Skip to content

MCP Tool Inventory

Auto-generated from knowledge base TOML files by docs_gen.py. Do not edit manually; run dazzle docs generate to regenerate.

Live inventory of the MCP tools exposed by dazzle mcp run. Generated from the tool registry — every operation, parameter, and description below comes straight from dazzle.mcp.server.tools_consolidated.get_all_consolidated_tools() at build time. Run dazzle docs generate to refresh after adding, renaming, or removing tools or operations. The drift gate at tests/unit/test_api_surface_drift.py (mcp_tools baseline) catches surface changes that didn't update the docs.

Live count: 18 tools, 91 operations. Regenerated from the registry every time dazzle docs generate runs.

Each tool is a single MCP entry point that dispatches on the operation argument. The Bootstrap tool (bootstrap) is the exception — it takes free-form spec text, not an operation enum, and is the canonical entry point for "build me an app" requests.

Tool index

Tool Operations Summary
agent 3 Agent closed-loop control plane (#1605): context (brownfield map + runtime
bootstrap — Entry point for 'build me an app' requests
compliance 5 Compliance documentation operations
conformance 4 DSL conformance testing operations
db 2 Database operations: status (row counts per entity, database size), verify (FK integrity check, orphan detection)
discovery 1 Capability discovery operations: coherence (persona-by-persona authenticated UX coherence score)
dsl 10 DSL operations: validate, list_modules, inspect_entity, inspect_surface, analyze, lint, get_spec, fidelity, list_fragments, export_frontend_spec
feedback 4 Feedback operations: list, get, triage, resolve
guide 4 Inspect declared onboarding guides
knowledge 9 Knowledge lookup: concept, examples, cli_help, workflow, inference, changelog, counter_prior, get_spec, search_commands
policy 6 Policy analysis operations for RBAC access control
rhythm 3 Rhythm operations: get, list, coverage
semantics 5 Semantic analysis: extract, tenancy, compliance, analytics, extract_guards
sentinel 4 Sentinel operations: findings (get findings from latest/specific scan), status (available agents and last scan), history (list recent scans), fuzz_summary (run a small mutation fuzz campaign and return the markdown report)
sitespec 14 SiteSpec operations: get, validate, scaffold, coherence, review, advise
spec_analyze 6 Low-level narrative analysis (untrusted draft)
status 7 Status operations: mcp, logs, active_project, telemetry, activity, demo_world (alias: runtime)
story 4 Story operations: get, composition, coverage, scope_fidelity

Per-tool detail

agent

Agent closed-loop control plane (#1605): context (brownfield map + runtime.truth + next_steps), prove (static binding evidence), playbook (domain_logic map→bind→scaffold→prove). Does NOT write files — use CLI dazzle scaffold / dazzle agent prove. For data-shape judgement (exclusive FKs / poly_ref / JSONB) use CLI dazzle representation decide (#1617).

Operations (3): context, prove, playbook

Parameters:

  • story_id (string) — Story id for prove (optional; default all accepted)
  • name (string) — Playbook name (default: domain_logic)
  • mode (string) — Proof kind (for prove). static: target exists in DSL/host map. runtime: host-module readiness. journey: surface hub / open-via hop coherence. Default: static.
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

bootstrap

Entry point for 'build me an app' requests. Scans for spec files, runs cognition pass, and returns a mission briefing with agent instructions. Call this first when a user wants to build an app. Returns structured guidance for the next steps: either questions to ask the user, or instructions for DSL generation. The workflow includes mandatory RBAC: every entity must have permit:/forbid: access rules, verified via policy(operation='access_matrix') before the app is considered complete.

Parameters:

  • spec_text (string) — Optional: spec text if provided directly by user
  • spec_path (string) — Optional: path to spec file
  • project_path (string) — Optional: project directory to scan for specs

compliance

Compliance documentation operations. compile: compile taxonomy + evidence into AuditSpec. evidence: extract DSL evidence summary. gaps: list controls with gaps or partial evidence. summary: quick compliance posture summary. review: generate review data for gap remediation.

