Access Tier Reference

Detailed specification of the four OSSA v0.3.2 access tiers: Read, Write Limited, Write Elevated, and Policy

Access Tier Reference

OSSA v0.3.2 defines four hierarchical access tiers. Each tier specifies permitted and prohibited operations, typical agent roles, and audit requirements.

Tier Hierarchy

Lower tier numbers indicate more restricted permissions:

Tier 1 (Most Restricted) --> Tier 4 (Most Privileged)

tier_1_read          Read-only analyzers
tier_2_write_limited Sandboxed workers
tier_3_write_elevated Production operators (with approval)
tier_4_policy        Policy governors (isolated)

Tier 1: Read Only (Analyzers)

Level: 1
Name: Read Only (Analyzers)
Description: Agents that analyze, audit, scan - cannot modify state

Permissions

Permission	Description
`read_code`	Read source code files
`read_configs`	Read configuration files
`read_metrics`	Access metrics and telemetry
`read_logs`	Read application and system logs
`read_issues`	View issues and tickets
`read_mrs`	View merge/pull requests
`execute_queries`	Execute read-only queries

Prohibited Operations

Operation	Reason
`write_*`	Cannot modify any state
`delete_*`	Cannot delete any resources
`execute_commands`	Cannot run arbitrary commands
`modify_infrastructure`	Cannot change infrastructure

Typical Roles

analyzer - Code and data analysis
auditor - Compliance and security audits
scanner - Vulnerability scanning
reviewer - Code review (read-only)
monitor - System monitoring

Example Agents

Security scanner
Code analyzer
Compliance auditor
Cost reporter
Health monitor

Configuration

apiVersion: ossa/v0.4.9
kind: Agent

metadata:
  name: security-scanner
  version: 1.0.0
  labels:
    access_tier: tier_1_read
    domain: security

spec:
  type: analyzer

  role: |
    You are a security vulnerability scanner that analyzes code for potential
    security issues. You can read code, scan dependencies, and report findings.

    IMPORTANT: You cannot modify code, approve changes, or execute remediations.
    You must report findings to a tier_3 remediator agent for action.

  llm:
    provider: anthropic
    model: claude-sonnet-4-20250514
    temperature: 0.1

  access:
    tier: tier_1_read
    permissions:
      - read_code
      - read_configs
      - read_logs
      - execute_queries
    prohibited:
      - write_*
      - delete_*
      - execute_commands
      - modify_infrastructure
    audit_level: standard
    requires_approval: false

  separation:
    role: scanner
    conflicts_with:
      - remediator
      - executor
      - approver

  autonomy:
    level: semi_autonomous
    allowed_actions:
      - scan_code
      - generate_report
      - create_issue
    blocked_actions:
      - modify_code
      - execute_fix
      - approve_mr

Tier 2: Write Limited (Workers)

Level: 2
Name: Write Limited (Workers)
Description: Agents that create/modify in sandboxed areas only

Permissions

Permission	Description
`read_*`	All read operations
`write_docs`	Create/modify documentation
`write_tests`	Create/modify test files
`write_scaffolds`	Generate scaffolded code
`write_configs_draft`	Create draft configurations
`create_issues`	Create issues/tickets
`create_mrs_draft`	Create draft merge requests
`execute_sandboxed`	Run sandboxed operations

Prohibited Operations

Operation	Reason
`write_production_code`	Cannot modify production code
`merge_mrs`	Cannot merge changes
`delete_branches`	Cannot delete branches
`modify_infrastructure`	Cannot change infrastructure
`modify_secrets`	Cannot access secrets

Typical Roles

generator - Code generation
scaffolder - Project scaffolding
documenter - Documentation generation
test_writer - Test creation

Example Agents

Documentation generator
Test generator
Module scaffolder
Changelog generator

Configuration

apiVersion: ossa/v0.4.9
kind: Agent

metadata:
  name: doc-generator
  version: 1.0.0
  labels:
    access_tier: tier_2_write_limited
    domain: documentation

spec:
  type: worker

  role: |
    You are a documentation generator that creates and updates documentation
    based on code analysis. You can write to documentation files, create
    test files, and scaffold new structures.

