This is a public, read-only report. Striff reads this repository's docs, turns each sentence that makes a claim about the code into a rule, and checks the rule against the code on the default branch. How this works
The system shall not expose a separate async entrypoint (e.g. acall) on ToolBot;
ToolBot may not have acall
Holds
Read Oct 11
AsyncToolBot uses native acompletion and async_stream_chunks—no asyncio.to_thread around the sync ToolBot.__call__.
AsyncToolBot may not use asyncio.to_thread
Holds
Read Oct 11
Ensure ChatMemory is top-level in llamabot/__init__.py ✅
ChatMemory is in llamabot
Holds
Read Oct 11
add_message() → append()
ChatMemory has an append method
Holds
Read Oct 11
get_messages() → retrieve()
ChatMemory has a retrieve method
Holds
Read Oct 11
New threading support with ChatMemory.threaded()
ChatMemory has a threaded method
Holds
Read Oct 11
save(), load(), export()
3 rules from this sentence
ChatMemory has a save method
checks “save()”
Holds
ChatMemory has a load method
checks “load()”
Holds
ChatMemory has an export method
checks “export()”
Holds
Read Oct 11
2026-03-28 — DecideNode.aexec constructs AsyncToolBot and awaits AsyncToolBot.__call__ (removed ToolBot.acall).
DecideNode depends on AsyncToolBot
Holds
Read Oct 11
Routing is implemented by DecideNode, which uses ToolBot in exec and AsyncToolBot in aexec to map shared["memory"] to a single tool call per decision step.
DecideNode depends on ToolBot
Holds
Read Oct 11
The system shall implement DecideNode.exec such that the routing step uses ToolBot, and DecideNode.aexec such that the routing step uses AsyncToolBot, each with the same tool list and model configuration as the AgentBot instance (subject to Ollama-specific prompt and tool_choice adjustments).
3 rules from this sentence
AgentBot depends on DecideNode
Holds
DecideNode depends on ToolBot
checks “DecideNode.exec such that the routing step uses ToolBot”
Holds
DecideNode depends on AsyncToolBot
checks “DecideNode.aexec such that the routing step uses AsyncToolBot”
Holds
Read Oct 11
async tool selection shall live on AsyncToolBot.__call__ only.
AsyncToolBot has __call__
Holds
Read Oct 11
Class QueryBot, Module llamabot.bot.querybot, Base SimpleBot
QueryBot extends SimpleBot
Holds
Read Oct 11
Class AsyncQueryBot, Module llamabot.bot.querybot, Base QueryBot
AsyncQueryBot extends QueryBot
Holds
Read Oct 11
Field docstore, Type AbstractDocumentStore, Role Primary retrieval corpus required
QueryBot has a docstore field
Holds
Read Oct 11
Field memory, Type Optional[AbstractDocumentStore], Role Optional second retrieval source also receives append of assistant text after each successful call when set
The system shall provide QueryBot with a required docstore of type AbstractDocumentStore and an optional memory of the same abstract type.
QueryBot depends on AbstractDocumentStore
Holds
Read Oct 11
AsyncQueryBot.__call__ shall delegate to QueryBot.__call__ via asyncio.to_thread with the same query and n_results.
AsyncQueryBot depends on QueryBot
Holds
Read Oct 11
SimpleBot is the base completion stack for most LlamaBot bots: LiteLLM (completion / acompletion) behind the shared message model, optional recorder spans, and SQLite logging of turns.
SimpleBot imports litellm
Holds
Read Oct 11
Class AsyncSimpleBot, Module llamabot.bot.simplebot, Base SimpleBot
AsyncSimpleBot extends SimpleBot
Holds
Read Oct 11
SimpleBot.__call__: Creates a Span (child of current span when present), records metadata, calls make_response + stream_chunks, builds AIMessage with content and tool calls, sqlite_log, and appends to memory when configured (last processed user message + assistant message).
3 rules from this sentence
SimpleBot has __call__
checks “SimpleBot.__call__”
Holds
SimpleBot depends on Span
checks “Creates a **Span”
Holds
SimpleBot depends on AIMessage
checks “builds AIMessage”
Holds
Read Oct 11
AsyncSimpleBot.__call__: Uses token streaming via stream_tokens_for_messages and a finalize callback to assemble the final AIMessage, span fields, logging, and memory.
2 rules from this sentence
AsyncSimpleBot has __call__
checks “AsyncSimpleBot.__call__”
Holds
AsyncSimpleBot depends on AIMessage
checks “assemble the final AIMessage”
Holds
Read Oct 11
SimpleBot.__call__ shall append the last processed user message and the assistant AIMessage to memory when memory is configured.
SimpleBot depends on AIMessage
Holds
Read Oct 11
StructuredBot subclasses SimpleBot but does not use SimpleBot.__call__ for its main path.
StructuredBot extends SimpleBot
Holds
Read Oct 11
Class AsyncStructuredBot, Module llamabot.bot.structuredbot, Base StructuredBot
AsyncStructuredBot extends StructuredBot
Holds
Read Oct 11
Class ToolBot, Module llamabot.bot.toolbot, Base SimpleBot
ToolBot extends SimpleBot
Holds
Read Oct 11
Class AsyncToolBot, Module llamabot.bot.toolbot, Base ToolBot
AsyncToolBot extends ToolBot
Holds
Read Oct 11
AsyncToolBot.__call__: make_async_response + async_stream_chunks, then same extraction and memory updates.
ToolBot.__call__ shall return the list from extract_tool_calls after make_response and stream_chunks, and shall append the first user message and an AIMessage whose content is the string representation of that tool-call list to chat_memory.
2 rules from this sentence
ToolBot depends on AIMessage
checks “ToolBot.__call__ shall return the list from extract_tool_calls after make_response and stream_chunks, and shall append the first user message and an AIMessage”
Holds
ToolBot depends on ChatMemory
Holds
Read Oct 11
AgentBot.close_mcp() or AsyncAgentBot.close_mcp()
2 rules from this sentence
AgentBot has a close_mcp method
Holds
AsyncAgentBot has a close_mcp method
Holds
Read Oct 11
decide_node (DecideNode):
AgentBot has decide_node
Holds
Read Oct 11
tools (List[Callable]):
AgentBot has tools
Holds
Read Oct 11
docstore (AbstractDocumentStore): The document store used for retrieval
QueryBot has docstore
Holds
Read Oct 11
memory (Optional[AbstractDocumentStore]): The chat memory component (if any)
QueryBot has memory
Holds
Read Oct 11
memory (Optional[AbstractDocumentStore], default: None): An optional chat memory component to use. If provided, the bot will retain chat history. For conversational memory, use ChatMemory (e.g., lmb.ChatMemory()). For RAG/document retrieval, use an AbstractDocumentStore (such as BM25DocStore or LanceDBDocStore).
Message lists are built from shared compose_* helpers on the sync base classes (for example SimpleBot.compose_messages_for_human_messages, QueryBot.compose_rag_messages).
2 rules from this sentence
SimpleBot has compose_messages_for_human_messages
Holds
QueryBot has compose_rag_messages
Holds
66 rules from 19 docs
See these where the change is. The browser extension puts a pull request's doc findings, and a diagram of what it changed, on the GitHub page itself, so the code and what your docs say about it are side by side.