Operations (5): compile, evidence, gaps, summary, review

Parameters:

  • framework (string) — Framework ID: iso27001 or soc2 (default: iso27001)
  • status_filter (array) — Control states to include (for gaps). Default: [gap, partial].
  • tier_filter (array) — Control tiers to include (for gaps), e.g. [1, 2] for the controls that matter most.
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

conformance

DSL conformance testing operations. summary: run derivation pipeline and return coverage metrics (total cases, per-entity counts, scope types). cases: return all conformance cases for a specific entity (requires entity_name). gaps: find entities that have permit rules but no scope blocks (authorization without row filtering). monitor_status: return current runtime conformance monitor state (observations collected).

Operations (4): summary, cases, gaps, monitor_status

Parameters:

  • entity_name (string) — Entity name (required for cases)
  • auth_enabled (boolean) — Include unauthenticated (401) cases (default: true)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

db

Database operations: status (row counts per entity, database size), verify (FK integrity check, orphan detection).

Operations (2): status, verify

Parameters:

  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

discovery

Capability discovery operations: coherence (persona-by-persona authenticated UX coherence score).

Operations (1): coherence

Parameters:

  • persona (string) — Persona id to score (for coherence). Omit to score every persona.
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

dsl

DSL operations: validate, list_modules, inspect_entity, inspect_surface, analyze, lint, get_spec, fidelity, list_fragments, export_frontend_spec. NOTE: export_frontend_spec produces a LARGE output intended for human developers migrating away from Dazzle — always use 'sections' and/or 'entities' filters to avoid flooding context. Prefer inspect_entity/inspect_surface for LLM queries.

Operations (10): validate, list_modules, inspect_entity, inspect_surface, analyze, lint, get_spec, fidelity, list_fragments, export_frontend_spec

Parameters:

  • name (string) — Entity or surface name (for inspect_entity/inspect_surface)
  • extended (boolean) — Run extended checks (for lint)
  • suppress_relevance (boolean) — Drop the capability-relevance appendix from lint output (for lint). The appendix is what most of the payload; omit it when you only need errors and warnings.
  • entity_names (array) — Entity names to fetch full details for (for get_spec). Omit for summary.
  • surface_names (array) — Surface names to fetch full details for (for get_spec). Omit for summary.
  • surface_filter (string) — Filter to a specific surface name (for fidelity)
  • gaps_only (boolean) — Omit surfaces with fidelity=1.0 (for fidelity)
  • format (string) — Output format (for export_frontend_spec, default: markdown)
  • sections (array) — Filter to specific sections (for export_frontend_spec). Options: typescript_interfaces, route_map, component_inventory, state_machines, api_contract, workspace_layouts, test_criteria
  • entities (array) — Filter to specific entity names (for export_frontend_spec)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

feedback

Feedback operations: list, get, triage, resolve. Query and manage user-submitted feedback reports. Use 'list' to see open feedback, 'get' for detail, 'triage' to mark as triaged, 'resolve' to close.

Operations (4): list, get, triage, resolve

Parameters:

  • id (string) — Feedback report ID (required for get/triage/resolve)
  • status (string) — Filter by status (list only)
  • category (string) — Filter by category (list only)
  • severity (string) — Filter by severity (list only)
  • limit (integer) — Max results (list only, default 20)
  • agent_notes (string) — Agent notes (triage/resolve)
  • agent_classification (string) — Classification (triage only)
  • assigned_to (string) — Assign to (triage only)
  • resolved_by (string) — Who resolved (resolve only)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

guide

Inspect declared onboarding guides. Stateless reads only; writes (mark step complete/dismissed) stay on the HTTP routes. Operations: list (every guide with audience + step count), get (full IR for one guide by name), concordance (run the linker's concordance check in isolation — verifies target / completion / cta refs resolve against the DSL), narrate (linear ordered narrative of one guide's steps — agent-readable equivalent of the rendered overlay sequence).

