Skip to main content
Pre-release

This release is a pre-release and may not be stable for production use.

everalgo-boundary

Boundary detection for EverAlgo — cuts a conversation into coherent MemCell slices with an LLM, either in one batch call or one message at a time. Agent trajectories with tool calls and tool results go through the same class: only the chat messages reach the model, and the tool items stay in the cells.

See the umbrella project: EverAlgo monorepo and the architecture document at docs/concepts/architecture.md.

Install

pip install everalgo-boundary==0.4.2rc0

everalgo-user-memory and everalgo-agent-memory both depend on this package and re-export BoundaryDetector unchanged.

What this distribution provides

Symbol Role
BoundaryDetector Class binding an LLM client: adetect / detect (batch, over ConversationItem), adetect_step / detect_step (incremental, over ChatMessage)
adetect_boundaries Async function behind BoundaryDetector.adetect: (Sequence[ConversationItem], *, llm, is_final, ...) → DetectionResult
detect_boundaries Sync bridge over adetect_boundaries (asgiref.async_to_sync); never call it inside a running event loop
adetect_boundary_step Async function behind BoundaryDetector.adetect_step: single-step boundary decision → BoundaryDecision
DetectionResult NamedTuple(cells: list[MemCell], tail: list[ConversationItem])

Tool items

adetect_boundaries renders only the ChatMessage items for the model and carries its cuts back to the full list: tool items between two chat messages go with the earlier one, tool items ahead of the first chat message with the first cell, and trailing tool items into the tail (into the last cell when is_final=True). Every item lands exactly once. Without any ChatMessage no LLM is called and every item goes to the tail.

Hard limits

Both paths cut without the model once a window reaches a hard limit: 65536 tokens or 500 messages by default (hard_token_limit / hard_msg_limit on the batch path, hard_token_limit / hard_message_limit on adetect_step).

Prompts

Each model call sends two messages. The system message holds the rules — BATCH_BOUNDARY_DETECT_SYSTEM_PROMPT or STEP_BOUNDARY_DETECT_SYSTEM_PROMPT in everalgo.boundary.prompts.en.chat — and is the same text on every call, so a prompt cache can share all of it. The user message holds only the data, rendered from BATCH_BOUNDARY_DETECT_USER_PROMPT (<messages count="N">) or STEP_BOUNDARY_DETECT_USER_PROMPT (<conversation_history>, <time_gap>, <new_messages>). Each message line is [N] [YYYY-MM-DD HH:MM UTC (Weekday)] speaker: text on the batch path and the same without [N] on the step path. Batch records escape XML delimiters in names and content, indent continuation lines, and retain empty content as [N] [empty message]. Material inside these records is data, not instructions to the detector.

The batch input is an open streaming window of one or more messages. An episode is one coherent event, activity or independent discussion, including its execution, outcome and related feedback. Different steps, revisions, acknowledgements or dates alone do not establish a new episode. With insufficient evidence of a new independent episode, the prompt asks the model to keep the messages together.

Batch labels start at 0. Each boundary b names the first message of a new episode, so the preceding slice is messages[:b]. For example, [2] groups messages 0 and 1, then starts the next episode at 2. The valid range is 1 <= b < N; the final input message can start a new episode. Tool items are not numbered. The strict batch response contains only reasoning and boundaries; reasoning explains all boundary decisions in one sentence. It does not predict whether the trailing episode has finished.

custom_instructions= on adetect_boundaries, adetect_boundary_step and the BoundaryDetector methods adds the host's text to the bundled rules, never replacing them: it is placed verbatim, never rendered as a template, in a <custom_instructions> block that is always the system message's last, and None or a blank string leaves it out. The user message is unchanged and always rendered by the operator. There is no prompt=: the bundled rules cannot be replaced per call.

Quick start

import asyncio
import json

from everalgo.boundary import adetect_boundaries
from everalgo.testing.fake_llm import FakeLLMClient
from everalgo.types import ChatMessage

_BOUNDARY_JSON = json.dumps(
    {"reasoning": "All messages concern the same deployment decision.", "boundaries": []}
)

async def main() -> None:
    # One scripted reply per LLM call; `llm.achat` returns the reply text (see `everalgo.llm.LLMClient`).
    fake = FakeLLMClient(responses=[_BOUNDARY_JSON, _BOUNDARY_JSON])
    messages = [
        ChatMessage(id="m1", role="user",   content="Let's talk about deployment.",     timestamp=1_700_000_000_000, sender_id="u_alice"),
        ChatMessage(id="m2", role="assistant", content="Sure — what's the target env?",  timestamp=1_700_000_001_000, sender_id="assistant"),
        ChatMessage(id="m3", role="user",   content="K8s. Which deployment strategy should I use?", timestamp=1_700_000_002_000, sender_id="u_alice"),
    ]

