Flyto Indexer

Flyto2 Indexer is built for teams that let AI agents touch real code. It gives Codex, Claude Code, Cursor, Windsurf, and other MCP clients a local map of your repository before they rename symbols, edit shared APIs, delete files, or ship a pull request. Grep finds text; Flyto2 Indexer answers the question that matters: what depends on the thing you are changing?

Use it for AI code review, monorepo impact analysis, cross-repo dependency tracking, safe refactors, security taint scanning, release-readiness checks, documentation audits, and GitHub/GitLab merge gates. If you are looking for a local MCP server for AI-assisted development, this is the guardrail layer: it lets the agent search, inspect impact, and verify the work before it says done.

Good fit if you searched for:

• code intelligence MCP server
• AI coding agent impact analysis
• monorepo dependency graph for safe refactors
• GitHub or GitLab merge gate for AI-generated code

Try it in 60 seconds

pip install flyto-indexer
flyto-index setup .
flyto-index verify . --strict

That builds the local index, writes agent instructions, configures MCP for supported clients, and runs the same closed-loop gate Flyto2 uses before shipping AI-assisted code.

Without Flyto2 Indexer

You:    "Rename validateOrder to validate_order"

AI:     *renames the function*
        *greps for "validateOrder"*
        *finds 3 matches in the same project*
        *misses 4 callers in the frontend repo*
        *misses the API endpoint that routes to it*
        *pushes broken code*

With Flyto2 Indexer

You:    "Rename validateOrder to validate_order"

AI:     → impact(target="validateOrder", change_type="rename")

        ⚠️ 7 call sites across 3 projects:
          backend/checkout.py:42     — calls validateOrder()
          backend/api/orders.py:18   — imports validateOrder
          frontend/Cart.vue:55       — calls via useCheckout()
          frontend/QuickBuy.vue:23   — calls via useCheckout()
          mobile/OrderScreen.tsx:67  — API call to /api/validate
          tests/test_orders.py:12    — unit test
          tests/test_api.py:88       — integration test
          Risk: HIGH — 3 projects affected
          Test file: tests/test_orders.py
          Cross-project: 2 other repos affected

        *renames all 7 call sites, updates tests, pushes clean code*

That's the difference. grep finds text. This finds dependencies.

Install

pip install flyto-indexer
flyto-index setup .

That's it. One command does everything:

1. Scans your project and builds the code index
2. Writes CLAUDE.md with tool usage instructions
3. Configures Claude Code MCP settings (~/.claude/settings.json)

Restart Claude Code and start using it. Works with any MCP client — Claude Code, Cursor, Windsurf, etc.

API / MCP Tools

Flyto2 Indexer exposes the tools AI agents need before they make a risky edit:

See the feature guide, MCP integration guide, and source-backed tool reference for every published tool, input schema, compatibility definition, and implementation location.

Usage

Use flyto-index verify as the default gate before an AI agent finishes a code change. It runs the local indexer, validates graph integrity, checks context and impact loops, verifies CI/package/MCP runtime/working-tree closure, and runs the built-in weak scanners without Semgrep, Checkov, or network access.

flyto-index scan . --full
flyto-index verify . --strict
flyto-index verify-workspace /Users/chester/flytohub --project flyto-code --project flyto-engine --project flyto-indexer
flyto-index verify . --save-baseline .flyto-baselines/flyto-indexer.json --json
flyto-index verify . --baseline .flyto-baselines/flyto-indexer.json --regression-only
flyto-index verify-workspace . --changed-only --base origin/main
flyto-index verify . --report verify.sarif --report-format sarif
flyto-index verify-baseline compare . --baseline .flyto-baselines/flyto-indexer.json
flyto-index impact MySharedSymbol --path .
flyto-index context --path . --query "auth routes query keys"
flyto-index secrets . --json
flyto-index taint . --json --max-results 100

For CI, use verify --strict to fail on incomplete graph closure, unresolved impact references, high-risk secret findings, high-risk taint flows, missing agent instructions, incomplete CI gates, generated/high-risk changed files, runtime dependency drift, package manifest drift, suspicious baselines, or MCP tool/runtime drift. Use --baseline with --regression-only when a project has known warnings but new AI-generated regressions must still be blocked.

