§ 06Capability Catalog — full index

Twelve capabilities.
Each scoped in SEU, each with a delivery target,
each signed off through your PMO tool.

The Catalog is the entire customer-facing surface. There is no portal where you submit work requests. You raise a ticket in Jira, Linear, ServiceNow, Azure DevOps, or monday.com — pre-installed during Sprint 0 — pick the Catalog item, scoped in SEU, with a target lead time. Your Authorized Sponsor approves. VISystems delivers. Eval Cloud writes the audit row.

SEUSenior Engineering Unit

Every Catalog item is scoped in SEU — our standard measure of senior engineering effort. SEU is not hours. It reflects the seniority of the engineer, the complexity of the work (Claude API, MCP integration, eval harness design, compliance documentation), and the full operational overhead of production-grade delivery: code review, eval gating, audit-ledger instrumentation, and PMO reconciliation. One Sprint = 100 SEU. A 40‑SEU capability is a focused engagement; an 80‑SEU capability involves broader architectural scope, more integration surfaces, and longer stabilization cycles.

The SEU scope on each item below is a baseline. Final sizing is confirmed during Sprint 0 based on your specific environment, integration surface, and compliance requirements.

BUILD
01

Eval Suite Build & Migration

BUILD · ESTABLISH OR MIGRATE EVAL HARNESS

SEU

40

Target

5 BD lead

Design, build, or migrate eval suites for Claude-powered systems. Covers eval-spec authoring, grading mode selection (exact-match, LLM-as-judge, semantic), historical baseline import, and CI integration hooks.

Deliverables

  • Eval-spec files (TypeScript/YAML)
  • Baseline dataset
  • Grading rubric
  • CI config

Evidence Artifacts

  • Eval Cloud suite record
  • Baseline snapshot
  • Migration diff report
02

Context Engineering & Optimization

BUILD · CONTEXT ARCHITECTURE, MEMORY, TOKEN BUDGET

SEU

30

Target

3 BD lead

Context architecture design for Claude workflows. Prompt architecture, memory management (compaction, persistent state), token budget optimization, context window utilization, tool orchestration review. Cache-aware, cost-aware — prompt engineering is a sub-task within context engineering.

Deliverables

  • Context architecture doc
  • Versioned prompt library
  • Memory & compaction strategy
  • Token budget analysis

Evidence Artifacts

  • Eval Cloud context audit record
  • Cost-per-call trend
  • Cache hit rate dashboard
03

MCP Server Implementation

BUILD · SCOPED, ISOLATED, EVAL-GATED

SEU

60

Target

5 BD lead

Scoped, isolated MCP servers with eval-gated deployment. Tool registration, permission boundaries, integration testing, registry publication.

Deliverables

  • MCP server package
  • Tool schema
  • Integration test suite
  • Registry entry

Evidence Artifacts

  • Eval Cloud MCP registry record
  • Deployment trace
  • Test coverage report
04

Agent Architecture & Orchestration

BUILD · FROM PRD TO PRODUCTION

SEU

80

Target

10 BD lead

End-to-end agent system design from PRD to production. Multi-agent orchestration, tool-use patterns, guardrails, human-in-the-loop flows, observability instrumentation.

Deliverables

  • Architecture doc
  • Agent implementation
  • Orchestration config
  • Guardrail specs

Evidence Artifacts

  • Eval Cloud architecture record
  • Agent trace logs
  • Orchestration topology map
RUN
05

Production Observability Setup

RUN · TRACES, SLOs, DRIFT, COST

SEU

40

Target

5 BD lead

Trace ingestion, SLO definition, drift detection rules, cost dashboards. Connects Claude API telemetry to your observability stack.

Deliverables

  • Trace pipeline config
  • SLO definitions
  • Drift detection rules
  • Cost dashboard

Evidence Artifacts

  • Eval Cloud observability record
  • SLO compliance report
  • Drift alert history
06

Eval-as-CI Integration

RUN · GITHUB / GITLAB / BITBUCKET

SEU

30

Target

3 BD lead

Wire eval suites into GitHub Actions, GitLab CI, or Bitbucket Pipelines. Eval gates block merge on regression. Results feed Eval Cloud.

Deliverables

  • CI pipeline config
  • Eval gate rules
  • Merge-block policy
  • Results webhook

Evidence Artifacts

  • Eval Cloud CI integration record
  • Gate pass/fail history
  • Regression log
07

Capacity & Cost Right-sizing

RUN · CACHE ROI, BATCH, MODEL-TIER

SEU

50

Target

5 BD lead

Cache ROI analysis, batch vs streaming optimization, model-tier selection (Opus/Sonnet/Haiku routing). Quarterly cost review cadence.

Deliverables

  • Cost analysis report
  • Cache strategy doc
  • Model routing config
  • Optimization plan

Evidence Artifacts

  • Eval Cloud cost record
  • Cache hit trend
  • Model-tier usage breakdown
RELY
08

Model Migration Concierge

RELY · NEW CLAUDE VERSION → RE-BASELINE

SEU

40

Target

2 BD · Pro target

Concierge service on every new Claude release. Re-baseline evals, regression analysis, prompt adjustments, release-readiness sign-off. VISystems-billed, included at no additional cost for the migration diff.

Deliverables

  • Migration assessment
  • Re-baselined eval suite
  • Regression report
  • Sign-off doc

Evidence Artifacts

  • Eval Cloud migration record
  • Before/after eval comparison
  • Release-readiness certificate
09

SR 11-7 Alignment Authoring

RELY · MODEL RISK MGMT DOCUMENTATION

SEU

60

Target

10 BD lead

Model risk management documentation aligned to SR 11-7 / OCC 2011-12. Covers model inventory, validation framework, ongoing monitoring plan, board reporting template.

Deliverables

  • Model inventory
  • Validation framework doc
  • Monitoring plan
  • Board report template

Evidence Artifacts

  • Eval Cloud compliance record
  • Audit trail
  • Document version history
10

Incident Forensics & Remediation

RELY · TRACE-BACKED ROOT-CAUSE WORK

SEU

50

Target

1 BD · sev1

Trace-backed root-cause analysis for production incidents. Forensic trace reconstruction, remediation plan, post-incident review, prevention controls.

Deliverables

  • Root cause analysis
  • Remediation plan
  • Post-incident review doc
  • Prevention controls

Evidence Artifacts

  • Eval Cloud incident record
  • Trace reconstruction
  • Remediation verification
EVOLVE
11

Quarterly Reliability Review

EVOLVE · CFO-READY OUTCOMES REPORT

SEU

20

Target

15 BD lead

CFO-ready outcomes report. Sprint utilization, eval pass rates, cost trends, capability roadmap co-authoring, next-quarter planning.

Deliverables

  • Quarterly report
  • Utilization analysis
  • Roadmap update
  • Next-quarter plan

Evidence Artifacts

  • Eval Cloud quarterly record
  • Utilization dashboard
  • Roadmap diff
12

Custom PMO Integration

EVOLVE · BUILDS NEW ADAPTER ON FRAMEWORK

SEU

80

Target

10 BD lead

Build a new PMO adapter on the framework for tools beyond the top 5 (Jira, Linear, ServiceNow, ADO, monday.com). Webhook config, bi-directional sync, Catalog template installation.

Deliverables

  • PMO adapter package
  • Webhook config
  • Sync rules
  • Catalog templates

Evidence Artifacts

  • Eval Cloud PMO integration record
  • Sync audit log
  • Adapter test results

Need to size which capabilities fit your team?

dev@visystems.com →See pricing tiers →