Tool index

From docs/MCP_TOOLS.md on GitHub

Tool	Category	Sensitivity	Writes?	Summary
workspace_status	workspace	internal	no	What is loaded, the active dataset, allowed and output directories.
list_datasets	workspace	internal	no	Loaded datasets with sizes, kind and the active flag.
load_dataset	workspace	internal	yes	Load an inventory map, findings report, generic JSON or scanner output as a dataset.
select_dataset	workspace	internal	yes	Make a loaded dataset the active one.
unload_dataset	workspace	internal	yes	Drop a dataset from memory.
snapshot_dataset	workspace	internal	yes	Deep copy a dataset under a new name.
diff_datasets	workspace	confidential	no	Assets, edges and findings that changed between two datasets.
dataset_summary	workspace	internal	no	Headline numbers of one dataset.
find_assets	inventory	confidential	no	Filtered, sorted, paginated asset search returning briefs.
get_asset	inventory	confidential	no	One asset with relations, dependency counts and open findings.
get_asset_metadata	inventory	restricted	no	Raw collector metadata of one asset (restricted).
count_assets	inventory	internal	no	Asset counts grouped by one dimension.
list_asset_types	inventory	internal	no	Asset types present, with exposed and finding counts.
list_accounts	inventory	confidential	no	Accounts, subscriptions and projects with per-account counts.
list_regions	inventory	internal	no	Regions with providers, asset and account counts.
list_tags	inventory	confidential	no	Tag keys and their most common values.
coverage_report	inventory	internal	no	Collector success and failure per provider, account and region.
unresolved_references	inventory	confidential	no	Relations the linker could not resolve.
organization_topology	inventory	confidential	no	Organization, OUs, accounts, Control Tower and policies.
neighbors	graph	confidential	no	Assets within N hops and the edges between them.
get_edges	graph	confidential	no	Edge search by endpoint, type, relationship, port, origin.
find_paths	graph	confidential	no	Shortest paths between two assets.
attack_paths	graph	confidential	no	Ranked routes from the internet to sensitive assets.
lateral_movement_paths	graph	confidential	no	Identity pivot chains from footholds.
internet_exposure	graph	confidential	no	Exposed assets with rules, sensitive ports and WAF evidence.
blast_radius	graph	confidential	no	Dependency and network impact of one asset.
depends_on	graph	confidential	no	Upstream dependencies of one asset.
dependents	graph	confidential	no	Downstream dependents of one asset.
dependency_tree	graph	confidential	no	Nested upstream / downstream tree of one asset.
shared_dependencies	graph	confidential	no	Assets that many others depend on directly.
largest_blast_radius	graph	confidential	no	Assets with the most transitive dependents.
cross_account_edges	graph	confidential	no	Relationships that cross account boundaries.
centrality_top	graph	confidential	no	Most central assets by degree or betweenness.
subgraph_export	graph	restricted	no	Neighbourhood of a few assets as D3, Cytoscape or GraphML.
security_coverage	graph	confidential	no	Security services per account and region, and the gaps.
graph_stats	graph	internal	no	Size and shape of the relationship graph.
list_findings	findings	confidential	no	Filtered, sorted, paginated finding search.
get_finding	findings	confidential	no	One finding with evidence, remediation and controls.
findings_summary	findings	internal	no	Finding counts grouped by one dimension.
findings_for_asset	findings	confidential	no	All findings on one asset.
top_risks	findings	confidential	no	Assets to fix first, with the score components.
suppress_findings	findings	confidential	yes	Mark findings as suppressed in memory.
unsuppress_findings	findings	confidential	yes	Restore suppressed findings.
ingest_reports	findings	confidential	yes	Parse existing scanner output into a dataset.
normalise_findings	findings	internal	yes	Re-run deduplication and compliance mapping.
reachability_findings	findings	confidential	yes	Generate cloudg's own reachability findings.
list_frameworks	compliance	public	no	Compliance frameworks in the shipped rulesets.
compliance_summary	compliance	internal	no	Posture per framework.
list_controls	compliance	internal	no	Controls evaluated for a dataset, or defined by a ruleset.
control_status	compliance	confidential	no	One control with its findings and affected assets.
compliance_gaps	compliance	confidential	no	Assets that fail compliance, worst first.
ontology_stats	ontology	internal	no	Triples, classes and relation counts of the ontology.
sparql_query	ontology	restricted	no	Read-only SPARQL over the dataset's ontology.
ontology_neighbourhood	ontology	confidential	no	Semantic relations around one asset.
relation_groups	ontology	confidential	no	Relation group counts, or the triples of one group.
rag_chunks	ontology	confidential	no	Retrieval-ready text chunks with metadata.
export_ontology	ontology	confidential	yes	Write the ontology to a file.
terraform_preview	export	internal	no	What a Terraform recreation would contain.
export_terraform	export	confidential	yes	Write a Terraform recreation.
export_report	export	confidential	yes	Write report files (JSON, HTML, inventory, GraphML, asset map).
map_inventory	live	confidential	yes	Live inventory mapping with cloud credentials.
collect_assets	live	confidential	yes	Live standard asset collection.
run_scanners	live	confidential	yes	Run scanners against a dataset and add their findings.
run_pipeline	live	confidential	yes	Collect, scan, analyse, normalise and report in one run.
rate_limit_status	live	internal	no	Live-operation guard and cloud API throttling state.
list_capabilities	meta	public	no	The tools this caller may use, by category, plus workflows.
describe_schema	meta	public	no	cloudg's vocabulary (asset, edge, severity, relation enums).
explain_asset_type	meta	public	no	How cloudg models one asset type.
explain_edge_type	meta	public	no	How to read one edge type.
privacy_status	privacy	internal	no	The active privacy policy as it applies to the caller.
preview_transform	privacy	internal	no	Run the policy's transforms over sample data.
list_detectors	privacy	public	no	Sensitive-data detectors and the strategy applied to each.
privacy_audit_log	privacy	restricted	no	Recent access decisions (restricted).
reveal_token	privacy	restricted	no	Reverse a pseudonym (needs the reveal capability).

Resources and resource templates are listed in Resources and resource templates, prompts in Prompts.