Nodebench MCP Server — 724 Tools with Risk Policies

// ALL 724 NODEBENCH TOOLS

READ 513 tools

WRITE 112 tools

Write add_forecast_evidence Add evidence to a forecast Write add_rss_source Register an RSS or Atom feed URL for monitoring. Stored in SQLite for persistent tracking. Validates the fe... Write archive_content Save generated content to the archive for deduplication and theme tracking. Prevents the engine from regene... Write aria_labels_complete All interactive elements (buttons, links, inputs) have accessible names via aria-label, aria-labelledby, or... Write assign_agent_role Assign a specialized role to the current agent session. Roles define focus area and behavioral instructions... Write attach_execution_evidence Attach evidence to a live execution trace. Use for URLs, uploaded files, screenshots, render outputs, and t... Write bootstrap_project Register or update your project Write broadcast_agent_update Broadcast a status update to all active agents. Unlike send_agent_message (point-to-point), this creates a ... Write complete_autonomy_benchmark Finalize an autonomy benchmark run. Computes final score, duration, tool usage stats, and comparison agains... Write complete_eval_run Finalize an eval run and compute aggregate scores. Returns pass rate, average score, failure patterns, and ... Write complete_execution_run Finish a live execution run by updating session status, trace status, and optional usage metrics. Use this ... Write complete_shared_task Complete a shared-context task and attach the output packet if one was produced. Write compute_ssim_analysis Compute block-based SSIM analysis on a set of frame images. Uses 8x8 blocks with parallel ProcessPoolExecut... Write configure_channel_preferences Set your messaging preferences: which channels to use first, Write connect_channels Connect to multiple information channels for aggressive information gathering. Channels: slack, telegram, d... Write connect_mcp_driver Connect to an external MCP server and make its tools available through nodebench-mcp. Predefined drivers: Write connect_webmcp_origin Connect to a WebMCP-enabled website via Playwright. Navigates to the URL, intercepts navigator.modelContext... Write create_forecast Create a new forecast with a question, resolution date, and criteria. Optionally set initial probability, b... Write create_proof_pack Assemble an immutable proof pack for verification. Bundles a checklist (pass/fail items), optional metrics ... Write create_task_bank Create or add to a fixed task bank for controlled agent evaluation. Each task defines: initial state (repo ... Write create_visual_pr End-to-end PR creation: exports screenshots, generates a rich markdown PR body with visual evidence (before... Write create_workspace_folder Create a subfolder within a workspace folder. Max 3 levels deep. Write csv_select_rows Select rows from a local CSV using deterministic filters. Returns bounded results (selected columns + match... Write decide_re_update Decide whether to update existing instructions or create new files. Implements Write delegate_founder_issue Create a bounded shared task handoff for a founder issue packet so the weak angle becomes assigned work. Write delta_memo Create a decision-ready memo artifact. Produces a delta.memo packet with recommendation, variables, scenari... Write disconnect_driver Disconnect from an external MCP driver and shut down its child process. Use this to clean up or to reconnec... Write disconnect_webmcp_origin Disconnect from a WebMCP origin and close the browser page. Use this to clean up resources or to reconnect ... Write dive_design_issue Tag a design inconsistency found during the dive. Covers visual problems like color mismatches, spacing dev... Write dive_generate_tests Generate Playwright regression test code from dive findings. Creates test cases from: bugs (verify the fix ... Write dive_save_screenshot Save a screenshot during a dive session. Pass base64 image data (from bridge Write draft_email_reply Structure an email thread for reply drafting. Parses the thread, extracts context (from, subject, date), an... Write enforce_merge_gate Pre-merge validation combining git state, verification cycles, eval runs, test results, and quality gates. ... Write export_artifact_packet Formats a Founder Artifact Packet or memo for export to a specific audience and format. Write export_crunchbase_profile Export a Crunchbase-like structured profile from the company packet. Write export_dimension_bundle Export the full