Whilora
Coding agents need a sense of human time.
Whilora is an availability-aware strategy layer for coding agents. It helps them ask better questions while you are present, continue safely while you are away, and return with a clear handoff instead of a pile of half-finished context.
Session rhythm
Human attention becomes agent strategy.
Ask blocker questions, confirm autonomy, define stop conditions.
Work independently, checkpoint progress, queue questions for return.
Quota event
Checkpoint, wait, resume
Return digest
Done, tested, assumed, next
The concept
Turn availability into an operating contract.
Today, coding agents often ask at the wrong time, pause for avoidable clarification, or lose momentum when quota resets. Whilora imagines a thin coordination layer that converts time, attention, quota, and risk into instructions the agent can actually follow.
Input signals
What Whilora understands
Availability
18m now, away 3h
Autonomy
Moderate, reversible work
Quota
Balanced spend, reset aware
Risk
Ask before destructive moves
Whilora layer
Convert context into agent strategy
Strategy engine
Agent operating contract
When to ask
What can proceed
When to checkpoint
How to resume
Agent behavior
Four modes across the session
- 1
Pairing
Ask concise blocker questions.
- 2
Pre-departure
Lock decisions and stop conditions.
- 3
Autonomous
Work, test, checkpoint, queue questions.
- 4
Return
Summarize changes, assumptions, and next action.
Session loop
Present time becomes autonomous momentum.
Whilora frontloads decisions while the human is present, preserves progress during interruptions, and returns with the smallest useful handoff.
Human present
Clarify blockers
Leaving soon
Agree boundaries
Away window
Work independently, checkpoint, resume after quota reset
Human returns
Report done, assumed, next
Use the live attention window
Whilora starts from a simple truth: the human may have ten minutes now and two hours away after that. The agent should behave differently in each window.
Frontload decisions
When the human is present, the agent should ask the questions that unlock progress, surface risky assumptions, and agree on what autonomy is allowed.
Work while the human is away
When the human leaves, the agent should keep moving on reversible work, run checks, save checkpoints, and queue non-blocking questions for later.
Recover after limits
If quota or context interrupts the run, the session should checkpoint, wait for the reset or approved fallback, and resume from the next useful action.
Why it matters
Better collaboration is not just a stronger model.
The most capable agent still needs to know when to ask, when to act, how much risk is allowed, and what to do when a quota window closes. Whilora focuses on that missing layer: the rhythm between human judgment and autonomous execution.
Attention-aware
Agents should know whether the human is available, leaving soon, away, or returning.
Autonomy with boundaries
The agent can decide reversible details while preserving explicit approval for destructive or high-risk changes.
Quota-conscious
Model limits, reset windows, and cost should shape strategy instead of surprising the user halfway through work.
Return-ready
When the human comes back, the agent should offer a concise handoff: what changed, what was assumed, what passed, and what needs a decision.
Human-agent work
A calmer way to run autonomous coding sessions.
Whilora points toward agents that respect limited human attention, preserve context across interruptions, and return with exactly the summary needed to make the next decision.