export declare const HELP_TEXT = "# SpecShield MCP tools\n\nSpecShield answers one question inside your AI coding agent: **\"Is it safe to ship\nthis API change to my consumers?\"** All tools are **read-only** \u2014 they analyze API\ncontract compatibility and never modify your code.\n\n## Tools (in order)\n\n1. **is_change_safe** \u2014 THE DEPLOY GATE. Given a base and target spec, returns\n   `safeToMerge`, a risk level, and the exact blocking reasons. Use this before\n   merging or releasing an API change.\n2. **explain_breaking_changes** \u2014 plain-language explanation of what breaks, the\n   developer/consumer impact, and suggested migration.\n3. **generate_migration_guide** \u2014 a migration guide (markdown) + safe rollout steps.\n4. **generate_release_notes** \u2014 release notes for developer/customer/internal audiences.\n5. **compare_specs** \u2014 the raw diff (breaking / additions / modifications / warnings)\n   with a risk score. Prefer is_change_safe for a go/no-go decision.\n6. **run_governance_review** \u2014 deterministic API governance ruleset (design/policy\n   checks beyond breaking changes: missing operationId, undocumented error responses,\n   no security scheme, missing pagination, and more). Reviews a single spec via\n   `specContent`/`specPath`. **Paid feature** (Team plan or above; a FREE key\n   returns a `payment_required` error).\n\nTools 1\u20135 accept specs as inline content (`baseSpecContent`/`targetSpecContent`)\nor file paths (`baseSpecPath`/`targetSpecPath`); run_governance_review takes a\nsingle spec (`specContent`/`specPath`).\n\n## Privacy\n\nRequires a SpecShield API key (`SPECSHIELD_API_KEY`). Specs are sent to your\nconfigured SpecShield backend for analysis. This server never logs spec content or\nAPI keys.\n";