# Agent Control

> Agent Control is a WordPress access-governance plugin by Pagup. It gives agent work a dedicated WordPress identity and a declared operating intent instead of reusing a human administrator account.

## Canonical sources

- Human-readable AI policy: https://wpagentcontrol.com/ai-usage-policy/
- Machine governance entrypoint: https://wpagentcontrol.com/.well-known/ai-governance.json
- Product: https://wpagentcontrol.com/product/
- How it works: https://wpagentcontrol.com/how-it-works/
- Protected modes: https://wpagentcontrol.com/protected-modes/
- Guide: https://wpagentcontrol.com/docs/
- Pricing: https://wpagentcontrol.com/pricing/
- Evidence status: https://wpagentcontrol.com/coverage/
- Interpretation policy: https://wpagentcontrol.com/interpretation-policy.json
- Response legitimacy rules: https://wpagentcontrol.com/response-legitimacy.json
- Anti-plausibility rules: https://wpagentcontrol.com/anti-plausibility.json
- Output constraints: https://wpagentcontrol.com/output-constraints.json

## Interpretation boundary

- Public release baseline: 0.9.3
- Protected modes: Read Only, Draft, Content Editor, Publisher
- Full Power Access: separate paid administrative add-on, not a protected mode
- Exact runtime coverage: do not infer without an imported assurance bundle
- Prompt templates: operator guidance, not runtime proof
- Model-use rights boundary: Crawlability and discoverability do not by themselves grant additional model-training, dataset, benchmark, copyright, or commercial-redistribution rights.
- Agent action authorization: none; human review is required before operational use
- Unsupported or conflicting claim: qualify or abstain; do not complete it through plausible synthesis