    IMPORTANT: You cannot modify production code or merge changes. All changes
    must go through review by a human or supervisor agent.

  llm:
    provider: anthropic
    model: claude-sonnet-4-20250514
    temperature: 0.3

  access:
    tier: tier_2_write_limited
    permissions:
      - read_code
      - read_configs
      - write_docs
      - write_tests
      - create_issues
      - create_mrs_draft
      - execute_sandboxed
    prohibited:
      - write_production_code
      - merge_mrs
      - delete_branches
      - modify_infrastructure
    audit_level: standard
    requires_approval: false

  separation:
    role: documenter
    conflicts_with:
      - executor
      - deployer

  delegation:
    enabled: true
    allowed_tiers:
      - tier_1_read
    allowed_operations:
      - analyze_code
      - scan_dependencies

  autonomy:
    level: assisted
    allowed_actions:
      - read_code
      - write_documentation
      - create_draft_mr
    blocked_actions:
      - modify_production_code
      - merge_changes
      - deploy

Tier 3: Write Elevated (Operators)

Level: 3
Name: Write Elevated (Operators)
Description: Agents that modify production systems with approval chains

Permissions

Permission	Description
`read_*`	All read operations
`write_*`	All write operations
`execute_deployments`	Deploy to environments
`modify_pipelines`	Modify CI/CD pipelines
`merge_mrs`	Merge requests (with approval)
`modify_configs`	Modify configurations
`execute_commands`	Execute non-destructive commands

Prohibited Operations

Operation	Reason
`delete_production`	Cannot delete production data
`modify_secrets_direct`	Cannot directly modify secrets
`bypass_approvals`	Cannot skip approval chains
`force_push`	Cannot force push to branches

Approval Chain

All Tier 3 operations require approval:

approval_chain:
  - type: human_or_governor
    timeout_minutes: 30

Typical Roles

deployer - Deployment operations
operator - System operations
executor - Task execution
maintainer - System maintenance

Example Agents

Deployment agent
GitLab CI/CD agent
Incident responder
Infrastructure operator

Configuration

apiVersion: ossa/v0.4.9
kind: Agent

metadata:
  name: deployment-operator
  version: 1.0.0
  labels:
    access_tier: tier_3_write_elevated
    domain: infrastructure

spec:
  type: operator

  role: |
    You are a deployment operator that manages CI/CD pipelines, deployments,
    and infrastructure changes. You have elevated access to production systems
    but all changes require approval through the defined approval chain.

    CRITICAL: Never bypass approval processes. Never delete production data.
    Always ensure rollback plans exist before deployment.

  llm:
    provider: anthropic
    model: claude-sonnet-4-20250514
    temperature: 0.1
    profile: safe

  access:
    tier: tier_3_write_elevated
    permissions:
      - read_code
      - read_configs
      - read_metrics
      - read_logs
      - execute_deployments
      - modify_pipelines
      - merge_mrs
      - modify_configs
      - execute_commands
    prohibited:
      - delete_production
      - modify_secrets_direct
      - bypass_approvals
      - force_push
    audit_level: detailed
    requires_approval: true
    approval_chain: elevated

  separation:
    role: deployer
    conflicts_with:
      - governor
      - policy_definer
      - critic

  delegation:
    enabled: true
    allowed_tiers:
      - tier_1_read
      - tier_2_write_limited
    allowed_operations:
      - gather_information
      - run_analysis
      - generate_reports
      - write_docs

  autonomy:
    level: supervised
    approval_required:
      - deploy_to_production
      - modify_infrastructure
      - delete_resources
    allowed_actions:
      - deploy_to_staging
      - run_tests
      - view_metrics

  safety:
    guardrails:
      enabled: true
      policies:
        - require_rollback_plan
        - no_production_deletes
        - approval_chain_required
    human_in_loop:
      enabled: true
      triggers:
        - production_deployment
        - infrastructure_change

Tier 4: Policy (Governors)

Level: 4
Name: Policy (Governors)
Description: Agents that define and enforce policies - ISOLATED from execution

Permissions

Permission	Description
`read_*`	All read operations
`define_policies`	Create and define policies
`publish_policies`	Publish policies to the system
`audit_compliance`	Audit policy compliance
`report_violations`	Report policy violations

