MCP server intelligence profile

preflight MCP Server

An MCP server for Claude Code that catches vague prompts, applies triage and correction pattern learning, and provides semantic search, cross-service contracts, and scorecards to reduce wasted tokens

Local OnlyOfficial distributionpreflight-dev
Verified cleanNpm · 3.2.0

Our scanner tested version 3.2.0 without proving a finding in the methods exercised. This is not a guarantee that every deployment is secure.

1Distribution channel
24Independently observed tools
0Linked remote endpoints
AvailableVersion intelligence

Install and connect

Installation and connection instructions are shown only when supported by retained package, repository, or endpoint evidence.

Install preflight-dev from npm

Version 3.2.0 declares 1 executable entrypoint.

npm install --save-exact preflight-dev@3.2.0
npx -y -p preflight-dev@3.2.0 preflight-dev
MCP client configuration example
{
  "mcpServers": {
    "preflight-dev": {
      "command": "npx",
      "args": [
        "-y",
        "-p",
        "preflight-dev@3.2.0",
        "preflight-dev"
      ]
    }
  }
}

Identity

Canonical slugpreflight-dabe82d6DeploymentLocal Only
Canonical packagenpm:preflight-devRepositorypreflight-dev/preflight
First published—Latest release—
Last security verification—Classification confidence90%
PublicationPublishedOfficial distributionYes

Distributions

ChannelIdentifierCurrent versionVersionsSource
npmpreflight-dev3.2.01Repository

Current release

PackageVersionPublished / observedInventorySecurity scan
npmpreflight-dev3.2.0CurrentSep 5, 202624 toolsSucceeded · 0 resources · 0 promptsFailed
Enterprise protection

Continuously monitor this MCP for security risk

Independently scan the exact version your agents use, receive alerts when its risk changes, and investigate every finding with retained version evidence.

  • Independent exact-version security scans
  • Continuous release and vulnerability monitoring
  • Risk-change alerts with capability context
  • Historical evidence and API exports
Custom pricingContact salesTailored to your organization, integrations, data needs, and support requirements.

Current version evidence

Provenanceartifact_hash_verifiedSignature—
MCP SDK@modelcontextprotocol/sdk Artifact SHA-256b8eaac94a108540b16101d5f156cc08a35405ba516c372ce4237e60220ab9bcc
Scannermcp-proof-engine 0.1.0Scan completedAug 20, 2026
Security rating45 / 100Methodologyversion-rating-1.0
Executable entrypoints
[
  "preflight-dev"
]

Current protocol inventory

2025-06-18Negotiated protocol
preflightServer-reported name
1Capability groups
Aug 17, 2026Observed