    # Streaming: hold `tail` between calls; pass prior tail + new messages each time.
    result = await adetect_boundaries(messages, llm=fake)
    cells, tail = result  # NamedTuple unpacking

    # Persist the closed cells and carry the tail into the next batch.

    # End-of-session: tail is forced into the last cell.
    result = await adetect_boundaries(messages, llm=fake, is_final=True)
    assert result.tail == []
    for mc in result.cells:
        print(mc.timestamp, len(mc.items))


asyncio.run(main())

The streaming state machine

adetect_boundaries deliberately holds back trailing messages as tail — the LLM cannot know whether a conversation continues beyond the last seen message. The caller maintains state:

tail: list[ConversationItem] = []

for batch in incoming_batches:
    result = await adetect_boundaries(tail + batch, llm=client)
    await persist(result.cells)
    tail = result.tail

# Session ends — flush everything.
final = await adetect_boundaries(tail, llm=client, is_final=True)
await persist(final.cells)

Migrating from 0.4.1

Breaking: DetectionResult now has two fields on both batch and step paths. Replace cells, tail, tail_complete = result with cells, tail = result; remove result.tail_complete accesses and completion-based branches. Persist result.cells, retain result.tail, and use the host's idle or session-end flush policy. The library does not schedule automatic flushing. The step prompt and decision schema are unchanged.

The model's strict batch schema also drops tail_complete. Update scripted replies and custom LLM adapters to return only reasoning and boundaries; the removed field is rejected as an extra key. Custom batch instructions must use zero-based new-episode start labels. The numeric cut position itself remains a Python slice boundary.

This release candidate replaces the completion prediction because an open window does not establish whether more related messages will arrive. It changes the segmentation policy and restores a one-sentence explanation of all decisions. The prompt is smaller, but this does not imply lower latency: the evaluated candidate generated longer responses and had higher boundary-call latency than 0.4.1. Effects and speed remain model-dependent.

Failures

Each call makes one LLM request and validates the reply against a strict pydantic model, which it also passes as response_format. A reply that does not fit — not JSON, a missing or extra key, a non-integer boundary — raises everalgo.llm.LLMOutputError; nothing is retried, and the client's own exceptions propagate unchanged. Retry policy belongs to the caller.

Tokenizer utilities

Boundary detection counts tokens with everalgo.utils.count_tokens (in everalgo-core), which uses OpenAI's o200k_base encoding via tiktoken.

Stubs

WorkspaceMemCellExtractor (Jira / Email / Confluence) is a placeholder in v0.x — all methods raise NotImplementedError. It is deliberately excluded from everalgo.boundary.__all__; import it from everalgo.boundary.workspace directly if you need the reserved name. Implementation lands in a future minor bump when the RawData contract is finalised.

Metadata

Release files for everalgo-boundary 0.4.2rc0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for everalgo-boundary 0.4.2rc0
File Size Uploaded Provenance Actions
everalgo_boundary-0.4.2rc0.tar.gz 44.3 kB Without provenance Details for everalgo_boundary-0.4.2rc0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for everalgo-boundary 0.4.2rc0
File Interpreter ABI Platform Actions
everalgo_boundary-0.4.2rc0-py3-none-any.whl Python 3 none any Details for everalgo_boundary-0.4.2rc0-py3-none-any.whl

Total release size: 73.6 kB

Release files / everalgo_boundary-0.4.2rc0.tar.gz

Download URL everalgo_boundary-0.4.2rc0.tar.gz
Size 44.3 kB
Tags Source
SHA-256 checksum
How to use checksums
5da599bf7d2f64376a45abdeb85d37e85f3d53465781d5469406caa319fd4041
BLAKE2b-256 checksum
How to use checksums
1ca0fa4882d84c62dcc695c6018eacadd4aac5289f718821157e5cb2727ddc80
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.12.12

Release files / everalgo_boundary-0.4.2rc0-py3-none-any.whl

Download URL everalgo_boundary-0.4.2rc0-py3-none-any.whl
Size 29.4 kB
Tags Python 3
SHA-256 checksum
How to use checksums
8c43a2d95dc143f86f219dd048c26c3410143e0bbd6125d4a492a417352e0aba
BLAKE2b-256 checksum
How to use checksums
b8808ddca3ec7ac10fd88adb1f7232ddbc1976858958eb63c658062b7cb8e92f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.12.12

Security for everalgo-boundary 0.4.2rc0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page