Skip to main content

​
What This Does

Agent packs are grouped AI workflows that use LoopIQ context to answer operational questions, assess risk, recommend actions, and produce reusable signals. Instead of calling one isolated agent at a time, LoopIQ can route a workflow to a pack that combines the right agents, evidence sources, and output sections for the job. Examples:
  • Assess release readiness
  • Create remediation plan
  • Generate release dossier
  • Plan sprint from backlog
  • Triage incident
  • Generate test coverage

​
Why Agent Packs Matter

Agent packs help make AI behavior more consistent across LoopIQ. They give users:
  • the same readiness sections in web and mobile
  • traceable evidence read by the pack
  • findings and risks grounded in LoopIQ records
  • recommendations that can become governed actions
  • reusable agent signals for dashboards and Prediction Center
  • auditability for what ran, why it ran, and what it produced

​
Common Agent Packs

​
Release Governance Pack

Use this pack to assess release readiness, blockers, evidence gaps, provider findings, approvals, and release governance. Typical outputs:
  • verdict
  • readiness gaps
  • provider findings
  • evidence read
  • pending approvals
  • recommended actions

​
Sprint Planning Pack

Use this pack to refine backlog, prioritize work, estimate effort, break work into tasks, and create sprint planning recommendations. Typical outputs:
  • candidate backlog items
  • planning assumptions
  • prioritized work
  • owner or team recommendations
  • action plan

​
Code and Security Pack

Use this pack to review pull requests, static analysis, dynamic analysis, secrets, open-source license risk, API compatibility, and security findings. Typical outputs:
  • code or security findings
  • severity and confidence
  • impacted services
  • remediation recommendations
  • evidence links

​
Testing Pack

Use this pack to generate test cases, classify test types, validate test assets, evaluate traceability, and review automation coverage. Typical outputs:
  • test coverage gaps
  • generated manual test case suggestions
  • automation recommendations
  • traceability findings
  • quality risks

​
Incident and Change Pack

Use this pack to summarize incidents, classify impact, generate runbooks, support rollback decisions, and create postmortem context. Typical outputs:
  • incident summary
  • blast radius
  • change impact
  • recommended next actions
  • post-incident follow-up

​
Product and Customer Pack

Use this pack to analyze ideas, requirements, feedback, market signals, personas, roadmap context, and business cases. Typical outputs:
  • product themes
  • customer sentiment
  • market or competitive insights
  • roadmap recommendations
  • requirement drafts

​
Platform and Cost Pack

Use this pack to review cloud cost, budget risk, environment readiness, MLOps, architecture, infrastructure, and feature flags. Typical outputs:
  • cost risks
  • optimization ideas
  • environment findings
  • platform readiness
  • architecture recommendations

​
Collaboration Pack

Use this pack to summarize meetings, standups, decisions, action items, notifications, and response drafts. Typical outputs:
  • meeting summary
  • action items
  • owner follow-ups
  • status update draft

​
What Is an Agent Signal?

An agent signal is a structured output saved by LoopIQ after an agent or pack run. Signals may include:
  • tenant
  • workflow
  • entity scope
  • evidence read
  • findings
  • risks
  • recommendations
  • actions
  • confidence
  • approval requirement
  • trace ID
  • cost or latency metadata
Agent signals let LoopIQ dashboards and Prediction Center reuse the same evaluated outputs instead of repeatedly rebuilding logic from raw data.

​
Where Users See Pack Results

Pack results may appear in:
  • Helix web chat
  • LoopIQ Helix mobile
  • release certification detail pages
  • release dossier generation
  • remediation planning
  • sprint planning
  • incident triage
  • test coverage workflows
  • Prediction Center
  • analytics dashboards
  • admin scorecards
The exact surfaces depend on your organization’s permissions, subscription, and enabled workflows.

​
Triggered and Scheduled Runs

Agent packs can run from user actions, scheduled jobs, or source-system events. Examples:
  • a release certification is refreshed
  • an approval status changes
  • a release dossier is generated
  • a GitHub pull request event arrives
  • a CI pipeline finishes
  • a Cloud Deploy rollout changes phase
  • a PagerDuty incident changes status
  • a nightly release risk scan runs
LoopIQ records durable trigger jobs so pack execution can be retried, monitored, and audited.

​
Review Pack Output

When reviewing pack output, check:
  • the tenant, team, release, or entity scope is correct
  • the evidence read matches the records you expected
  • recommendations are tied to findings
  • confidence is appropriate for the evidence
  • write actions require approval where needed
  • stale or missing evidence is clearly identified
Do not approve actions when the pack output is not grounded in the expected LoopIQ records.
  • Use release packs for release decisions, not ad hoc prompts.
  • Prefer pack sections over free-form markdown for readiness and audit workflows.
  • Review low-confidence outputs before acting.
  • Treat missing evidence as a signal, not as permission to proceed.
  • Use scorecards to identify agents that need schema or grounding improvements.
  • Keep provider integrations healthy so packs have fresh evidence.