Agently Docs

EN

中文

EN

中文

Appearance

Agent Integration ​

Default Agent already includes SessionExtension.

Attach / Detach ​

python
agent.attach_session(session)
agent.detach_session()

Convenience helpers ​

python
agent.enable_session_lite(chars=12000, messages=8)
agent.enable_session_memo(chars=6000, messages=12)
agent.enable_quick_session()
agent.disable_quick_session()

chat_history is auto-proxied ​

When a Session is attached, these methods write into Session:

  • set_chat_history(...)
  • add_chat_history(...)
  • reset_chat_history()

Before each request, session.current_chat_history is injected into the prompt.

Record control ​

python
agent.settings.set("session.record.input.paths", ["input", ["input", "user.id"]])
agent.settings.set("session.record.input.mode", "all")

agent.settings.set("session.record.output.paths", ["reply", "data.result"])
agent.settings.set("session.record.output.mode", "first")

Custom record handler ​

python
def my_record_handler(result):
    return [
        {"role": "user", "content": "custom input"},
        {"role": "assistant", "content": "custom output"},
    ]

agent.set_record_handler(my_record_handler)

Practical scenario: trigger different summaries after multi-turn chat ​

1) Full setup + auto trigger (lite / deep) ​

python
from agently import Agently
from agently.core import Session

agent = Agently.create_agent()

session = Session(agent=agent).configure(
    mode="memo",
    limit={"chars": 6000, "messages": 12},
    every_n_turns=4,
)
agent.attach_session(session)

Default policy order:

  • over char limit → deep
  • over message limit → lite
  • every_n_turns → lite

2) Manual trigger ​

python
agent.session.resize(force="deep")
agent.session.resize(force="lite")

3) Different handlers for lite vs deep ​

python
def lite_resize_handler(full_history, current_history, memo, settings):
    return full_history, current_history[-6:], memo

def deep_resize_handler(full_history, current_history, memo, settings):
    return full_history, current_history[-4:], memo

session.set_resize_handlers("lite", lite_resize_handler)
session.set_resize_handlers("deep", deep_resize_handler)

How summaries flow back into the next turn (long-term memory) ​

Key points:

  1. current_chat_history is injected automatically
  2. memo is not injected automatically — you must add it

Option A: inject memo before every request ​

python
async def inject_memo(prompt, settings):
    memo = agent.session.memo
    if memo:
        prompt.set("info.memo", memo)

agent.extension_handlers.append("request_prefixes", inject_memo)

Option B: manual injection ​

python
agent.info({"memo": agent.session.memo})
agent.input("continue...").get_text()

This creates the loop:

  • resize updates memo
  • next request injects memo
  • model keeps stable preferences and facts