Verity Mod Error Messages: What Each One Means

How to use this page

Copy the message exactly as it appears, then search this page for a distinctive part of it (press Ctrl+F, or Find in page on a phone). The tables quote the part of each message that stays the same; names, numbers, and file paths in the middle change from setup to setup.

Error text from Minecraft, Forge, Groq, Ollama, and LiteLLM can change slightly between versions. If your wording is close but not identical, the meaning is usually the same. If nothing here matches, the troubleshooting guide works forward from the last step that succeeded.

Never paste your API key, .env file, or a full log into a public post. The useful part is usually two or three lines.

Where the real message is

A summary such as “Verity is not working” or “exit code 1” is not the error. Find the underlying line first.

Setup	Where to look	What to search for
Java, game crashed	crash-reports folder inside the instance folder; the newest file	Description: near the top, then the first Caused by:
Java, game running but Verity is silent	logs/latest.log inside the instance folder	ERROR, Exception, groq, ollama, litellm, or 401/429
Java, failed before the title screen	The loading-error screen itself, then logs/latest.log	Mod ID:, Requested by:, Failure message:
Bedrock add-on	Minecraft Settings > Creator: turn on the content log GUI, then reload the world	Red lines naming a script, pack, or entity
PnTMC helper	The terminal window running the helper (CMD, Termux, or a-Shell)	The last lines printed before it stopped
Ollama / LiteLLM	The terminal running ollama serve or the LiteLLM gateway	error, not found, refused, address already in use

In most launchers, Open folder on the instance takes you to the instance folder. In the CurseForge app, use the three-dot menu on the profile and Open Folder.

Java: Verity JE’s own messages

These are the exact texts built into Verity JE 5.7.4. The first two appear in chat; the rest appear in logs/latest.log.

The message	What it means	Fix
API connection failed. Status: 429. Check console for details.	The AI service returned an error code (here 429, rate limit)	Read the code in the Groq messages table below
AI connection error. You might need to replace your API Key.	Shown for any failed or empty reply, not only bad keys	Find the [Verity AI] HTTP Error : or [Verity AI] API Error: line in the log for the real cause
Ollama requires a custom model name. (6.0 beta)	AI Provider is OLLAMA but no model is typed	Turn on Use Custom Model and enter the tag from ollama ls, without ollama/
[Verity AI] Failed to parse AI JSON:	The model answered, but not in the format Verity expects	Go back to the default AI Intelligence Level (or the beta’s default preset); a custom model that ignores the expected format is the usual cause
[Verity TTS] Failed to play voice. Is Groq unreachable?	Cloud voice failed	Text still works; check the internet connection, or turn on the offline voice
[Verity Local TTS] No audio generated (engine may have failed to load).	The built-in offline voice did not start	Restart the game; on low-memory PCs, turn Use Offline AI Voice off
[Verity Native TTS] Failed to use OS Narrator.	The operating system’s narrator option failed	Turn Use Native TTS off, or enable the narrator in your OS
[Verity] Failed to load Sherpa model. Check your model filenames!	The built-in offline speech recognition model failed to load	Turn the built-in STT option off and use cloud speech recognition
[Verity STT] Failed to process speech locally with Sherpa.	Offline speech recognition failed on this recording	Speak again; if it repeats, switch back to cloud speech recognition

Java: Forge loading and crash messages

The message contains	What it means	Fix
Actual version: '[MISSING]'	A library listed under Mod ID is required by the mod named in Requested by, but is not installed	Install the named dependency for Forge 1.20.1 from the file page’s dependency list
Mod ID: 'geckolib' or Mod ID: 'yet_another_config_lib_v3' with [MISSING]	Verity JE needs GeckoLib 4.4+ and YACL 3.6+ (YACL on the client only), plus Forge 47+ on Minecraft 1.20.1	Install both libraries for Forge 1.20.1; these are the requirements written in the jar itself
Expected range: followed by an Actual version that is not [MISSING]	The dependency or Minecraft version is installed, but outside the range the mod accepts	Install the dependency version that falls in the range, or use the Verity file built for your Minecraft version
needs language provider	The jar was built for a newer Forge than the one installed, or for a different loader	Use a Forge 1.20.1 file in a Forge 1.20.1 profile; do not put the NeoForge 1.21.1 file into it
Found duplicate mods	Two files with the same mod ID are in the mods folder, usually an old and a new Verity jar	Delete the older jar and keep one copy
java.lang.OutOfMemoryError	Minecraft ran out of the memory the launcher assigned	Raise the allocation moderately, or close other programs; see memory checks
Exit Code: 1 (launcher window)	The game stopped; the code alone does not say why	Open the newest crash report and read Description: and the first Caused by:

How to read the dependency line. A typical loading-error entry has four parts:

Mod ID: 'examplelib', Requested by: 'examplemod', Expected range: '[1.2.0,)', Actual version: '[MISSING]'

Read it as: examplemod needs examplelib version 1.2.0 or newer, and examplelib is not installed. A square bracket means “including this version”; a round bracket or empty end means “no upper limit”. Install exactly what the Mod ID names, for the same Minecraft version and loader. Do not install every library a video mentions; the file page’s dependency list is the authority for the release you downloaded.

