CodeSift -- Token-efficient code intelligence for AI agents

CodeSift indexes your codebase with tree-sitter AST parsing and gives AI agents 150 MCP tools (55 core + 95 discoverable) via CLI or MCP server. It uses 61-95% fewer tokens than raw grep/Read workflows on typical code navigation tasks.

Works with: Claude Code, Cursor, Codex, Gemini CLI, Zed, Aider, Continue — any MCP client.

Install

Bulletproof one-liner (clears stale cache, installs latest, auto-configures all platforms):

npm cache clean --force && npm i -g codesift-mcp@latest

npm cache clean --force clears stale registry metadata that can cause ETARGET errors. The postinstall script then runs codesift setup all automatically.

Restart your AI client (close + reopen) so the new MCP server is picked up. New terminal sessions in your IDE work fine — no need to quit the IDE itself.

To configure individual platforms manually:

codesift setup claude    # Claude Code — config + rules + hooks + CLAUDE.md
codesift setup codex     # Codex CLI — config + AGENTS.md rules
codesift setup cursor    # Cursor IDE — config + .cursor/rules
codesift setup gemini    # Gemini CLI — config + GEMINI.md rules
codesift setup antigravity # Google Antigravity — config only
codesift setup all       # All platforms at once

Verify installed version:

codesift --version

What setup installs (all by default):

Additionally, every MCP client receives ~800 tokens of compact guidance automatically via the MCP instructions field — zero setup needed.

Related MCP server: astllm-mcp

Update

npm update -g codesift-mcp
codesift setup all              # Updates rules files to latest version
codesift setup all --force      # Force-update even if you modified rules

setup writes the absolute codesift-mcp bin path when available. This avoids per-session npx cache drift and stale dist/server.js paths. Re-run setup after updating to refresh rules files and repair old MCP config entries.

Quick start

# Index a project
codesift index /path/to/project

# Search for a function
codesift symbols local/my-project "createUser" --kind function --include-source

# Semantic search (requires embedding provider)
codesift retrieve local/my-project \
  --queries '[{"type":"semantic","query":"how does caching work?"}]'

# Build the project wiki — agents read it on every new session
codesift wiki-generate

Wiki tip: setup --hooks (on by default) wires a SessionStart hook that injects a project overview into every new agent session, plus a PostToolUse hook that auto-refreshes it. Both are silent no-ops until you run codesift wiki-generate once in the repo — skip it and the wiki never reaches your agent. This is the #1 reason wiki adoption stays at zero.

Benchmark results

Combo benchmark (real-world tool sequences)

772 real tasks from usage.jsonl — exact query sequences agents used across 33+ repos. Native (grep/find/read) vs CodeSift.

Per-tool (single-tool benchmark)

Performance features

CLI commands

Indexing

Search

Outline

Symbol retrieval

Graph & analysis

Code analysis

Wiki & Lens

Output goes to .codesift/wiki/ in the repo root. Includes markdown pages with [[wikilinks]], backlinks, community summaries, and a self-contained codesift-lens.html with D3 chord diagram and force-directed graph.

Cross-repo

Diff

Batch & utility

MCP tools (146 total — 55 core + 95 discoverable)

When running as an MCP server, CodeSift exposes 51 core tools directly. The remaining 95 niche tools are discoverable via discover_tools and describe_tools, or via plan_turn which routes a natural-language task to the best-fit tools and auto-reveals any hidden ones.

Conversation search

Search past Claude Code conversation history — the decisions, rationale, and debugging sessions that shaped your code.