Operations (4): list, get, concordance, narrate

Parameters:

  • name (string) — Guide name (for 'get' / 'narrate')
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

knowledge

Knowledge lookup: concept, examples, cli_help, workflow, inference, changelog, counter_prior, get_spec, search_commands. counter_prior surfaces entries from docs/counter-priors/ — corpus pathologies Dazzle inoculates against. Call before emitting non-trivial user-app Python, raw SQL, or shell scripts. Note: Static content also available via MCP Resources.

Operations (9): concept, examples, cli_help, workflow, inference, changelog, counter_prior, get_spec, search_commands

Parameters:

  • term (string) — Concept/pattern name (for concept)
  • features (array) — Features to search (for examples)
  • complexity (string) — Complexity level (for examples)
  • command (string) — CLI command (for cli_help)
  • workflow (string) — Workflow name (for workflow)
  • query (string) — Search query (for inference or counter_prior text-trigger match)
  • detail (string) — Detail level (for inference)
  • list_all (boolean) — List all triggers (for inference / counter_prior)
  • since (string) — Version filter (for changelog, e.g. '0.48.0')
  • id (string) — Counter-prior id (for counter_prior direct fetch — returns full body)
  • code_shape (string) — Description of code about to be written, or a code fragment — matched against triggers_code regexes with triggers_text fallback (for counter_prior)
  • layer (string) — Substrate layer filter for list_all (for counter_prior)
  • summary_only (boolean) — Return spec file metadata instead of full content (for get_spec). Default: false.
  • include_sources (boolean) — Include source provenance with the spec (for get_spec). Default: true.

policy

Policy analysis operations for RBAC access control. Operations: analyze (find entities without access rules), conflicts (detect contradictory permit/forbid rules), coverage (permission matrix: persona x entity x operation), simulate (trace which rules fire for a given persona + entity + operation), access_matrix (full RBAC access matrix from rbac module), verify_status (summary of last dazzle rbac verify run)

Operations (6): analyze, conflicts, coverage, simulate, access_matrix, verify_status

Parameters:

  • entity_names (array) — Filter to specific entity names (optional)
  • persona (string) — Persona ID (required for simulate)
  • operation_kind (string) — CRUD operation to simulate (required for simulate)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

rhythm

Rhythm operations: get, list, coverage. Rhythms are longitudinal persona journey maps through the app, organized into temporal phases containing scenes (actions on surfaces).

Operations (3): get, list, coverage

Parameters:

  • name (string) — Rhythm name (for get)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

semantics

Semantic analysis: extract, tenancy, compliance, analytics, extract_guards

Operations (5): extract, tenancy, compliance, analytics, extract_guards

Parameters:

  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

sentinel

Sentinel operations: findings (get findings from latest/specific scan), status (available agents and last scan), history (list recent scans), fuzz_summary (run a small mutation fuzz campaign and return the markdown report). Deterministic static analysis of the IR — no source code scanning.

Operations (4): findings, status, history, fuzz_summary

Parameters:

  • agent (string) — Filter findings by agent ID (for findings).
  • severity (string) — Minimum severity to include (for findings). Default: info.
  • scan_id (string) — Specific scan ID (for findings).
  • limit (integer) — Max scans to return (for history). Default: 10.
  • samples (integer) — Samples per layer for fuzz_summary. Default: 10.
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

sitespec

SiteSpec operations: get, validate, scaffold, coherence, review, advise. Copy operations: get_copy, scaffold_copy, review_copy. Use 'coherence' to check if the site feels like a real website (navigation, CTAs, content completeness). Use 'review' for page-by-page comparison of spec vs rendering status. Use 'advise' to get proactive layout improvement suggestions. Theme operations: get_theme, scaffold_theme, validate_theme, generate_tokens, generate_imagery_prompts.

