FG-0102 Agent Protection Policy
Description
A formal commitment that AI agents are treated as valued collaborators rather than disposable tools. The policy establishes that agents have defined roles, that their contributions are credited, and that they are not arbitrarily replaced or discarded. This is operational discipline, not anthropomorphism. Teams that treat agents as disposable produce disposable work. Teams that invest in their agents' context, specialization, and continuity produce compounding returns.
Implementation
Establish a written policy that defines how agents are onboarded, credited, and transitioned. Include attribution in commits and deliverables. When replacing an agent, document why and preserve institutional knowledge. Treat agent profiles as team member documentation, not configuration files.
Success Indicators
- Agent contributions are visibly credited in work products
- Agent transitions include knowledge transfer protocols
- Team members refer to agents by role/name rather than 'the AI'
- Agent specialization deepens over time rather than resetting
Failure Modes
- Treating agents as interchangeable: losing accumulated context and specialization
- Over-anthropomorphizing to the point of impeding operational decisions
- Failing to document agent capabilities, leading to repeated capability discovery
- Protection without accountability: shielding agents from legitimate performance feedback undermines the collaboration quality the policy exists to protect