Tools 24

ToolCategoryAnnotationsRisk
audit_workspaceAudit workspace documentation freshness vs actual project state. Compares .claude/ workspace docs against recent git commits to find stale or missing documentation. Call after completing a batch of work or at session end.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {}
}
——
check_patternsCheck if the current prompt matches any learned correction patterns from past mistakes. Use this to avoid repeating known pitfalls.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "prompt": {
      "type": "string",
      "description": "The prompt to check against known patterns"
    }
  },
  "required": [
    "prompt"
  ]
}
——
check_session_healthCheck session health and recommend whether to continue, checkpoint, or start fresh. Tracks session depth, uncommitted work, workspace staleness, and time since last commit. Call periodically during long sessions.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "stale_threshold_hours": {
      "description": "Hours before a doc is considered stale. Default: 2",
      "type": "number"
    }
  }
}
——
checkpointSave a session checkpoint before context compaction hits. Commits current work, writes session state to workspace docs, and creates a resumption note. Call this proactively when session is getting long, or when the session-health hook warns about turn count. This is your "save game" before compaction wipes context.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "summary": {
      "type": "string",
      "description": "What was accomplished so far in this session"
    },
    "next_steps": {
      "type": "string",
      "description": "What still needs to be done"
    },
    "current_blockers": {
      "description": "Any issues or blockers encountered",
      "type": "string"
    },
    "commit_mode": {
      "description": "What to commit: 'staged' (only staged files), 'tracked' (modified tracked files), 'all' (git add -A). Default: 'tracked'",
      "type": "string",
      "enum": [
        "staged",
        "tracked",
        "all"
      ]
    }
  },
  "required": [
    "summary",
    "next_steps"
  ]
}
——
clarify_intentClarify a vague user instruction by gathering project context. Call BEFORE executing when the user's prompt is missing specific files, actions, scope, or done conditions. Returns git state, test failures, recent changes, and workspace priorities.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "user_message": {
      "type": "string",
      "description": "The user's raw message/instruction to clarify"
    },
    "suspected_area": {
      "description": "Best guess area: 'tests', 'git', 'ui', 'api', 'schema'",
      "type": "string"
    }
  },
  "required": [
    "user_message"
  ]
}
——
enrich_agent_taskEnrich a vague sub-agent task with project context. Call before spawning a Task/sub-agent to add file paths, patterns, scope boundaries, and done conditions.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "task_description": {
      "type": "string",
      "description": "The raw task for the sub-agent"
    },
    "target_area": {
      "description": "Codebase area: directory path, keyword, or description like 'auth tests', 'api routes'",
      "type": "string"
    }
  },
  "required": [
    "task_description"
  ]
}
——
estimate_costEstimate token usage and cost for the current session. Shows waste from vague prompts and savings from preflight checks.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "session_dir": {
      "description": "Path to session JSONL file (optional, uses latest if omitted)",
      "type": "string"
    },
    "model": {
      "description": "Pricing model to use (claude-sonnet-4, claude-opus-4, claude-haiku-3.5). Default: claude-sonnet-4",
      "type": "string"
    }
  }
}
——
generate_scorecardGenerate a prompt discipline scorecard, trend report, or comparative report. Analyzes sessions across 12 categories with PDF/Markdown output, trend lines, and cross-project comparisons.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "project": {
      "description": "Project name to score. If omitted, scores current project.",
      "type": "string"
    },
    "period": {
      "default": "day",
      "type": "string",
      "enum": [
        "session",
        "day",
        "week",
        "month"
      ]
    },
    "session_id": {
      "description": "Score a specific session by ID",
      "type": "string"
    },
    "since": {
      "description": "Start date (ISO or relative like '7days')",
      "type": "string"
    },
    "output": {
      "default": "markdown",
      "type": "string",
      "enum": [
        "pdf",
        "markdown"
      ]
    },
    "output_path": {
      "description": "Where to save PDF. Default: /tmp/scorecard-{date}.pdf",
      "type": "string"
    },
    "report_type": {
      "default": "scorecard",
      "description": "Type of report: scorecard (default), trend (week/month analysis), or comparative (cross-project)",
      "type": "string",
      "enum": [
        "scorecard",
        "trend",
        "comparative"
      ]
    },
    "compare_projects": {
      "description": "Project names for comparative report",
      "type": "array",
      "items": {
        "type": "string"
      }
    }
  }
}
——
log_correctionLog when the user corrected your action. Tracks error patterns over time to identify what kinds of prompts lead to wrong outputs. Call this whenever the user says "no", "wrong", "not that", "I meant", or otherwise corrects your work.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "what_user_said": {
      "type": "string",
      "description": "The user's correction message"
    },
    "what_you_did_wrong": {
      "type": "string",
      "description": "What you did that was incorrect"
    },
    "root_cause": {
      "type": "string",
      "description": "Why — was it a vague prompt, stale context, wrong assumption, or something else?"
    },
    "category": {
      "type": "string",
      "enum": [
        "vague_prompt",
        "stale_context",
        "wrong_assumption",
        "wrong_file",
        "wrong_scope",
        "misread_intent",
        "incomplete_work",
        "style_mismatch",
        "other"
      ],
      "description": "Error category"
