build_enrichment_proposal
ChatGPTUse this before any write tool when you need a read-only proposal for what to enrich next. It lightly composes upstream search, agent-package, and optional implementation-context without creating drafts.
create_agent_instruction
ChatGPTCreates a new agent instruction draft upstream. Provide instruction_key, title, agent_target, instruction_type, status=draft, content_body, and change_summary. generated_by is normalized to chatgpt when omitted or set to cursor_agent. Use only when the target instruction is intentionally being drafted, and use an agent_target/instruction_type supported by the upstream project.
create_bootstrap_project
ChatGPTCreates only the base upstream BootstrapProject shell. Use this when human intent to start a new project is explicit. The project is born as draft or analyzing, never active, does not create a GitHub repository, and does not create initial documents, rules, or agent instructions.
create_business_rule
ChatGPTCreates a new business rule draft upstream. Provide rule_key, title, description, rule_type, severity, status=draft, content_body, and change_summary. generated_by is normalized to chatgpt when omitted or set to cursor_agent. Use only for explicit draft creation and do not treat this as approval or publish.
create_document_version
ChatGPTCreates a new draft version for an existing document upstream. Provide document_key, document_status=draft, content_format, content_body, and change_summary. generated_by is normalized to chatgpt when omitted or set to cursor_agent. Use only when updating a known document and do not treat this as approval or publish.
create_project_document
ChatGPTCreates a new document draft upstream in bootstrap-yaml. Provide a stable document_key, title, upstream-supported document_type, document_status=draft, content_format, content_body, and change_summary. generated_by is normalized to chatgpt when omitted or set to cursor_agent; use origin_tool/source for more specific provenance. Use only after human intent is clear, and do not treat this as approval or publish. Discover valid document_type values from existing project documents or upstream validation errors; do not invent a type for the project.
create_trigger_rule
ChatGPTCreates a new trigger rule draft upstream. Provide trigger_key, title, description, trigger_event, action_description, status=draft, content_body, and change_summary. generated_by is normalized to chatgpt when omitted or set to cursor_agent. Use only for deliberate draft creation and expect human review before any downstream execution.
enrich_bootstrap_project
ChatGPTRequests controlled upstream BootstrapProject enrichment. This tool defaults to dry_run=true when omitted. Applied enrichment is blocked by default and requires the extra MCP environment flag. Even when applied, upstream creates only minimal missing drafts, never activates the project, never approves artifacts, and never provisions a GitHub repository.
get_agent_package
ChatGPTUse this when you need the packaged agent context for one project_key, including documents, rules, and agent instructions assembled upstream.
get_api_contract_context
ChatGPTUse this to retrieve a read-only API contract context package for one project, preferably by operation_id when known. It assembles upstream API metadata only and must not be treated as sync, import, or publication logic.
get_implementation_context
ChatGPTUse this when you need an assembled implementation context package for one project_key. Provide goal or related hints when useful.
get_project
ChatGPTUse this when you need the full Bootstrap Project context for one project_key, including optional structural flags.
get_project_creation_context
ChatGPTUse this before modeling a new BootstrapProject. It returns the official upstream project creation patterns, lifecycle defaults, project key rules, and validation checklist. This tool is read-only and never creates a project.
get_project_document
ChatGPTUse this when you already know the project_key and document_key and need the full sanitized document body.
list_agent_instructions
ChatGPTUse this when you need the agent instructions published for one project_key.
list_business_rules
ChatGPTUse this when you need the published business rules for one project_key.
list_code_endpoints
ChatGPTUse this to inspect read-only HTTP endpoints derived from imported code for one project. Prefer repository or service filters to keep the response narrow and reduce stale-context ambiguity.
list_code_files
ChatGPTUse this to inspect imported file inventory for one project without retrieving mutative repository features. Results reflect the upstream repository snapshot and may lag behind the source repository until refreshed upstream.
list_project_documents
ChatGPTUse this when you need the documents for one project_key. You can filter by document_type such as trd.
list_projects
ChatGPTUse this when you need the available Bootstrap Projects in DataDolphin. Optionally filter by a short search term.
list_trigger_rules
ChatGPTUse this when you need trigger or automation rules for one project_key.
search_api_operations
ChatGPTUse this to search imported API contract metadata for one project when you need read-only operation matches by query, method, path, tag, repository, or service. This does not trigger sync or contract import.
search_code_context
ChatGPTUse this to retrieve read-only code evidence for one project when you need files, symbols, or implementation snippets. Results depend on upstream imported code context and may be stale until bootstrap-yaml refreshes its repository snapshot.
search_project_documents
ChatGPTUse this when you need semantic or keyword-style search within one project_key. Prefer this before fetching a large document set.
validate_bootstrap_project
ChatGPTUse this after create_bootstrap_project or before human review to validate an existing BootstrapProject against upstream readiness and lifecycle expectations. This tool is read-only: it does not alter state, does not promote lifecycle_status, and does not create documents, rules, or agent instructions.