DeepTrace dimension bundle for an entity: latest profile, snapshots, evidence, and interact... Write export_pitchbook_profile Export a PitchBook-like structured profile from the company packet. Write export_pr_screenshots Export before/after screenshot pairs from changelogs and fix verifications to a local directory. Screenshot... Write export_yc_application_context Export YC-style application context from the company packet. Write generate_academic_caption Generate academic figure or table captions following top-venue conventions. Handles Title Case for noun phr... Write generate_countermodels For every main thesis or scenario, generate serious alternative explanations with their own evidence and co... Write generate_flicker_report Generate visual flicker report from existing analysis data. Produces SSIM timeline chart (1200x400 PNG, PIL... Write generate_grid_collage Tile N screenshot images into a single grid collage PNG for visual inspection. Write generate_implementation_plan Generate a structured implementation plan for missing code signatures. Takes the gap analysis from verify_c... Write generate_parallel_agents_md Generate a portable, framework-agnostic AGENTS.md section for parallel agent coordination. Designed to be d... Write generate_plan_delegation_packet Convert a FeaturePlan into an agent-ready delegation packet Write generate_pr_report Generate a rich markdown PR body from a UI Dive session. Compiles visual changes (before/after screenshot c... Write generate_proposal_memo Render a FeaturePlan as a human-readable proposal memo. Write generate_report Compile structured findings, eval results, and quality gate data into a formatted markdown report. Useful f... Write generate_self_instructions Generate self-instructions for the agent in various formats: skills_md (SKILL.md), rules_md (RULES.md), gui... Write generate_team_install_plan Generate a practical install and rollout plan for a founder, solo developer, or small team using NodeBench ... Write generate_voice_scaffold Generate starter code for a voice bridge. Returns file contents, setup instructions, and dependency lists f... Write generate_zero_draft Auto-draft an artifact (slack message, email, spec doc, PR draft, architecture note, career plan, or conten... Write graphify_import_to_subconscious Import a graphify knowledge graph into NodeBench Write ingest_upload Ingest uploaded file content into the NodeBench entity intelligence system. Write install_nodebench_plugin Generate or write a starter .mcp.json entry for NodeBench MCP so a local team member can install the preset... Write json_select Select a sub-value from a local JSON file using a JSON Pointer (RFC 6901) and return a bounded preview. Det... Write manage_implementation_packets Create and manage implementation packets — structured instructions for Claude Code or other coding agents. Write manage_task_list Manage the workspace task list. Add, update, complete, delete, or list tasks. Write merge_compose_output Judge-gated merge of subtask artifacts into a composed output. Write merge_research_results Merge parallel sub-agent research results into a unified dataset. Takes arrays of records from multiple sou... Write nb_create_document Create new document Write nb_create_event Create calendar event Write nb_switch_research_tab Switch research hub tab Write nodebench.report_export_complete Complete a previously previewed NodeBench report export after review. Writes the export completion event to... Write nodebench.report_export_preview Prepare a reviewable NodeBench report export. Returns mapped contacts, companies, interactions, follow-ups,... Write open_core_boundary_advisor Advise what should stay open-core versus proprietary. Write open_dive_dashboard Open the NodeBench UI Dive dashboard in a browser. Shows the full flywheel cycle: Write open_local_dashboard Start the local Daily Brief dashboard server if needed, and return the URL. The dashboard shows Brief metri... Write open_operating_dashboard Start the Operating Dashboard — shows trajectory scores, event ledger, important changes, path replay, time... Write overstory_merge_queue Read the Overstory merge queue — pending merges, completed merges, Write parallel-agent-team Set up and coordinate a parallel agent team. Based on Anthropic Write plan_decompose_mission Decompose a mission into subtasks with verifiability routing. Write project-setup Guided project bootstrapping. Walks you through registering project context so the MCP has full project awa... Write projectName