Local-first Replay Suite Private Telemetry
Recorded review ready

Rehearse the failure story before the live probe ever touches a private target.

Aegis-Air is built for air-gapped incident response: review the stored evidence pack, inspect the strongest replay, then wire the local engine only when the trust boundary is understood.

4 cases ยท 32/32 checks

Stored replay evidence stays visible even when the local engine is not wired.

Telemetry stays local

Public Pages only shows review data and explicit reachability state, never fake live telemetry.

Pack โ†’ replay โ†’ handoff

Reviewers can move from the brief to the top replay and commander handoff without leaving the surface.

Recorded mode SEV1 rehearsal

Checkout dependency outage

A 42.9% failure rate stays pinned to a dependency outage story, with schema-backed evidence and next actions ready for handoff.

Failure bucket dependency-outage
Confidence 86%
  • Read the runtime brief before discussing live target readiness.
  • Use the replay suite to prove taxonomy and severity accuracy.
  • Unlock live probe only after the local engine is intentionally wired.
01

Review the runtime brief and trust boundary.

02

Open the strongest replay and inspect the RCA summary.

03

Copy the commander brief only after the report schema reads cleanly.

Local-First Readiness

Recorded profile
Mode recorded-review
Replay Score 100%
Report Schema aegis-air-incident-report-v1
Target Reachability recorded
aegis-air-runtime-brief-v1 cases:4 checks:32 bucket:100%

Start with Review Path or Top Replay, then open the full pack only when you need deeper deployment proof.

Start with the pack, then read the strongest replay before you talk about the target.

Use the highest-scoring replay as the first trust-building surface.

Copy the commander brief only after the schema-backed report reads cleanly.

Keyboard: R routes ยท P pack ยท B proof bundle ยท X commander brief ยท T top replay ยท L load replay ยท ? help.

  • Open /health to confirm whether the live loop is intentionally wired.
  • Read /api/runtime/brief for replay score, trust boundary, and target reachability.
  • Use recorded review first, then switch to live probing only after the boundary is clear.
  • Validate the schema before integrating downstream handoff flows.
  • Treat replay score and live-probe evidence as separate signals.
  • Keep the structured RCA valid even if the local narrative model is unavailable.
  • Prefer schema-backed sections over free-form prose during handoff.
  • severity
  • failure_bucket
  • summary
  • supporting_evidence
  • immediate_actions
  • The target service is intentionally unreachable in recorded mode.
  • Pages deployments only show stored review data without the local engine.
  • The schema-backed report remains the source of truth even without narrative streaming.

Reviewer-first pack for replay evidence, target reachability, and downstream handoff readiness in air-gapped environments.

Proof Bundle 100% / 32 checks
Target Boundary recorded ยท aegis-air-target-api
  • [Proof] Replay suite โ†’ /api/replay-suite
  • [Proof] Runtime brief โ†’ /api/runtime/brief
  • [Proof] Schema contract โ†’ /api/schema/report
  • Read the runtime brief before claiming target readiness.
  • Inspect the highest-scoring replay and confirm severity/bucket alignment.
  • Copy the commander brief only after the RCA summary reads cleanly.
  • live-probe
  • webhook-alert
  • recorded-review

Target Infrastructure

โ— STANDBY
P95 Latency --
Error Rate --

Shortcut bar: R routes ยท P pack ยท C claim ยท T top replay ยท L load replay ยท ? help

Run Objective

Probe the target, capture failure signals, classify the incident, and return a report with evidence and immediate actions.

Live Probe Loop

Run a short probe loop against the target API. Aegis-Air will still return a structured report if Ollama is unavailable.

Latest Incident Report

No run yet
Severity --
Failure Bucket --
Confidence --

Run the live probe loop or inspect the replay suite to populate a structured incident report.

  • Replay and live runs both populate evidence here.
  • Actionable next steps appear here after each incident review.

Operator Trace

Waiting for incident triggers...

Replay Suite

Loading...
Overall Score --
Rubric Checks --
Severity Accuracy --
Taxonomy Coverage --
Loading replay cases.
Top Bucket --
Attention Runs --
Worst Score --
Visible Runs --
  • Loading replay drift signals.
Loading drift board.