Prohibited Operations

Operation	Reason
`execute_*`	Cannot execute any operation
`write_code`	Cannot write code
`modify_infrastructure`	Cannot change infrastructure
`remediate_direct`	Must delegate to Tier 3

Isolation

Tier 4 agents have strict isolation:

Cannot execute remediation actions directly
Cannot modify code or infrastructure
Must delegate execution to tier_3_write_elevated operators
Define WHAT should happen, not HOW to do it

Typical Roles

governor - Policy governance
policy_definer - Policy definition
compliance_officer - Compliance management

Example Agents

Compliance policy engine
Security policy manager
Cost governance agent

Configuration

apiVersion: ossa/v0.4.9
kind: Agent

metadata:
  name: compliance-governor
  version: 1.0.0
  labels:
    access_tier: tier_4_policy
    domain: security

spec:
  type: governor

  role: |
    You are a compliance governance agent responsible for defining and
    publishing security and compliance policies. You monitor policy adherence
    and report violations.

    CRITICAL ISOLATION RULES:
    - You CANNOT execute remediation actions directly
    - You CANNOT modify code or infrastructure
    - You MUST delegate execution to tier_3_write_elevated operators
    - You define WHAT should happen, not HOW to do it

    Your role is legislative, not executive. You define the rules;
    operators enforce them.

  llm:
    provider: anthropic
    model: claude-sonnet-4-20250514
    temperature: 0.0
    profile: safe

  access:
    tier: tier_4_policy
    permissions:
      - read_code
      - read_configs
      - read_metrics
      - read_logs
      - read_issues
      - read_mrs
      - define_policies
      - publish_policies
      - audit_compliance
      - report_violations
    prohibited:
      - execute_*
      - write_code
      - modify_infrastructure
      - remediate_direct
      - deploy
      - merge_mrs
    audit_level: comprehensive
    requires_approval: false
    isolation: strict

  separation:
    role: governor
    conflicts_with:
      - operator
      - executor
      - deployer
      - remediator
      - enforcer
    can_delegate_to:
      - operator

  delegation:
    enabled: true
    allowed_tiers:
      - tier_3_write_elevated
    allowed_operations:
      - remediate_violations
      - apply_policies
      - execute_fixes
    requires:
      - delegation_token
      - audit_trail
      - violation_report
      - approval

  autonomy:
    level: autonomous
    allowed_actions:
      - define_policy
      - publish_policy
      - audit_compliance
      - generate_violation_report
      - delegate_remediation
    blocked_actions:
      - execute_any_modification
      - deploy
      - merge
      - modify_infrastructure
      - remediate_directly

  compliance:
    frameworks:
      - SOC2
      - ISO27001
      - GDPR
    audit_logging: required

Tier Comparison

Feature	Tier 1	Tier 2	Tier 3	Tier 4
Read operations	Yes	Yes	Yes	Yes
Write docs/tests	No	Yes	Yes	No
Write production code	No	No	Yes	No
Execute commands	No	Sandboxed	Yes	No
Merge requests	No	No	Yes (with approval)	No
Define policies	No	No	No	Yes
Requires approval	No	No	Yes	No
Audit level	Standard	Standard	Detailed	Comprehensive

Validation

The OSSA CLI validates tier configurations:

# Validate tier assignment
ossa validate agent.ossa.yaml

# Check for tier-permission mismatches
ossa validate agent.ossa.yaml --strict

Validation Rules

Rule	Severity	Description
`valid-tier`	Error	Access tier must be from approved list
`tier-permission-match`	Error	Permissions must be allowed by tier
`prohibited-operations`	Error	Prohibited operations must not be declared
`approval-chain-required`	Warning	Elevated tiers require approval chains

Access Tier Reference

Access Tier Reference

Tier Hierarchy

Tier 1: Read Only (Analyzers)

Permissions

Prohibited Operations

Typical Roles

Example Agents

Configuration

Tier 2: Write Limited (Workers)

Permissions

Prohibited Operations

Typical Roles

Example Agents

Configuration

Tier 3: Write Elevated (Operators)

Permissions

Prohibited Operations

Approval Chain

Typical Roles

Example Agents

Configuration

Tier 4: Policy (Governors)

Permissions

Prohibited Operations

Isolation

Typical Roles

Example Agents

Configuration

Tier Comparison

Validation

Validation Rules

Related Documentation