Product edition policy, commercial packaging, and company release orchestration are intentionally outside this public, reusable code-intelligence package.

Configuration

Flyto2 Indexer works without credentials. Repository policy lives in .flyto-rules.yaml; MCP transport and optional analyzer settings are described in docs/CONFIGURATION.md. The generated docs/reference/configuration.md lists every environment variable, scanner default, and built-in rule source.

Contributing

Pull requests are welcome for scanners, MCP tools, docs, CI gates, and language support. Run the local checks before opening a PR:

Testing

python -m pytest
python -m ruff check src tests
flyto-index verify . --strict

Security issues should be reported to [email protected]. General developer questions can go to [email protected] or GitHub Discussions.

Project-specific verify budgets live in .flyto-rules.yaml under verify:. This stays stdlib-only; the verifier reads a small no-dependency subset:

verify:
  allow_warn: [docs_coverage]
  warn_as_fail: [agent_hygiene, generated_index_ignore, mcp_registry]
  min_docs_score: 60

If your MCP client doesn't use ~/.claude/settings.json, add this to your MCP config:

{
  "mcpServers": {
    "flyto-indexer": {
      "command": "python3",
      "args": ["-m", "flyto_indexer.mcp_server"]
    }
  }
}

Then scan and set up CLAUDE.md separately:

flyto-index scan .
flyto-index setup-claude .

git clone https://github.com/flytohub/flyto-indexer.git
./flyto-indexer/scripts/install-local-cli.sh
flyto-index --version
flyto-index setup .

flyto-index setup . --remove
pip uninstall flyto-indexer

What It Does

Impact Analysis — the core feature

Every tool an AI already has (grep, file read, glob) finds text. None of them answer "what depends on this?"

One call gives you everything — references, blast radius, cross-project impact, and test files:

→ impact(target="useAuth")

  References: 12 across 4 projects
    flyto-cloud:  LoginPage.vue, RegisterPage.vue, AuthGuard.ts, api.ts
    flyto-pro:    vscode_agent/tools.py, middleware/auth.py
    flyto-vscode: ChatHandler.ts, AuthProvider.ts
    flyto-core:   modules/auth/login.py
  Risk: HIGH — shared across 4 projects
  Cross-project: 3 other repos affected
  Test file: tests/test_auth.py

Works with uncommitted changes too:

→ impact(mode="unstaged")

  3 symbols affected by your changes:
    validate_order  — 5 callers, test: tests/test_orders.py
    OrderSchema     — used in 2 API endpoints
    format_receipt  — no callers (safe)

Cross-Language API Tracking

Python backend endpoints automatically linked to TypeScript/Vue frontend callers:

→ structure(focus="apis")

  POST /api/checkout
    Defined in: backend/routes/order.py (create_order)
    Called by:   frontend/Cart.vue, frontend/api/orders.ts
    Call count: 4

Detects FastAPI, Flask, Starlette decorators + fetch(), axios, $http calls.

Code Health & Security

One call audits everything — auto-expands weak dimensions with detailed findings:

→ audit()

  Health: 74/100 (C)

  ⚠️ Security (60/100) — auto-expanded:
    2 critical: hardcoded API keys in config.py, settings.py
    1 high: SQL string concatenation in query.py

  ⚠️ Complexity (65/100) — auto-expanded:
    process_data() — 87 lines, depth=6 → extract sub-functions

  ✓ Dead code (90/100) — passing
  ✓ Documentation (85/100) — passing

  Git hotspots: order.py (42 commits, complexity=8.5)

  Refactoring suggestions:
    [high]   process_data() → extract sub-functions
    [medium] dead_fn() — unreferenced, 45 lines → safe to remove

Taint Analysis — track data flow, not just patterns

AST-based engine that tracks how untrusted data flows from sources to dangerous sinks. Unlike regex pattern matching, it follows variables through assignments, f-strings, and function calls — with sanitizer awareness to eliminate false positives.

