LlamaIndex Tools Integration: past dev

past dev is a memory API for AI agents. This package gives a LlamaIndex agent two tools over past dev:

• recall: a past dev Recall. The agent asks a question and gets the ranked memory of the current user, with the date, the kind and the quoted evidence of each entry.
• ingest: a past dev Ingest. The agent keeps one piece of information as a data point, so that a later recall can find it.

Install

pip install llama-index-tools-pastdotdev

Get a key

1. Sign up at past.dev and open a project.
2. Make a project API key. It starts with past_sk_.
3. Set it in the environment as PAST_API_KEY, or give it to the spec as api_key.

Use

from llama_index.core.agent.workflow import FunctionAgent
from llama_index.llms.openai import OpenAI
from llama_index.tools.pastdotdev import PastDevToolSpec

past_dev = PastDevToolSpec(
    identity=current_user.id,  # your id for the end user, from your own user record
)

agent = FunctionAgent(
    tools=past_dev.to_tool_list(),
    llm=OpenAI(model="gpt-4.1-mini"),
    system_prompt="Recall what you know about the user before you answer.",
)
response = await agent.run("What did I decide about the launch date?")

The identity

past dev keeps memory per identity: your id for the end user. Recall needs an identity, and the first ingest for an identity creates it.

Set identity in the PastDevToolSpec constructor, from your trigger or your user record. It is not a tool argument: the model cannot see it or change it. The same applies to audience and to the recall budget. Make one spec for each user.

The tools

The argument names are the API's own field names.

Parameters

Credits

Each recall and each ingest spends credits on your past dev plan. The agent decides when to call the tools, so tell it in the system prompt when to use them.

Errors

An API error raises PastDevError, with the error code, the HTTP status, the message and the X-Request-Id of the request. The LlamaIndex agent gives the error to the model as the tool output. The spec retries a 429 once, after its Retry-After delay. It does not retry other errors, for example 402 out-of-credits or a 400.

Links

• Quickstart
• API reference
• LlamaIndex tools guide
• A memory block for the same API: llama-index-memory-pastdotdev

License

MIT. See LICENSE.