Tools / Memory

get_context_pack()

Key

The flagship read. One typed pack for a question: facts, beliefs, assumptions, hypotheses, causal edges, prior similar decisions with their outcomes, open contradictions — plus what's missing (missing_evidence) and where the system abstained. Every item is stamped available_at_decision_time, and the retrieval run is persisted so the pack can be reconstructed later.

get_context_pack(question, as_of, detail)→ CONTEXT PACK

Callable over MCP as get_context_pack(question, as_of, detail). REST mirror POST /v1/context-pack Planned — alpha is MCP-only

Why your agent calls it

The call before any judgment. One question in; one typed pack out — facts, beliefs, causal edges, prior similar decisions and how they turned out, open contradictions, and an explicit list of what's missing. Every item stamped with whether it was knowable at decision time. This is what "the agent checked before money moved" actually means.

What it returns

get_context_pack() · Everything knowable for a decision, stamped and typed.ILLUSTRATIVE

Call it from an agent

Over MCP, your agent invokes it on its own when the moment calls for it — just describe the situation. A Python SDK for headless runtimes is planned; this is what the call will look like.

agent · python sdkSDK · PLANNED — ALPHA IS MCP-ONLY

from cascadian import Cascadian

cx = Cascadian(api_key=os.environ["CASCADIAN_API_KEY"])

result = cx.get_context_pack(
    question="Should we cut the discount tier or raise onboarding spend?",
    as_of="2026-06-11T17:00:00Z",
    detail="summary",
)

Parameters

Param	Type	Description
questionREQUIRED	string	The decision question to assemble context for.
as_of	timestamp	Assemble only what was knowable then. Default now — pinned automatically inside an open decision.
detail	enum	"summary" (default, ≤2k tokens) or "full".

Example response

get_context_pack()JSON · ILLUSTRATIVE

{
  "question": "Should we cut the discount tier?",
  "as_of": "2026-06-11T17:00:00Z",
  "facts": [
    { "id": "fct_2231", "s": "discount tier live since 2025-09",
      "available_at_decision_time": true }
  ],
  "beliefs": [
    { "id": "mem_0871", "s": "discount pulls deals forward; lift unclear",
      "status": "approved" }
  ],
  "causal_edges": [
    { "edge": "discount → win_rate", "status": "approved", "confidence": 0.7 }
  ],
  "prior_decisions": [
    { "id": "dec_0093", "q": "raise list price 5%", "variance": "within interval" }
  ],
  "contradictions": [],
  "missing_evidence": ["no enterprise win-rate data after 2026-04"],
  "abstained": false,
  "retrieval_run_id": "ret_5531"
}

Try it — sandbox

Run this call right here. The response is canned and spec-true — the exact shape your agent gets back over MCP, on sample data.

Sandbox — canned, spec-true responses. No key, no server, no live data.

Sandbox — sample responseget_context_pack()

// Press RUN to call get_context_pack() in the sandbox.

Sample — canned from the spec, not a live server

Related — Memory

keybootstrap()The ~200-token session card your agent reads first.keyingest_source()Store a raw source — chunked, embedded, indexed.keyassert_fact()Write a temporal fact with an honest available_at.keypropose_memory()Propose a memory item — contradiction-checked on write.keylink_memories()A typed relationship edge, with rationale.keysupersede_memory()Revise without erasing — new row, old row closed.keysearch_memory()Hybrid retrieval with an honest empty result.keyget_facts()What was true, as of any moment.keyget_entity()The versioned dossier for one entity.keyget_memory()The full evidence card, by id.keylist_contradictions()Open conflicts waiting on a human.