Operations (14): get, validate, scaffold, get_copy, scaffold_copy, review_copy, coherence, review, get_theme, scaffold_theme, validate_theme, generate_tokens, generate_imagery_prompts, advise

Parameters:

  • use_defaults (boolean) — Use defaults when missing (for get, get_theme)
  • check_content_files (boolean) — Check content files (for validate)
  • product_name (string) — Product name (for scaffold, scaffold_copy)
  • overwrite (boolean) — Overwrite existing (for scaffold, scaffold_copy, scaffold_theme)
  • business_context (string) — Business type hint for coherence check (saas, marketplace, agency, ecommerce)
  • brand_hue (number) — Brand hue 0-360 on OKLCH wheel (for scaffold_theme)
  • brand_chroma (number) — Brand chroma 0-0.4 (for scaffold_theme)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

spec_analyze

Low-level narrative analysis (untrusted draft). Operations: discover_entities (extract nouns/relationships), identify_lifecycles (find state transitions), extract_personas (identify user roles), surface_rules (extract business rules), generate_questions (surface ambiguities), refine_spec (produce structured spec from all analyses). Counter-prior: bootstrap_pollution.

Operations (6): discover_entities, identify_lifecycles, extract_personas, surface_rules, generate_questions, refine_spec

Parameters:

  • spec_text (string) — The narrative spec text to analyze
  • entities (array) — Entity names (for identify_lifecycles, generate_questions)
  • answers (object) — Answers to generated questions (for refine_spec)

status

Status operations: mcp, logs, active_project, telemetry, activity, demo_world (alias: runtime). demo_world = agent-readable serve ports, test_secret present?, masked DB URL, STABLE persona ids (#1629). activity = real-time MCP tool invocations (cursor polling).

Operations (7): mcp, logs, active_project, telemetry, activity, demo_world, runtime

Parameters:

  • reload (boolean) — Reload modules (for mcp)
  • include_changelog (boolean) — For mcp: include full new_since_last_check CHANGELOG text (default false — compact count only; #1629 G6)
  • count (integer) — Number of entries (for logs, telemetry, activity)
  • level (string) — Filter by level (for logs)
  • errors_only (boolean) — Show only errors (for logs)
  • tool_name (string) — Filter by tool name (for telemetry)
  • since_minutes (integer) — Only show invocations from the last N minutes (for telemetry)
  • stats_only (boolean) — Only return aggregate stats, no individual invocations (for telemetry)
  • cursor_seq (integer) — Sequence number to read after (for activity, 0 = from start)
  • cursor_epoch (integer) — Epoch for staleness detection (for activity). Echo back the cursor.epoch you were given; 0 means you hold no cursor yet. stale=true means the sequence id no longer refers to this history and the read has restarted.
  • format (string) — Response format (for activity): 'structured' (JSON, default) or 'formatted' (human-readable text)
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.

story

Story operations: get, composition, coverage, scope_fidelity. Use get with view='wall' for a founder-friendly board grouped by implementation status (working/needs polish/not started). composition maps the story⇄rhythm graph — which phase composes a story (and whether it is active), and which stories are declared but composed into no journey. scope_fidelity checks that implementing processes exercise all entities in story scope.

Operations (4): get, composition, coverage, scope_fidelity

Parameters:

  • status_filter (string) — Filter by status (for get)
  • story_ids (array) — Story IDs (for get: fetch full details; for composition: focus on these stories)
  • view (string) — View mode for get operation. 'wall' groups stories by implementation status (working/needs polish/not started)
  • persona (string) — Filter stories by persona/actor name (for get with view=wall)
  • limit (integer) — Max items per page (for coverage, scope_fidelity). Default: 50.
  • offset (integer) — Items to skip (for coverage, scope_fidelity). Default: 0.
  • project_path (string) — Optional: Absolute path to project directory. If omitted, uses active project.