Skip to main content
The ask-with-history tool goes beyond answering “what is X right now?” — it shows you the full arc of how something evolved. It returns both the current state of a topic and a chronological view of how it got there, drawn from Hyper’s predecessor graph of memory documents. Use it when you need to understand context that requires knowing what changed, when it changed, and why it changed — not just the final answer.

Parameters

Returns

current_state
string
A summary of what Hyper currently knows about the topic — the most recent, authoritative answer.
evolution
string
A chronological account of how the topic changed over time, drawn from the predecessor graph of memory documents. Each entry in the evolution reflects a meaningful state change with context about when and why it occurred.

How It Works

ask-with-history combines vector search with traversal of Hyper’s predecessor graph — a linked structure that connects each memory document to the earlier documents it superseded. This lets Hyper reconstruct not just what’s true now, but what was true at each earlier point, and what caused the transition.
ask-with-history is only available on v2 workspaces. If your workspace is on v1, calling this tool automatically falls back to ask() and returns the current state without history.

Example Calls

Pricing evolution

ask_with_history("how has our pricing evolved?")
Current state:
- Sprint plan: $30/seat/month with 4M token cap
- Blitz plan: $120/seat/month with 20M token cap

How pricing evolved:
- Previously: Sprint was $40/seat/month (reduced in April 2025)
- Previously: Blitz was $150/seat/month (reduced in April 2025)
- Original pricing launched at $50/$200 in beta (January 2025)
  Rationale: reduced pricing across the board to improve conversion
  after beta data showed price sensitivity in the $40–$50 seat range

Past decisions on a technical choice

ask_with_history("how has our approach to session management changed?")
Current state:
Sessions are stored in Redis with a 7-day TTL and server-side invalidation
on password reset or explicit logout.

How it evolved:
- Previously (March 2025): Sessions stored in Postgres. Switched to Redis
  after write latency spikes under concurrent login load.
- Originally (January 2025): JWT-only, stateless sessions. Abandoned because
  we couldn't invalidate tokens server-side — a requirement that emerged from
  the compliance review.

What someone was working on before their current project

ask_with_history("what was Marcus working on before AML?")
Current state:
Marcus is working on the AML (anti-money laundering) integration for the
payments module, targeting code-complete by May 9th.

How it evolved:
- Before AML (February–April 2025): Marcus led the payments webhook
  reliability project — reduced webhook delivery failures from 4.2% to 0.3%.
- Before webhooks (Q4 2024): Marcus was on the core billing engine,
  building the invoice generation pipeline.

When to Use ask-with-history vs. ask

Choosing between the two tools is straightforward:
Question typeTool
”What is X right now?”ask — faster, sufficient for current state
”How has X changed over time?”ask-with-history
”Why did we reverse the decision about X?”ask-with-history
”What did the team think about X before?”ask-with-history
”What is Marcus working on?”ask
”What was Marcus working on before his current project?”ask-with-history
If you’re unsure which to use, start with ask. It’s faster and handles most questions. Reach for ask-with-history when the user’s question explicitly involves change, history, or reversal — words like “before”, “used to be”, “changed”, “evolved”, “why did we reverse”, or “what happened to X” are good signals.

v1 Workspace Fallback

If your workspace hasn’t been upgraded to v2, ask-with-history gracefully degrades to ask — you’ll get the current state of the topic without the evolution history. The response won’t include a historical breakdown, but it won’t error either. Contact your workspace admin or the Hyper team to upgrade to v2 if you want full history traversal.