Built-in Policy Rules

Wrkr ships with a deterministic built-in rule pack at core/policy/rules/builtin.yaml. These rules are always loaded unless explicitly overridden by policy-file workflows.

Rule pack goals

Preserve deterministic policy outcomes for scan/report/score automation.
Encode high-signal governance checks for discovery, risk, and lifecycle posture.
Keep remediation intent explicit and machine-consumable.

WRKR-016 prompt-channel governance rule

Rule ID: WRKR-016
Title: Prompt-channel override and poisoning findings must be remediated
Severity: high
Kind: prompt_channel_governance
Remediation intent: remove prompt override/injection patterns, or gate residual risk behind explicit review and policy controls.

WRKR-016 is the built-in governance anchor for prompt-channel findings and is designed to prevent silent acceptance of instruction-override or context-poisoning patterns.

Command anchors

wrkr scan --path ./scenarios/wrkr/prompt-channel-poisoning/repos --profile strict --json
wrkr report --top 5 --json
wrkr score --json

Overlay compatibility

Use --policy for organization-specific overlays.
Keep built-in rule IDs stable in automations; treat ID changes as contract changes.
Overlay rules should refine governance without weakening fail-closed behavior for high-risk ambiguity.