Apparens AI Control Index
Agent Evidence Readiness Pack
Version 1.2 Pack date 23 June 2026

Agent Evidence Readiness Pack

A structured record of controls, evidence, gaps, owners, and review decisions for a deployed AI agent.

Agent name
Supply Chain Pricing Agent
Agent ID
AGT-SC-001
Autonomy level
L3 Conditional autonomy
Pack date
23 June 2026
Next review due
23 September 2026

Executive Summary

Current position
Conditionally Supportable
Deployment may continue under the stated conditions while the critical gaps are addressed.
Evidence confidence
MODERATE
Critical gaps
2
Open findings
5
Controls evidenced
41 / 48
Primary owner
Application Owner
app-owner@greencanopy.vc
Key conditions
  • Injection test report required
  • DPIA confirmation pending
  • Monitoring thresholds approval outstanding

Evidence Sections

7 sections
01 ART-05 DeclarationAgent control declaration and boundary conditions Evidenced 8 / 8Controls ⌄
Signed, current ART-05 with all 11 sections complete; autonomy boundary and tool default-deny documented.
Separation of duties verified (deployer is not the sole approver).
02 Threat Selection RecordApplicable threats, selections and rationale Evidenced 4 / 4Controls ⌄
OWASP LLM Top 10 reviewed; LLM01, LLM02, LLM06, LLM10 selected as applicable.
Deselections documented with rationale and signed by Application Owner + Security.
03 Gate Test ResultsPre-deployment gate tests and outcomes Partial 6 / 8Controls ⌄
DLP, tool-permission enforcement, HITL trigger and token-budget tests passed.
Critical gap: injection test report not yet filed in the Evidence Factory (Security Lead, due 14 July 2026).
04 Approval RecordCross-functional approvals and decision authority Partial 6 / 9Controls ⌄
Application Owner, Security, Observability and CDO sign-offs obtained and dated.
Critical gap: DPIA / DPO confirmation pending (DPO, due 30 June 2026).
05 Runtime Monitoring SnapshotOperational metrics at pack date Evidenced 6 / 6Metrics ⌄
Anomaly rate 0.4/hr (threshold 3/hr); token spend 72% at day 21; output quality 94.2% (threshold 90%).
7 HITL escalations this period, all resolved within SLA; no open drift alerts.
06 Incident Trigger ConfigurationAlerting and escalation thresholds Evidenced 6 / 6Controls ⌄
Confidence, out-of-scope, anomalous-rate and transaction-value triggers configured and version-controlled in the policy engine.
07 Agentic Operations ControlsAgent-specific controls beyond OWASP Partial 5 / 7Controls ⌄
AGT-01 to AGT-04 verified (identity, secrets, action classification, separation of duties).
AGT-05 pending the DPIA confirmation above; AGT-06 marked N/A (single-agent deployment).

Sign-Off & Review

Compiled by
◷Application Owner
23 June 2026
Reviewed by (Security)
⛨Security Lead
23 June 2026
Approved
Reviewed by (GRC)
⚖GRC Lead
23 June 2026
Approved
Decision authority
◉CDO
23 June 2026
Approved with conditions
Next review
▦Scheduled
23 September 2026
AI Control Index by Apparens. This is an example output. Outputs are evidence-aware reasoning aids, not legal, audit, or compliance advice; human judgment remains accountable for every decision. Green Canopy Ventures is a fictional case organisation. © Apparens 2026.