→ audit(focus="security")

  Taint flows detected:
    [high] src/api/users.py:42
      SQL injection: request.args.get('id') → cursor.execute(query)
      Flow: request.args.get('id') → user_id → query (f-string) → cursor.execute()
      Fix: Use parameterized query: cursor.execute(sql, params)

    [critical] src/api/admin.py:18
      RCE: request.form.get('code') → eval(code)
      Fix: Never eval() user-controlled strings

• Python: Full AST analysis — tracks taint through assignments, f-strings, concat, for loops
• Cross-function: Detects when tainted data is passed as an argument to a function with a dangerous sink
• JS/TS/Go: Regex-based fallback for common taint patterns
• Sanitizer-aware: int(), html.escape(), shlex.quote(), parameterized queries all break the taint chain
• Project DSL: Declare custom sources / sinks / sanitizers in .flyto-rules.yaml under taint: (merged on top of built-in defaults; taint_rules.yaml is also still honored for backward compat)

# .flyto-rules.yaml
taint:
  sources:
    - pattern: "ctx.payload"
      language: python
      taint_type: user_input
  sinks:
    - pattern: "dangerousEval("
      vuln_type: rce
      severity: critical
      recommendation: "Use sandbox runner"
  sanitizers:
    - pattern: "safe_html("
      cleanses: ["xss"]

flyto-index add-taint-source . --pattern "ctx.payload" --taint-type user_input
flyto-index add-taint-sink   . --pattern "dangerousEval(" --vuln-type rce --severity critical
flyto-index list-taint-rules .

Project Rules — AI learns from your corrections

.flyto-rules.yaml — structured, versionable project conventions that audit enforces automatically.

# .flyto-rules.yaml
architecture:
  - rule: "i18n files must be in flyto-i18n/"
    glob_deny: ["flyto-cloud/**/*.locale.json"]

style:
  - rule: "Frontend does no data processing"
    grep_deny: [{ pattern: '\breduce\s*\(', glob: "*.vue" }]

conventions:
  - rule: "Commit messages in English"

When you correct the AI ("don't put i18n files there"), it auto-writes a verifiable rule — so the mistake never happens again, for any AI, any tool:

User corrects AI → add_rule() writes .flyto-rules.yaml → audit checks compliance

• glob_deny — files in wrong locations
• grep_deny — forbidden code patterns in specific file types
• conventions — text-only guidance (no automated check)
• Rules accumulate over time — no upfront config needed

Architecture Layers — declare who may import whom

Declarative layer membership + import constraints. The indexer walks the import graph (Python / TS / JS / Vue / Go) and flags every edge that crosses a forbidden boundary. No plugin, no runtime — just .flyto-rules.yaml and audit.

layers:
  - name: ui
    paths: ["src/pages/**", "src/components/**"]
    can_import: [lib, hooks, types]
    reason: "UI is the top layer"

  - name: lib
    paths: ["src/lib/**"]
    cannot_import: [ui]
    reason: "lib must be UI-agnostic"

  - name: db
    paths: ["src/db/**"]
    can_import: [types]

cross_imports_deny:
  - from: "src/features/a/**"
    to:   "src/features/b/**"
    reason: "features must not cross-import — use shared/"

flyto-index layers .                      # human-readable report
flyto-index layers . --json --fail-on-violation   # CI gate (exits non-zero)
flyto-index add-layer --name ui --paths "src/ui/**" --cannot-import db

• can_import — whitelist (only these layers + self allowed)
• cannot_import — blacklist (overrides the whitelist)
• Path aliases from tsconfig.json paths and Go module paths from go.mod are resolved automatically
• audit picks up layer violations with no extra flag

Task Analysis — plan before you code

Scores risk across 6 dimensions and generates an execution plan:

→ task(action="plan", description="Rename validateOrder to validate_order", intent="refactor")

  Dimensions:
    blast_radius:      HIGH (8.0)  — 7 callers across 3 projects
    breaking_risk:     HIGH (7.0)  — public API, used by external consumers
    test_risk:         MEDIUM (5.0) — 2/7 callers have test coverage
    cross_coupling:    HIGH (8.0)  — referenced in 3 projects
    complexity:        LOW (2.0)   — straightforward rename
    rollback_difficulty: MEDIUM (5.0) — multi-project change

  Execution Plan:
    1. scope_callers       → find_references("validateOrder")
    2. verify_test_coverage → find_test_file("checkout.py")
    3. check_cross_project → cross_project_impact("validateOrder")
    4. ⛔ gate_before_plan → task_gate_check(phase="plan")
    5. preview_changes     → edit_impact_preview("validateOrder", "rename")
    6. ⛔ gate_before_apply → task_gate_check(phase="apply")

Each step has pre-filled arguments — AI follows the data structure, not prompts. Server-side enforcement blocks skipping gates.

Tools

Five core tools cover the normal inspect-before-edit workflow. Focused smart tools add verification, security, licensing, documentation, framework, call hierarchy, PR-risk, and repository-policy operations without forcing clients to work through the granular compatibility surface.

search replaces: search_code, semantic_search, fulltext_search, get_file_info, get_file_symbols, get_symbol_content, get_file_context

impact replaces: find_references, impact_analysis, batch_impact_analysis, edit_impact_preview, cross_project_impact, impact_from_diff

audit replaces: code_health_score, security_scan, taint_analysis, rules_check, find_dead_code, find_complex_functions, find_duplicates, suggest_refactoring, find_stale_files, find_todos, coverage_gaps

task replaces: analyze_task, task_gate_check, validate_changes

structure replaces: list_projects, list_apis, list_categories, dependency_graph, check_api_contracts, contract_drift, extract_type_schema

All granular compatibility tools remain available in dispatch for backward compatibility and execution-plan steps. The generated MCP reference records the current counts and schemas directly from the registries.

Languages

How It Works

flyto-index scan .

1. Parse — AST (Python) or regex (others) extracts every function, class, and import
2. Graph — Builds dependency graph + reverse index (caller → callee)
3. Serve — MCP server answers queries from the graph in memory
4. Incremental — Re-scans only changed files, incrementally patches reverse_index and BM25 (10-50x faster than full rebuild)
5. LSP — Optional type-aware references via pyright/tsserver/gopls/rust-analyzer (zero deps, graceful fallback)

.flyto-index/
├── index.json       # Symbols + dependency graph + reverse index
├── content.jsonl    # Source code (lazy-loaded)
├── bm25.json        # BM25 keyword search index
├── semantic.json    # TF-IDF + learned ConceptGraph (v2.2+)
└── manifest.json    # Change tracking

CI: Block Risky Changes

# Fail the PR if changes affect too many call sites
- run: pip install flyto-indexer
- run: flyto-index verify . --full-scan --strict
- run: flyto-index check . --threshold medium --base main

CLI

flyto-index setup .                       # One command: scan + CLAUDE.md + MCP config
flyto-index scan .                        # Index (or re-index)
flyto-index impact useAuth --path .       # Impact analysis
flyto-index check . --threshold medium    # CI gate
flyto-index demo .                        # 30-second demo
flyto-index install-hook .                # Auto-reindex on commit
flyto-index setup . --remove              # Uninstall

Privacy

Core indexing, search, impact, policy, and verification run locally and require no hosted service. Optional LSP enrichment starts local language servers. The optional LLM auditor can send selected analysis input to its configured provider and is disabled unless explicitly configured. Delete .flyto-index/ to remove generated index data. See the security model.

Limitations

• Static analysis only — dynamic imports and metaprogramming not tracked
• No type inference — complex TypeScript generics simplified
• Cross-project tracking requires all projects indexed together

License

Apache License 2.0. See NOTICE for attribution.