consensus-loop:tools

Run consensus-loop deterministic analysis tools via CLI — code_map, dependency_graph, audit_scan, coverage_map, rtm_parse, rtm_merge, audit_history, fvm_generate, fvm_validate. Use this skill whenever you need codebase analysis (symbol index, dependency DAG, pattern scan, coverage), RTM operations (parse, merge, query), or FVM operations (generate matrix, validate against server) — even if the MCP server is not configured. This skill replaces mcp__plugin_consensus-loop_consensus-loop__* tool calls with equivalent CLI commands.

consensus-loop:tools

CLI interface for the 9 deterministic analysis tools that power the consensus-loop workflow. These tools run via tool-runner.mjs — same logic as the MCP server, but invoked through Bash instead of JSON-RPC.

Tool Runner Path

${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs

Invocation pattern:

node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs <tool_name> --param value ...

Add --json to any command for structured JSON output instead of formatted text.

Tool Selection Guide

NeedToolReference
Find functions/classes/types with line rangescode_mapcode-map.md
Map import/export dependencies, detect cyclesdependency_graphdependency-graph.md
Scan for as any, hardcoded values, console.logaudit_scanaudit-scan.md
Get per-file test coverage percentagescoverage_mapcoverage-map.md
Read RTM rows, filter by req_id or statusrtm_parsertm-parse.md
Merge worktree RTM files into basertm_mergertm-merge.md
Query audit verdict history, detect patternsaudit_historyaudit-history.md
Generate FE×API×BE×Auth verification matrixfvm_generatefvm-generate.md
Execute FVM rows against live serverfvm_validatefvm-validate.md

Workflow

  1. Identify which tool matches the user's need using the table above
  2. Read the corresponding reference file for detailed parameters and examples
  3. Run the tool via node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs
  4. Parse the output (text to stdout, summary to stderr)

Quick Examples

# Symbol index for a directory
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs code_map --path src/

# Dependency graph with cycle detection
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs dependency_graph --path src/ --depth 3

# Pattern scan for type-safety issues
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs audit_scan --pattern type-safety

# Parse RTM, filter by requirement ID
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs rtm_parse --path docs/rtm.md --req_id EV-1

# Audit history summary with risk patterns
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs audit_history --summary

# FVM generation
node ${CLAUDE_PLUGIN_ROOT}/scripts/tool-runner.mjs fvm_generate --path /project/root

Error Handling

  • Exit code 0 = success (text to stdout, summary to stderr)
  • Exit code 1 = error (message to stderr)
  • Missing required params → error message lists what's needed
  • --json flag outputs structured JSON for programmatic use

MCP ↔ CLI Equivalence

Every MCP tool call maps 1:1 to a CLI command:

MCP CallCLI Equivalent
mcp__consensus-loop__code_map({path: "src/"})node tool-runner.mjs code_map --path src/
mcp__consensus-loop__dependency_graph({path: "src/"})node tool-runner.mjs dependency_graph --path src/
mcp__consensus-loop__audit_scan({pattern: "all"})node tool-runner.mjs audit_scan --pattern all
mcp__consensus-loop__rtm_parse({path: "x.md"})node tool-runner.mjs rtm_parse --path x.md

The output format is identical — same text, same summary.