If the error mentions a mod other than Verity, the problem is that mod. Remove everything except Verity and its listed dependencies, launch once, then add other mods back in small groups.

Groq messages

Groq returns a standard HTTP status code and a short JSON message. Verity may show only the code in game, while the log contains the full text. The meanings below follow Groq’s error documentation.

The message contains	What it means	Fix
401 or Invalid API Key	Groq did not accept the key: it is wrong, incomplete, revoked, or has extra spaces	Create a new key, paste it once into the Verity config, and remove leading or trailing spaces. Full steps
403 or permission	The key works, but your organization or project may not use that model	Check model permissions in the Groq console. A new key will not help. Full steps
404 or does not exist or you do not have access to it	The configured model ID is not available to you	Compare the model in the message with Groq’s current model list. Full steps
decommissioned	The model was retired by Groq; old tutorials and old configs often still name it	Update Verity to a release that uses a current model, or change the model setting if your release exposes it
413 or Request too large	This single request is bigger than your per-minute token allowance	Shorten the conversation: start a new session or world, then test with one short message. See below
429 or Rate limit reached	You hit a request or token limit for this time window	Stop sending, wait the time given in the message, then send one short message. See below
500, 502, or 503	A problem on Groq’s side	Wait and retry once; check Groq’s status page. Full steps
timed out or ConnectException	The request never got an answer from Groq	Check the PC’s internet connection, firewall, VPN, or school or work network filtering

Reading a Groq 429 message

A full 429 message names the exact limit you hit. It typically looks like this:

Rate limit reached for model `<model>` in organization `<org>` service tier `on_demand`
on tokens per minute (TPM): Limit 6000, Used 5400, Requested 1100. Please try again in 5s.

• The limit name tells you which allowance is exhausted. Tokens per minute (TPM) and requests per minute (RPM) recover within a minute. Requests per day (RPD) or tokens per day (TPD) will not recover until the daily window resets, so repeated retries only waste time.
• Requested is the size of the message that failed, in tokens. If it is large, your conversation history is large.
• Please try again in is the wait time. Wait that long once instead of pressing send again.

The numbers in the example are illustrative. Your own limits are on the Limits page of the Groq console, and the rate-limit calculator in the Groq guide turns them into messages per minute and per play session.

Why a 413 does not go away by waiting

A 413 says the request is larger than the limit itself, for example Limit 6000, Requested 7200. Waiting does not shrink the request. Verity sends instructions and recent conversation with your message, so a long session can grow past the limit even when you type one word. Starting a fresh conversation (a new session or world, depending on your release) is the reliable fix. Switching to a model with a higher per-minute token limit also works if your project has access to one.

Ollama and LiteLLM messages (local AI)

These appear in the terminal running Ollama or LiteLLM, and sometimes in Verity’s log. Stable Verity JE 5.7.4 talks to LiteLLM on port 4000, and LiteLLM talks to Ollama on port 11434. The Ollama guide shows that chain as a diagram.

The message contains	What it means	Fix
model "…" not found, try pulling it first	Ollama is running but does not have that exact model tag	Run ollama ls. Pull the tag you configured, or change the setting to a listed tag
LLM Provider NOT provided	LiteLLM received a model name without a provider prefix	Use ollama/<tag> in the LiteLLM command and in Verity’s Ollama model field, for example ollama/qwen2.5:1.5b
Connection refused and 4000	Nothing is listening on the LiteLLM port	Start the LiteLLM gateway and leave its terminal open
Connection refused and 11434	Ollama is not running	Start the Ollama app, or run ollama serve in its own terminal
address already in use (or Only one usage of each socket address on Windows)	You started a second copy of a server that is already running	Do not start another; use the one already running, or quit it from the tray icon first
Reply arrives after 30 seconds or more	The model is loading or is too large for your memory	Test with the default small model; see the memory calculator