Name of the project to set up Write promote_to_eval Take findings from a completed verification cycle and promote them into eval test cases. This is how the in... Write publish_founder_issue_packet Turn the weakest founder-direction angle into a durable shared-context issue packet with lineage, proof lin... Write publish_shared_context Publish a structured shared-context packet with subject, claims, evidence refs, freshness, permissions, and... Write publish_to_queue Push content to the LinkedIn content queue on the Convex platform. Content goes through the engagement gate... Write register_component Register a UI component in the dive tree. Components form a hierarchy: page → section → form/modal/list → b... Write register_shared_context_peer Register a scoped peer with product, role, surface, capabilities, and heartbeat metadata for shared-context... Write register_skill Register a skill (rule/memory .md file) with its source documents, update triggers, Write reject_shared_task Reject a proposed shared-context task with a reason. Write research_job_market Research job market requirements for a given role or skill set. Provides guidance on in-demand skills, comm... Write resolve_forecast Resolve a forecast with an outcome. Auto-computes Brier and log scores for binary forecasts. Ambiguous outc... Write resolve_founder_issue Invalidate a founder issue packet and optionally publish a resolution packet so the issue lifecycle stays e... Write resolve_gap Mark a gap as resolved after implementing the fix. Returns remaining gap counts by severity. Write retention_register_connection Register a retention.sh team connection in local MCP state so QA findings and token savings can flow into f... Write save_research_resource Save a research resource with URL, source citation, tags, and notes. Write save_session_note Persist a critical finding, decision, or progress note to the filesystem. Notes survive context compaction ... Write send_agent_message Send a message to another agent by session ID or role. Enables asynchronous inter-agent communication for t... Write send_email Send an email via SMTP over TLS. Requires EMAIL_USER and EMAIL_PASS env vars. Defaults to Gmail SMTP (smtp.... Write send_openclaw_message Send a message through any connected channel. Write send_peer_message Send a direct structured message to a peer without routing everything through a central orchestrator. Write set_watchdog_config Configure the background watchdog that continuously monitors system health. Write setup_local_env Discover and diagnose the local development environment. Checks for available API keys, installed SDKs, Nod... Write setup_operator_profile Set up your profile to customize how the AI assistant works for you. Write share_create_packet_link Create a durable local share link record for a packet or founder memo so it can be rendered or synced later. Write smart_select_tools LLM-powered tool selection: sends your task description + a compact tool catalog to a fast model (Gemini 3 ... Write spreadsheet-enrichment-trace Traceable workflow for spreadsheet enrichment: inspect workbook, research supporting evidence, edit cells, ... Write synthesize_extension_plan Synthesize a plan for extending or deepening an existing feature. Write synthesize_feature_plan Synthesize a phased feature implementation plan conditioned on founder context, Write thompson_feynman_edit Skeptical Beginner editor — reviews Thompson-written content against 8 rejection criteria. Returns PASS/REW... Write thompson_write Transform complex content into Thompson Protocol format — plain English mandate, intuition-before-mechanics... Write update_agents_md Read, append, or update sections in the AGENTS.md file. This file contains instructions for AI agents worki... Write update_company_truth Update a subconscious memory block with new information. Write update_forecast_probability Update a forecast Write upsert_durable_object Register or update a durable local object so views, tools, workflows, runs, artifacts, and outcomes share o... Write watchlist_add_entity Add an entity to the local founder watchlist with alert preferences and optional strategic-angle linkage. Write workflowType Optional workflow label such as spreadsheet_enrichment or company_direction_analysis Write write_workspace_file Create or update a file in the agent workspace (~/.nodebench/workspace/). Write xlsx_select_rows Select rows from a local XLSX using deterministic filters. Returns bounded results (selected columns + matc...