# Index conversations for current project (auto-detected from cwd)
# Also runs automatically at startup via auto-discovery
index_conversations()

# Index a specific project's conversations
index_conversations(project_path="/Users/me/.claude/projects/-Users-me-DEV-my-project")

# Search past conversations
search_conversations(query="auth middleware bug", limit=5)

# Find conversations that discussed a specific code symbol
find_conversations_for_symbol(symbol_name="processPayment", repo="local/my-project")

# In codebase_retrieval batch queries
codebase_retrieval(repo, queries=[
  {"type": "semantic", "query": "how does auth work"},
  {"type": "conversation", "query": "why we chose Redis over Postgres cache"}
])

Features:

• Auto-discovery at startup (zero config)

• Session-end hook for immediate re-indexing

• Noise filtering: tool_result dumps stripped, tool_use truncated, images → [image]

• Compaction-aware: skips summary injections, indexes last summary as meta-doc

• Cross-reference: link code symbols to the conversations that discussed them

Secret scanning

Detect hardcoded secrets (API keys, JWT tokens, passwords, connection strings) in your indexed codebase. Uses ~1,100 detection rules from TruffleHog via @sanity-labs/secret-scan, with CodeSift's tree-sitter AST for false-positive reduction.

# Scan entire repo for secrets
scan_secrets(repo="local/my-project")

# Filter by severity
scan_secrets(repo="local/my-project", severity="critical")

# Only high-confidence findings, including test files
scan_secrets(repo="local/my-project", min_confidence="high", exclude_tests=false)

# Scope to specific directory
scan_secrets(repo="local/my-project", file_pattern="src/config/**")

Features:

• Eager scanning on file change — results are cached and instant on query

• AST-aware confidence: test files, docs, placeholder variables auto-demoted to low

• Masked output — secrets shown as sk-p***hijk, raw values never in cache or logs

• Inline allowlist — add // codesift:allow-secret to suppress a finding

• Config files indexed — .env, .yaml, .toml, .json, .ini, .properties scanned

• Severity mapping: cloud keys (AWS, GCP) = critical, API keys (OpenAI, GitHub) = high

• Inline warnings in index_file responses when secrets detected

Wiki & Lens — auto-generated architecture documentation

Generate browsable wiki pages and an interactive HTML dashboard from your codebase's topology — zero manual writing.

# Generate wiki for current repo
codesift wiki-generate

# Scope to a directory
codesift wiki-generate --focus src/tools

# Check wiki integrity
codesift wiki-lint .codesift/wiki

What it generates (in .codesift/wiki/):

• Community pages — one per Louvain community (module), with members, cohesion score, cross-boundary edges

• Hubs page — top symbols by fan-in (load-bearing code)

• Surprises page — unexpected cross-community connections (structural + temporal coupling)

• Hotspots page — files ranked by git churn × complexity

• Framework pages — conditional pages for Next.js routes, Hono middleware, Astro islands (when detected)

• Index page — links to all pages with [[wikilinks]] and auto-generated backlinks

• Summaries — compact *.summary.md files (~400 tokens) for AI agent context injection

Lens HTML dashboard (codesift-lens.html):

• Self-contained single HTML file — open in any browser, no server needed

• D3 chord diagram showing cross-community connections

• D3 force-directed graph with community nodes

• 5 tabs: Overview, Communities, Hubs, Surprises, Wiki browser

• Dark/light theme, responsive

AI agent integration:

• Hook inject via handlePrecheckRead — when an agent reads a file, it automatically receives the file's community wiki summary as context

• Configurable token budget (2000 chars default)

• Staleness detection — warns when wiki is outdated vs current index

MCP tool:

generate_wiki(repo, focus?, output_dir?, include_lens?)

PHP / Yii2 support

Full PHP code intelligence with first-on-market Yii2 framework awareness. No other general-purpose MCP tool provides static Yii2 intelligence.