Bedrock: import and world messages

The message or symptom	What it means	Fix
Duplicate Pack Detected	A pack with the same identity is already installed, usually an older Verity revision	Delete the old packs in Settings > Storage, then import the new file again
Duplicate Pack Detected with no old Verity installed (Utlover65)	Utlover65’s 4.2.3 packs use example-style identifiers (a1b2c3d4-e5f6-7890-abcd-ef1234567890 for the behavior pack), which another add-on built from the same example could share	Check Settings > Storage for a non-Verity pack with the same identity, and keep only one of them in the world
Import Failed	The file is incomplete, renamed, or not a Minecraft pack	Download again from the project page; do not rename a .zip to .mcaddon
Only one pack appears after import	Half of the add-on imported; the behavior and resource packs are separate	Re-import the .mcaddon and confirm both packs are listed
Red content-log lines mentioning scripts	Beta APIs are off, or the add-on does not match this Bedrock build	Enable Beta APIs in the world and match the file’s game version to your Minecraft version
401 or no reply after verity ai key <key> (Utlover65)	Key mode sends requests to OpenAI, so a Groq key is rejected	Use an OpenAI key, or switch back with verity ai worker
No reply after verity ai local, Ollama shows model "llama3" not found (Utlover65)	The add-on asks for llama3, not the llama3.2:1b its guide suggests	ollama pull llama3, or ollama cp llama3.2:1b llama3
Kicked from the world right after chatting (Utlover65)	The profanity filter: one warning, then a kick	Rejoin and avoid swearing in chat near Verity
Verity shows ... (Utlover65)	The add-on loaded but the connection or online service did not answer	Repeat the project’s connection step; see three dots

Bedrock: the PnTMC connection and helper

PnTMC’s add-on needs its local helper running on the same device, and the /connect 127.0.0.1:3000 command in each world session. The helper messages below were checked against the v4.0.3 helper files; the Minecraft chat messages come from that setup, per PnTMC’s setup guide.

Where	The message contains	What it means	Fix
Minecraft chat	Could not connect to server	Minecraft could not reach the helper	Start the helper first and keep it open, then run /connect 127.0.0.1:3000 again
Minecraft chat	Connection established to server	The link to the helper works	Send one typed message to test the AI next
Minecraft chat	Command not allowed or unknown	Cheats are off in this world	Turn on cheats in the world settings
Helper window (Windows)	Python not found.	start_minecraft_stt.bat could not find Python	Install Python 3.10+ and tick Add python.exe to PATH, then run the script again
Windows CMD	'python' is not recognized as an internal or external command	Python is not installed, or was installed without adding it to PATH	Reinstall Python 3.10+ and tick Add python.exe to PATH, then open a new CMD window
Helper terminal	No module named 'websockets' (or dotenv, httpx)	The dependency install step was skipped or failed	Run the guide’s pip install step again in the helper folder
Termux	which termux-microphone-record prints nothing	Termux:API is missing or not from F-Droid	Install both Termux and Termux:API from F-Droid, as the guide requires
Helper terminal	WARNING: set GROQ_API_KEY in .env	The Groq key field in .env is empty	Paste your Groq key after GROQ_API_KEY= and restart the helper
Helper terminal	Fish TTS: OFF — set FISH_API_KEY in .env to enable voice	No FishAudio key, so Verity will reply in text without speech	Optional: add FISH_API_KEY and FISH_REFERENCE_ID for voice
Helper terminal	Empty AI reply	Groq answered but returned no text	Send the message again once; if it repeats, check the model setting in .env
Helper terminal	401 or Invalid API Key	The Groq or FishAudio key in .env is wrong	Fix the key in .env, then copy the folder again as the guide describes and restart

If Verity answers but the replies are short, generic, and silent, the add-on is probably using its built-in keyword replies because the helper is not connected. Check that the helper window is still running and repeat /connect 127.0.0.1:3000.

On iPhone and iPad, a missing microphone is not an error: PnTMC’s a-Shell setup supports typed chat only.

When there is no message at all

Silence usually means the request never left the game, or a filter hid the reply. On stable 5.7.4, check General > Require Verity (messages must include her name) and General > Immersive Mode before changing keys. The not-working guide covers that path, and on Bedrock the five-minute first test separates pack problems from connection problems.

Asking for help with an error

A short, precise report gets an answer far faster than a screenshot of the title screen. Include:

1. Edition, project name, and Project ID.
2. Verity release and Minecraft version (and loader on Java).
3. The two or three lines that contain the error, with keys removed.
4. What you did immediately before it appeared, and whether it happens in a new world.

Send it to the project’s own support channel listed on its project page.