retry with tighter vertical band.
if (not r.strip()) and ("fill" in c and "fire" in c):
out.append(f"execute at {player} run fill ~-25 ~-1 ~-25 ~25 ~3 ~25 minecraft:fire replace air")
if len(out) >= max_retry:
return out[:max_retry]
return out[:max_retry]
@@ -2582,22 +2640,23 @@ def _expand_tnt_commands_from_prompt(commands: list, prompt: str, player: str, c
if len(commands) >= target:
return commands
summons = [c for c in commands if "summon" in c and "tnt" in c]
if not summons:
first = None
wrappers = []
x = y = z = None
for c in commands:
w, tail = _split_execute_tail(c)
m = re.match(r'^summon\s+(?:minecraft:)?tnt\s+(\S+)\s+(\S+)\s+(\S+)(?:\s+\{.*\})?$', tail)
if not m:
continue
first = c
wrappers = w
x, y, z = m.groups()
break
if not first:
return commands
base = summons[0]
prefix = ""
body = base
m_pref = re.match(rf'^(execute\s+at\s+{re.escape(player)}\s+run\s+)(.+)$', base)
if m_pref:
prefix = m_pref.group(1)
body = m_pref.group(2)
m = re.match(r'^summon\s+(?:minecraft:)?tnt\s+(\S+)\s+(\S+)\s+(\S+)(?:\s+\{.*\})?$', body)
if not m:
return commands
x, y, z = m.groups()
x = x or "~"
y = y or "~1"
z = z or "~"
expanded = []
for i in range(target):
@@ -2609,7 +2668,9 @@ def _expand_tnt_commands_from_prompt(commands: list, prompt: str, player: str, c
xx = "~" if dx == 0 else f"~{dx}"
if z.startswith("~"):
zz = "~" if dz == 0 else f"~{dz}"
expanded.append(f"{prefix}summon minecraft:tnt {xx} {y} {zz}")
tail = f"summon minecraft:tnt {xx} {y} {zz}"
cmd = "".join(wrappers) + tail if wrappers else tail
expanded.append(cmd)
log.warning(f"Expanded TNT commands from {len(commands)} to {len(expanded)} (requested={requested}, cap={cap})")
return expanded
@@ -2865,9 +2926,15 @@ def process_sudo(player, prompt, config):
return True
return False
def _local_lookup_fallback_answer(query: str, ref_cmd: str) -> str:
def _local_lookup_fallback_answer(query: str, ref_cmd: str, last_feedback: Dict[str, Any]) -> str:
q = (query or "").lower()
rc = (ref_cmd or "").lower()
if re.search(r'\bdid that command do what i asked\b', q):
if not last_feedback:
return "I do not have enough recent execution context to verify that yet."
if bool(last_feedback.get("ineffective", False)):
return "Likely no. Recent execution results indicate the command was ineffective or partially failed."
return "Likely yes. Recent execution results indicate the command completed successfully."
if "invisible" in q and "mob" in q and "invisibility" in rc:
return "Invisibility greatly reduces mob detection, but it does not make you perfectly undetectable at close range or while making noise/actions."
return ""
@@ -2897,6 +2964,7 @@ def process_sudo(player, prompt, config):
if last_cmd:
_send_private(player, f"context command: {last_cmd}", config, "dark_gray")
last_fb = get_last_sudo_feedback(player)
wiki_rows = _info_lookup_wiki(lookup_query)
web_rows = _info_lookup_web(lookup_query)
gateway_msg = ""
@@ -2947,7 +3015,7 @@ def process_sudo(player, prompt, config):
_send_private(player, f"- {tool}: {q}", config, "dark_gray")
if not wiki_rows and not web_rows and not gateway_msg:
fb = _local_lookup_fallback_answer(lookup_query, last_cmd)
fb = _local_lookup_fallback_answer(lookup_query, last_cmd, last_fb)
if fb:
_send_private(player, f"- {fb}", config, "gray")
else:
@@ -3193,7 +3261,7 @@ def process_sudo(player, prompt, config):
ineffective = (len(executed) == 0) or (effective_hits == 0)
# Adaptive fallback for destructive intent when output appears ineffective.
if _is_destructive_intent(prompt) and ineffective:
if _is_destructive_intent(prompt) and ineffective and "tnt" not in prompt.lower():
fallback_cmds = _build_destructive_fallback(player, config)
log.warning(f"SUDO destructive fallback engaged for prompt={prompt!r}: {fallback_cmds}")
_send_private(player, "[SUDO] Initial plan was weak; applying destructive fallback.", config, "yellow")
@@ -3211,10 +3279,27 @@ def process_sudo(player, prompt, config):
effective_hits = sum(1 for _, res in results_seen if _sudo_result_is_effective(res))
ineffective = (len(executed) == 0) or (effective_hits == 0)
if _is_fire_intent(prompt) and ineffective:
fire_retry = [f"execute at {player} run fill ~-25 ~-1 ~-25 ~25 ~3 ~25 minecraft:fire replace air"]
log.warning(f"SUDO fire fallback engaged for prompt={prompt!r}: {fire_retry}")
for cmd in fire_retry:
resolved, is_safe = validate_command(cmd, online, player, config)
if not is_safe:
continue
log.info(f"SUDO fire fallback execute: {resolved}")
result = rcon(resolved, config["rcon_host"], config["rcon_port"], config["rcon_password"])
log.info(f"SUDO fire fallback result: {result!r}")
executed.append(resolved)
results_seen.append((resolved, str(result or "")))
time.sleep(0.15)
effective_hits = sum(1 for _, res in results_seen if _sudo_result_is_effective(res))
ineffective = (len(executed) == 0) or (effective_hits == 0)
for cmd, res in results_seen:
if not _sudo_result_is_effective(res):
add_sudo_failure(player, cmd, res)
set_last_sudo_feedback(player, prompt, results_seen, ineffective)
_report_sudo_feedback(player, prompt, commands, results_seen, ineffective, config)
add_sudo_history(player, prompt, commands, executed)