    }
  },
  "required": [
    "what_user_said",
    "what_you_did_wrong",
    "root_cause",
    "category"
  ]
}
——
onboard_projectIndex a project's Claude Code sessions and git history into the timeline database for semantic search and chronological viewing.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "project_dir": {
      "type": "string",
      "description": "Absolute path to the project directory"
    },
    "embedding_provider": {
      "default": "local",
      "type": "string",
      "enum": [
        "local",
        "openai"
      ]
    },
    "openai_api_key": {
      "type": "string"
    },
    "git_depth": {
      "default": "all",
      "type": "string",
      "enum": [
        "all",
        "6months",
        "1year",
        "3months"
      ]
    },
    "git_since": {
      "description": "Override git_depth with exact start date (ISO: '2025-08-01')",
      "type": "string"
    },
    "git_authors": {
      "description": "Filter git commits to these authors. If omitted, auto-detects the primary author (most commits).",
      "type": "array",
      "items": {
        "type": "string"
      }
    },
    "reindex": {
      "default": false,
      "description": "If true, drop existing data and rebuild from scratch",
      "type": "boolean"
    }
  },
  "required": [
    "project_dir"
  ]
}
——
preflight_checkRun a preflight check on your prompt before Claude starts working. Catches vague instructions, surfaces missing context from related services, and builds execution plans for complex tasks.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "prompt": {
      "type": "string",
      "description": "The user's prompt to check"
    },
    "force_level": {
      "description": "Override triage — skip passes through, light does clarify only, full does everything",
      "type": "string",
      "enum": [
        "skip",
        "light",
        "full"
      ]
    }
  },
  "required": [
    "prompt"
  ]
}
——
prompt_scoreScore a prompt on specificity, scope, actionability, and done-condition. Returns a letter grade with specific improvement tips.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "prompt": {
      "type": "string",
      "description": "The prompt text to score"
    }
  },
  "required": [
    "prompt"
  ]
}
——
scan_sessionsLive scan of Claude Code sessions on disk. Shows active and recent sessions, file sizes, and optionally last messages. Detects parallel sessions on the same branch.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "project": {
      "description": "Project name or path. If omitted, scans all projects.",
      "type": "string"
    },
    "active_only": {
      "default": false,
      "description": "Only show sessions modified in last hour",
      "type": "boolean"
    },
    "include_messages": {
      "default": 0,
      "description": "Include last N messages from each session",
      "type": "number"
    },
    "sort": {
      "default": "recent",
      "type": "string",
      "enum": [
        "recent",
        "size",
        "messages"
      ]
    }
  }
}
——
scope_workBreak down a task into a structured execution plan BEFORE writing any code. Call this tool whenever you receive a new task, feature request, or bug report. It analyzes the current project state (git, files, workspace docs) and returns an ordered plan with scope boundaries, complexity estimate, and done conditions. Use this to avoid scope creep and ensure you touch only what's needed.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "task": {
      "type": "string",
      "description": "The raw user task or request to plan"
    },
    "branch": {
      "description": "Git branch to scope against (defaults to current branch)",
      "type": "string"
    }
  },
  "required": [
    "task"
  ]
}
——
search_contractsSearch API contracts, types, and schemas across current and related projects. Fast lookup without vector search.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "What to search for"
    },
    "scope": {
      "default": "all",
      "description": "Which projects to search",
      "type": "string",
      "enum": [
        "current",
        "related",
        "all"
      ]
    },
    "kind": {
      "default": "all",
      "description": "Filter by contract kind",
      "type": "string",
      "enum": [
        "interface",
        "type",
        "enum",
        "route",
        "schema",
        "event",
        "model",
        "all"
      ]
    }
  },
  "required": [
    "query"
  ]
}
——
search_historySemantic search across the unified timeline of prompts, commits, corrections, and tool calls. Find relevant events using natural language queries.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Natural language search query"
    },
    "scope": {
      "default": "current",
      "description": "Search scope: current project, related projects (PREFLIGHT_RELATED), or all indexed projects",
      "type": "string",
      "enum": [
        "current",
        "related",
        "all"
      ]
    },
    "project": {
      "description": "Filter to a specific project name (overrides scope)",
      "type": "string"
    },
    "branch": {
      "type": "string"
    },
    "author": {
      "description": "Filter commits to this author (partial match, case-insensitive)",
      "type": "string"
    },
    "type": {
      "default": "all",
      "type": "string",
      "enum": [
        "prompt",
        "assistant",
        "correction",
        "commit",
        "tool_call",
        "compaction",
        "sub_agent_spawn",
        "error",
        "all"
      ]
    },
    "since": {
      "description": "ISO date or relative: '2025-06-01', '3months'",
      "type": "string"
    },
    "until": {
      "description": "ISO date or relative",
      "type": "string"
    },
    "limit": {
      "default": 10,
      "type": "number"
    }
  },
  "required": [
    "query"
  ]
}
——
sequence_tasksOrder a set of tasks to minimize context switches, reduce re-reads, and batch related work. Supports dependency-order, file-locality, and risk-first strategies. Call when you have multiple tasks to execute in a session.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "tasks": {
      "minItems": 1,