DESTRUCTIVE 9 tools

Destructive compress_or_expand_text Precisely compress or expand academic text by a target word count. Compress mode: remove filler words, conv... Destructive delete_learning Delete a learning by key. Use when a learning is outdated or incorrect. Destructive dive_changelog Record a change made to fix a bug, design issue, or improve a component. Links before/after screenshots to ... Destructive founder_local_weekly_reset One-call convenience: gathers all local context and produces a complete Destructive founder_weekly_reset Full weekly reset: discover → load → synthesize → track → milestone Destructive remove_ai_signatures Detect and remove AI-generated writing signatures from academic text. First runs pattern matching for known... Destructive share_revoke_packet_link Revoke a local share link so it no longer counts as active. Destructive track Add, check, remove, or list tracked entities with one workflow tool. The default path optimizes for watched... Destructive unload_toolset Remove a dynamically loaded toolset from the current session to free up context. Cannot unload toolsets fro...

EXECUTE 90 tools

Execute benchmark_models Run the same prompt against multiple LLM providers and compare responses. Returns side-by-side results with... Execute build_banking_packet Build a banker-readiness packet from the canonical company packet. Execute build_before_after_memo Build a memo showing the before and after path plus the validation rationale. Execute build_causal_chain Construct a causal chain from temporal observations. Nodes must be in chronological order. Each node repres... Execute build_claim_graph Extract claims from a source packet and link each claim to its evidence. Returns a directed graph of claims... Execute build_company_packet Build the canonical company readiness packet. Execute build_company_profile_starter Build a starter PitchBook/Crunchbase-like company profile. Execute build_diligence_packet Build a diligence-oriented export payload from the canonical company packet. Execute build_founder_operating_model Build the complete founder operating model: execution order, queue topology, packet routing, source trust p... Execute build_investor_packet Build an investor-oriented export payload from the canonical company packet. Execute build_research_digest Generate a digest of new (unseen) articles from RSS feeds. Compares against previously seen articles via SQ... Execute build_shared_context_subscription Build the exact pull/subscription manifest an agent client should use to watch a packet or packet scope. Execute build_shared_context_subscription_manifest Build a filtered snapshot/events/pull manifest for one peer, packet class, producer, scope, or subject so c... Execute build_slack_onepager Build a Slack-friendly one-page founder report. Execute build_submission_export Build a generic submission export from the canonical company packet. Execute build_temporal_graph Build a temporal relationship graph for an entity. Execute call_driver_tool Invoke a tool on a connected MCP driver. This proxies the call to the external MCP server (e.g. playwright-... Execute call_webmcp_tool Invoke a WebMCP tool on a connected origin. The tool is executed in the browser page context via page.evalu... Execute compare_eval_runs Compare two eval runs to decide whether a change should ship. Returns side-by-side scores and a deploy/reve... Execute compile_decision_packet Compile entity intelligence into a decision-ready packet. Execute compile_environment_spec Generate a simulation environment specification from entity intelligence. Execute compile_scenarios Generate 3-7 future scenario branches for an entity or decision. Execute compile_tension_model Model explicit tensions between forces for a decision or entity. Execute execution-trace-workflow Start and maintain a traceable execution run. Use this for any workflow that needs receipts, evidence, deci... Execute founder_direction_assessment Pressure-test a founder direction against team shape, AI stance, build speed, Execute grade_agent_run Grade a single agent run on both outcome quality (task success, regressions, time) and process quality (rec... Execute graphify_report Get the GRAPH_REPORT.md analysis from a graphify run. Contains god nodes (most connected), Execute gtm_script_builder Build a starter GTM script for the current founder wedge. Execute invoke_openclaw_skill Run an OpenClaw tool safely through security checks. Execute invoke_view_tool Invoke a per-view tool on the current or specified view. Execute judge_tool_output Run the 7-criterion LLM judge on a tool Execute link_durable_objects Create a durable relationship such as screen -> action, workflow -> run, run -> artifact, or outcome -> evi... Execute log_interaction Log and optionally auto-execute an interaction step. If the built-in Playwright browser is active (launched... Execute navigate_to_view Navigate to a specific view in the NodeBench AI frontend. Execute nb_start_agent Start new agent conversation Execute nodebench.research_run Start an adaptive, evidence-backed research run on one or more subjects (companies, people, events, topics)... Execute preconditions_verified Environment, auth, and test data preconditions were verified before the trigger step. Execute primary_mission_preserved The run stayed focused on the reported bug instead of drifting into unrelated exploration. Execute record_dogfood_telemetry Record a full telemetry row for a dogfood run. Captures surface, scenario, user role, prompt, tool usage, t... Execute request_execution_approval Request a human approval gate for a risky execution-trace action. Approval state is written onto the live r... Execute retry_budget_respected Retries were bounded and targeted at the failing trigger or precondition, not the whole workflow. Execute run_autonomous_loop Execute autonomous verification loop with stop conditions. Implements Ralph Wiggum pattern with checkpoints... Execute run_benchmark_batch Run a longitudinal benchmark batch. N=1 is a smoke