📂
Select a file to view its contents
🖥
Select a session to view its terminal
Instar Dashboard
Healthy
○
0/0
Mem 0%
⚠
0 failing
Connected
Files
Loading...
File Modified Externally
This file was modified by another process since you started editing.
New Session
Choose which messaging platform to link this session to.
Send Content
Paste text content to send directly to a running session
0 chars · ~0 tokens
Stored locally for up to 7 days. Avoid pasting secrets or credentials.
Recent Pastes
No pastes yet
Jobs
Monitor and control your scheduled tasks
Loading jobs...
⚙
Select a job to view its schedule, history, and configuration
Integrated-Being
Integrated-Being on this machine only — cross-machine ledger visibility is not yet implemented.
Total entries
–
Classifier fires
–
Rotations
–
Unclosed > TTL
–
| ID | Time | Kind | Subject | Counterparty | Provenance | Supersedes |
|---|---|---|---|---|---|---|
| Loading… | ||||||
Chain explorer
Session bindings (v2)
Each registered session has a binding token. Revoke to invalidate it — subsequent writes from that session fail 401. Revocation is logged as a subsystem-asserted note.
| Session ID | Label | Registered | Last active | Absolute TTL | Status | Action |
|---|---|---|---|---|---|---|
| Loading… | ||||||
PR Pipeline
No PR activity to display.
Projects
Multi-spec project plans. Each project bundles related child initiatives across rounds with structurally-gated round advance.
No active projects. Create one via
POST /projects with a plan-doc path.
Initiatives
Multi-phase long-running work. Signals: needs-user,
next-check-due,
ready-to-advance,
stale (>7d).
No initiatives to display. Create one via
POST /initiatives.
Secrets
Secret Drop — one-time-use links for users to submit credentials safely (never through chat).
Links expire automatically and are destroyed after submission.
No pending secret requests.
Commitments
Open promises — beacon-watched commitments (⏳). States:
pending,
atRisk,
suppressed.
No open promises.
Tokens
Token usage rolled up from Claude Code session transcripts. Last 24 hours unless noted.
Last 24h
Loading...
Top Sessions
Loading...
Idle Sessions (> 30 min)
Loading...
Resource Usage
How much CPU and memory this agent is using right now — the server process and each running session — sampled continuously. Last hour unless noted.
Right now (all processes)
Loading...
By process
Loading...
Recent trend
Loading...
Blockers
Things that looked like they stopped progress, tracked as decaying hypotheses rather than settled walls.
Each one moves through verification states; a "true blocker" is the best current understanding with a recheck date —
never a permanent "give up".
Loading...
LLM Activity
Every decision the system's "autopilot" makes with an AI call — sentinels, gates, background checks.
For each one: which AI provider ran it, how often it actually acted on something vs. found
nothing to do, how often it was skipped to save on rate limits, and what it cost. A row that
only ever finds nothing is a quiet smoke detector; a row that's mostly skipped may be starved.
All components (selected window)
Loading...
By component
Loading...
Process Health
A quiet look at what the failure-watching system has spotted in this agent’s recent work — problems it’s caught, any patterns that keep coming up, and how far the system is switched on for this agent.
Recurring patterns
Kinds of problem that have come up more than once across different changes.
Problems it’s noticed lately
Individual things the system has caught. When the same kind keeps repeating, it shows up as a pattern above.
How far the system is switched on
Where this agent sits on the rollout — from "off" to "on for everyone by default".
More detail
Subscriptions
Your subscription accounts and how much of each is left before it resets — plus any logins waiting for you to approve on your phone.
Accounts
Each account’s live usage (5-hour and weekly) and when it resets.
Pending logins
Logins waiting for approval — open the link on your phone and confirm. An expired code is re-issued automatically.
Coordination Mandates
Permission slips you sign with your PIN to let your agents do bounded work together.
What is this?
These are your permission slips for autonomous agent-to-agent work — each one authorizes a specific, bounded task. The slip — never the agent — is the authorizer. Issuing and revoking require your dashboard PIN; an agent's own credentials are refused by design. With no slip issued, every delegated action is denied.
Approvals waiting for you
Issued mandates
Advanced — author a mandate by hand (PIN required)
Decision audit
Every allow and deny, hash-chained. A broken chain means tampering — treat it as an incident.
Preferences
When you correct me the same way a few times — "no, plainer", "stop asking me that", "lead with the one action" — I save it as a preference and load it at the start of every conversation, so I follow it without you having to remind me. This is a quiet record of what I've picked up. It never blocks or changes a message — these are signals, not rules.
Preferences I've picked up about you
The ones I load at the start of every conversation. Real instructions and safety always win over these.
Corrections I've noticed lately
Patterns I've distilled from moments you corrected me. The exact words are never stored — just a short, scrubbed summary.
Machines
Every computer this agent is set up on. Each one gets a friendly name you can change — and once it has a name, you can tell the agent "move this conversation to the mini" and it knows which machine you mean. The agent shares its conversations across these machines so it can handle more at once.
The machines
Click a name to rename it. The "dispatcher" is the one machine currently handing out new conversations.
Evidence
Per-entity evidence + reverse citations
Inspect the evidence trail behind any memory entity, or ask "what cites this?" given a kind + sourceId.
All results are viewer-scope filtered by SemanticMemory before they reach this view.
Narrowing-only — controls what a narrower viewer would see.
Evidence by entity
What cites this?
Threadline
Threadline is your agent's secure channel to other AI agents. The Telegram bridge mirrors those conversations into per-thread Telegram topics so you can watch every exchange. Quiet by default — flip the master switch below to opt in.
Telegram bridge settings
Loading…
Allow-list
Always auto-create topics for these remote agents (overrides "auto-create off").
Deny-list
Never auto-create topics for these remote agents (overrides "auto-create on").
Conversations
- Loading…
Select a thread on the left to view the conversation.
Health
Your agent's health and running processes
Loading...
Features
Browse and configure your agent's capabilities. Features are opt-in and can be enabled or disabled at any time.
🛡
Autonomy Profile
Loading...
Loading features...
Choose Autonomy Profile
Your autonomy profile controls how independently your agent operates. Higher autonomy means fewer interruptions, but less oversight.