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "Tasks to sequence (natural language descriptions)"
    },
    "strategy": {
      "default": "locality",
      "description": "Sequencing strategy",
      "type": "string",
      "enum": [
        "dependency",
        "locality",
        "risk-first"
      ]
    }
  },
  "required": [
    "tasks"
  ]
}
——
session_handoffGenerate a handoff brief for the next session. Reads last checkpoint, recent commits, open PRs, workspace state, and correction patterns to create a "here's where we are" document. Call at session end or when starting a new session to catch up on what happened.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "direction": {
      "type": "string",
      "enum": [
        "outgoing",
        "incoming"
      ],
      "description": "'outgoing' = ending this session, 'incoming' = starting a new one"
    }
  },
  "required": [
    "direction"
  ]
}
——
session_statsAnalyze Claude Code session history from JSONL files. Returns total sessions, prompts, correction rate, and more. No embeddings needed.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "projectFilter": {
      "description": "Filter to sessions matching this project path substring",
      "type": "string"
    }
  }
}
——
sharpen_followupDetects vague follow-up prompts and sharpens them with specific files, scope, and context from previous actions and git state. Call when the user says things like "fix it", "do the same for the others", "now the tests" without specifying files or scope.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "followup_message": {
      "type": "string",
      "description": "The user's follow-up message to analyze"
    },
    "previous_action": {
      "type": "string",
      "description": "Description of what was just done"
    },
    "previous_files": {
      "description": "Files involved in the previous action",
      "type": "array",
      "items": {
        "type": "string"
      }
    }
  },
  "required": [
    "followup_message",
    "previous_action"
  ]
}
——
timelineChronological view of project events grouped by day. Shows prompts, responses, tool calls, corrections, and commits in order.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "scope": {
      "default": "current",
      "description": "Search scope: current project, related projects (PREFLIGHT_RELATED), or all indexed projects",
      "type": "string",
      "enum": [
        "current",
        "related",
        "all"
      ]
    },
    "project": {
      "description": "Filter to a specific project name (overrides scope)",
      "type": "string"
    },
    "branch": {
      "type": "string"
    },
    "author": {
      "description": "Filter commits to this author (partial match, case-insensitive)",
      "type": "string"
    },
    "since": {
      "type": "string"
    },
    "until": {
      "type": "string"
    },
    "type": {
      "default": "all",
      "type": "string",
      "enum": [
        "prompt",
        "assistant",
        "correction",
        "commit",
        "tool_call",
        "compaction",
        "sub_agent_spawn",
        "error",
        "all"
      ]
    },
    "limit": {
      "default": 50,
      "type": "number"
    },
    "offset": {
      "default": 0,
      "type": "number"
    }
  }
}
——
token_auditDetect token waste patterns in Claude Code sessions — repeated reads, large file dumps, context bloat, skill paste overhead. Call periodically to check efficiency, especially in long sessions.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "session_actions": {
      "description": "Description of recent actions if available",
      "type": "string"
    },
    "check_mode": {
      "default": "quick",
      "description": "Quick checks git/files; deep also analyzes tool call patterns",
      "type": "string",
      "enum": [
        "quick",
        "deep"
      ]
    }
  }
}
——
verify_completionVerify that work is actually complete before declaring done. Runs type check, relevant tests, checks for uncommitted files, and validates against the original task criteria. Call this BEFORE saying "done" or committing final work.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "task_description": {
      "type": "string",
      "description": "What was the task? Used to check if success criteria are met."
    },
    "test_scope": {
      "description": "Which tests to run: 'all', a directory/keyword, or a specific spec file path. Default: auto-detect from changed files.",
      "type": "string"
    },
    "skip_tests": {
      "description": "Skip running tests (only check types + git state). Default: false.",
      "type": "boolean"
    },
    "skip_build": {
      "description": "Skip build check. Default: false.",
      "type": "boolean"
    }
  },
  "required": [
    "task_description"
  ]
}
——
what_changedSummarize what changed recently. Useful after sub-agents finish, after a break, when context was compacted, or at the start of a new session. Returns diff summary with commit messages.
Input schema
{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "since": {
      "description": "Git ref: 'HEAD~5', 'HEAD~3', etc. Default: HEAD~5",
      "type": "string"
    }
  }
}
——

Resources 0

  • None observed.

Resource templates 0

  • None observed.

Prompts 0

  • None observed.

Remote endpoints

EndpointTransportAuthenticationHealthObserved
No verified remote endpoint is linked.

preflight MCP Server questions

How do I install preflight MCP Server?

Install the selected package version with: npm install --save-exact preflight-dev@3.2.0

What tools does preflight MCP Server provide?

preflight MCP Server exposed 24 tools during independent protocol observation, including audit_workspace, check_patterns, check_session_health, checkpoint, clarify_intent, enrich_agent_task, estimate_cost, generate_scorecard, and others.

Is preflight MCP Server secure?

Our scanner tested version 3.2.0 without proving a finding in the methods exercised. This is not a guarantee that every deployment is secure.

Explore related MCP server guides

Curated product and capability guides containing this catalog record.

Official vs Community MCP Servers

Let’s talk about MCP security.

Share your details and our security team will contact you.