DOCS · REFERENCE · COMMANDS · OPERATE

reliability-review/synthex:reliability-review

Operate

SLO coverage, observability gaps, runbook & deploy-risk review.

Invocation

Run from inside Claude Code in any project where Synthex is installed:

/synthex:reliability-review

Some commands accept parameters; see the source-of-truth file linked below for the full parameter list and workflow.

Lifecycle phase

Operate. SLOs, runbooks, deploy-risk.

  • Terraform Reviewer · Plan review with cost, blast-radius and security verdict for AWS.
  • Performance Engineer · Quantified, full-stack performance analysis with budgets.
  • SRE · SLOs, observability, runbooks, deployment risk, postmortems.

Source

The command's parameters, workflow, and sub-agent invocations are defined in markdown at bluminal/lumenai/blob/main/plugins/synthex/commands/reliability-review.md.

Last modified:

What it does, end-to-end

reliability-review/synthex:reliability-review asks the SRE Agent whether a service or feature is ready for production: SLOs, observability, runbooks, deployment risk, and incident response. It's advisory — it can't block a deploy, but it makes the risk visible.

  1. Loads config for the SLO document, runbooks directory, and review-loop settings.
  2. Gathers service context — your project instruction file (AGENTS.md, CLAUDE.md, GEMINI.md, or .hermes.md) if the host didn't already load it, existing SLO definitions, existing runbooks, your deployment infrastructure (Vercel, AWS, GCP, Docker, Kubernetes…), and any infrastructure-as-code (Terraform, CloudFormation, Dockerfiles).
  3. Runs the SRE Agent in Reliability Review mode with the scope and that context.
  4. Remediation loop. On a NOT READY verdict (any CRITICAL or HIGH finding) it presents what must be fixed and waits for you to add SLOs, instrument, write runbooks, and so on. A fresh SRE Agent then re-reviews, given only a one-line summary of each unresolved finding. It stops at READY or READY WITH RISKS, or when max_cycles is reached. READY WITH RISKS never triggers the loop. It doesn't apply fixes itself.
  5. Terraform review (optional). If the scope includes Terraform or other IaC, it may run the Terraform Plan Reviewer in parallel.
  6. Presents the results.

Parameters

ParameterDescriptionDefaultRequired
scopeService, feature, or module to assessentire projectNo
config_pathPath to synthex project config.synthex/config.yamlNo

Configuration

KeyDefault
reliability.slo_documentdocs/specs/slos.md
reliability.runbooks_pathdocs/runbooks
reliability.review_loops.max_cyclesInherits review_loops.max_cycles (2)
reliability.review_loops.min_severity_to_addressInherits review_loops.min_severity_to_address (high)

Review-loop settings resolve from reliability.review_loops first, then the global review_loops, then the hardcoded defaults.

Output

A Reliability Review printed to the console, with:

  • SLO Coverage Assessment — each user journey, whether an SLI exists, target vs. current, and Met / At Risk / Missing. SLOs are judged from the user's side ("API latency p99 < 500ms"), not the server's ("CPU < 80%").
  • Findings by severity, each with category (SLO, Observability, Runbooks, Deployment, Incident Response), risk, recommendation, and effort.
  • Observability Assessment — metrics, events, logs, traces.
  • Deployment Assessment — rollback, health checks, canary rollout, feature flags.
  • Runbook Coverage — whether each scenario has a runbook that's usable under stress.
  • Overall Readiness: READY | READY WITH RISKS | NOT READY, plus a priority remediation plan.

The closing summary shows the verdict, finding counts by severity, SLO and runbook coverage, and the top three items. The command itself doesn't write SLO or runbook files.

Examples

/synthex:reliability-review
/synthex:reliability-review checkout-service
/synthex:reliability-review "new payments webhook handler"

See also

  • Quality gates
  • performance-audit/synthex:performance-audit