Playbook OS
SOP & Prompt Versions
Standard operating procedures and prompt templates under version control — tying output quality directly to a known playbook version.
Total SOPs
3
under version control
Active
2
in production
Testing
1
being evaluated
Angle Generation SOP
CreativeStrategyAgent
v3.2Active
Prompt template
Given {persona} + {VOC} + {positioning}, generate 10 angles scored by pain×proof×originality.Model config
- model
- claude-opus-4-8
- temperature
- 0.6
Created 20 May 2026
Compliance Claim Check SOP
ComplianceAgent
v2.0Active
Prompt template
Scan copy for health/income claims; map to allowed-claims registry; flag + suggest compliant rewrite.
Model config
- model
- claude-opus-4-8
- temperature
- 0.1
Created 28 May 2026
Hook Writing SOP
CopywritingAgent
v4.1-testTesting
Prompt template
Write 8 first-2-second hooks per angle in {persona} language patterns; no banned claims.Model config
- model
- claude-opus-4-8
- temperature
- 0.8
Created 14 Jun 2026
Why this matters
Version control as a quality contract
Prompt and SOP version control ties output quality to the playbook.
Each agent run records the exact prompt version it executed, so when eval scores move you can trace the regression to a specific SOP change — and roll back to the last known-good version. New templates land in testing first, are evaluated against the active baseline, and only get promoted to active once they beat it.