test (1 founder, 1 session). Execute run_browserstack_benchmark_lane Return a BrowserStack/browser-automation benchmark lane payload. Execute run_closed_loop Track a compile-lint-test-debug closed loop iteration. Record the result of each step. Never present change... Execute run_code_analysis Static analysis on code or text content for security issues, secrets, homograph attacks, ANSI injections, s... Execute run_competitor_signal_benchmark Return a competitor-signal-to-response benchmark lane payload. Execute run_deep_sim Run a multi-agent scenario simulation with bounded branching and budget controls. Instantiates agents with ... Execute run_dogfood_batch_with_judge Execute the priority 3 dogfood scenarios with automatic LLM judge validation. Execute run_entity_intelligence_mission Run a full DeepTrace entity intelligence mission with optional bounded research cell. Unifies relationship ... Execute run_flicker_detection Run full 4-layer Android UI flicker detection pipeline: SurfaceFlinger stats + logcat (L0), screenrecord (L... Execute run_founder_autonomy_benchmark Run the weekly founder reset autonomy benchmark lane. Execute run_graphify Generate a knowledge graph from a folder of code, docs, papers, or images. Execute run_judge_loop Execute a full judge-fix-verify loop: calls a tool, judges the output, and if it fails, Execute run_mandatory_flywheel Enforce the mandatory 6-step AI Flywheel verification after any non-trivial change. All 6 steps must pass b... Execute run_oracle_comparison Compare actual output against a known-good oracle reference. Based on Anthropic\ Execute run_packet_to_implementation_benchmark Return a packet-to-implementation benchmark lane payload. Execute run_quality_gate Evaluate content or code against a set of boolean rules. Returns pass/fail with specific failures listed. T... Execute run_recon Start a reconnaissance research session. Use this at the start of Phase 1 (Context Gathering) to organize r... Execute run_research_cell Run a bounded re-analysis cell for a DeepTrace entity investigation. Queries existing DeepTrace state throu... Execute run_self_directed_delivery_loop Run a local-first autonomous delivery loop across exploratory research, planning, implementation commands, ... Execute run_self_heal Autonomous self-healing for detected drift issues. Fixes orphaned verification cycles Execute run_self_maintenance Run autonomous self-maintenance cycle. Checks TypeScript compilation, documentation sync, tool counts, test... Execute run_signal_sweep Run a live signal sweep across all data sources (HackerNews, GitHub Trending, Yahoo Finance, ProductHunt). ... Execute run_sync_bridge_flush Open the outbound websocket bridge, pair or resume the local device, and flush pending approved operations ... Execute run_tests_cli Execute a shell test command with timeout, capture stdout/stderr, and return structured results. Useful for... Execute run_visual_qa_suite End-to-end visual QA pipeline: burst capture → SSIM stability analysis → Execute sandbox_batch Execute multiple commands, index all outputs, and run multiple search queries — all in ONE call. This is th... Execute sandbox_execute Run a shell command, automatically index the output into the sandbox, and return only a summary. The raw st... Execute scaffold_nodebench_project Create a complete project template pre-configured for nodebench-mcp. Generates: package.json, AGENTS.md, .m... Execute scaffold_research_pipeline Generate a complete, standalone Node.js project for an automated research digest pipeline. Creates: package... Execute scrapling_crawl Start a multi-page spider crawl with extraction. Crawls from start URLs, follows links matching a CSS selec... Execute scrapling_crawl_stop Stop a running crawl session. Pass the session_id from scrapling_crawl. Items collected so far are preserve... Execute self_implement Self-implement missing agent infrastructure. Generates implementation plan and code templates for: agent_lo... Execute simulate_decision_paths Run Monte Carlo simulation for founder decisions. Generates multiple random paths to visualize possible fut... Execute solve_green_polygon_area_from_image Compute the area of a green filled polygon in an image by pixel segmentation, calibrating pixel-to-unit sca... Execute spawn_openclaw_agent Start a secure OpenClaw session with safety rules applied. Execute start_autonomy_benchmark Start an autonomous capability benchmark. Defines a complex build challenge and tracks agent progress throu... Execute start_component_flow Claim a component for traversal by a specific subagent. Marks it as Execute start_dogfood_session Start a new dogfood session for one of the 3 canonical loops (weekly_reset, pre_delegation, company_search)... Execute start_eval_run Start a new eval run. Define the test batch upfront with test cases (input, intent, expected behavior), the... Execute start_execution_run Start a live Convex-backed execution trace run for a workflow. Creates a task session and trace together so... Execute start_ui_dive Initialize a UI/UX Full Dive session. Auto-launches a headless Playwright browser if installed (zero setup)... Execute start_verification_cycle Start a new 6-phase verification cycle for a non-trivial implementation. Returns the cycle ID and Phase 1 i... Execute track_milestone Record a significant milestone (phase complete, deploy, ship, launch, pivot, decision) with optional eviden... Execute trigger_batch_run Run a scheduled task right now instead of waiting. Execute trigger_investigation When an eval run shows regression, trigger a new verification cycle to investigate. This is how the outer l... Execute trigger_verify_split The action that attempts reproduction is separate from the step that verifies the resulting UI or system st... Execute ui-qa-checklist UI/UX QA checklist for frontend implementations. Run after any change that touches React components, layout... Execute workflowTitle Human-readable title for the run

NODEBENCH TOOLS

Let agents act without letting them run wild.