Symbol extraction (tree-sitter-based):

• Namespaces, classes, interfaces, traits, enums (PHP 8.1), functions, methods, properties, constants

• PHPDoc extraction, signature extraction with type hints and return types

• PHPUnit test detection: TestCase subclass = test_suite, test* methods = test_case, setUp/tearDown = test_hook

Yii2 framework awareness:

• Convention routing: trace_route("site/index") resolves to SiteController::actionIndex() (incl. module nesting)

• analyze_project detects Yii2 via composer.json and extracts: controllers, models, modules, widgets, behaviors, components, assets, config files

• 6 PHP-specific tools: namespace resolution (PSR-4), event/listener tracing, view mapping, service locator resolution, security scanning, project audit (meta-tool with ActiveRecord analysis, N+1 detection, god-model detection via checks= parameter)

• Auto-load: PHP tools are automatically enabled when composer.json is detected at CWD — no need to call discover_tools/describe_tools first

Laravel support:

• Route tracing via Route::get('/path', [Controller::class, 'method']) pattern matching

• Convention extraction: controllers, middleware, models, routes, migrations

# Trace a Yii2 route
trace_route(repo, path="site/about")

# Analyze ActiveRecord models (via php_project_audit)
php_project_audit(repo, checks=["activerecord"], model_name="User")

# PHP security scan (8 parallel checks)
php_security_scan(repo)

# Resolve PSR-4 namespace to file
resolve_php_namespace(repo, class_name="App\\Models\\User")

LSP bridge: Intelephense configured for go-to-definition, find-references, type-info, and rename across PHP files.

Next.js intelligence

Deep Next.js static analysis — 3 core tools covering routing, rendering, security, SEO, and architecture:

• framework_audit — one-call meta-audit: runs route map + metadata + server actions + boundary + data flow + middleware + component classification checks. Returns composite score with prioritized findings. Use checks= parameter to run individual checks (e.g., checks=["server-actions"], checks=["boundary"], checks=["link-integrity"], checks=["data-flow"], checks=["middleware"], checks=["components"], checks=["api-contract"])

• nextjs_route_map — maps all App Router and Pages Router routes with rendering strategy (SSG/SSR/ISR/PPR), dynamic params, route groups, parallel routes, and intercepting routes

• nextjs_metadata_audit — detects missing/incomplete metadata exports, OpenGraph gaps, missing robots/sitemap, and SEO anti-patterns across all routes

Auto-load: Next.js tools are automatically enabled when next is detected in package.json — no manual discovery needed.

When to use CodeSift vs grep

Built-in anti-patterns (33 total)

The patterns command searches for common code quality issues across your codebase:

Custom regex is also supported: codesift patterns local/project "Promise<.*any>".

Performance anti-patterns (find_perf_hotspots)

A separate tool scans for performance-specific issues with balanced-brace loop body extraction (not just regex):

# Scan all patterns
find_perf_hotspots(repo)

# Only N+1 and unbounded queries
find_perf_hotspots(repo, patterns="n-plus-one,unbounded-query")

# Scope to API directory
find_perf_hotspots(repo, file_pattern="src/api")

MCP server setup

CodeSift runs as an MCP server, exposing 146 tools to AI agents (55 core + 95 discoverable). The fastest setup method is codesift setup <platform> which handles everything automatically. Manual configuration is also supported:

OpenAI Codex

Add this to ~/.codex/config.toml:

[mcp_servers.codesift]
command = "/Users/you/.npm-global/bin/codesift-mcp"
args = []
tool_timeout_sec = 120
default_tools_approval_mode = "auto"

You can also add it manually or via the Codex CLI:

codex mcp add codesift -- /Users/you/.npm-global/bin/codesift-mcp

Claude Code

Add this to ~/.claude/settings.json:

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": []
    }
  }
}

With semantic search (OpenAI embeddings), add the env var manually:

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": [],
      "env": {
        "CODESIFT_OPENAI_API_KEY": "sk-..."
      }
    }
  }
}

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": []
    }
  }
}

Cursor

Add this to ~/.cursor/mcp.json, or to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": []
    }
  }
}

Gemini CLI

Add this to ~/.gemini/settings.json, or to .gemini/settings.json in your project:

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": []
    }
  }
}

You can also use the Gemini CLI:

gemini mcp add codesift -s user /Users/you/.npm-global/bin/codesift-mcp

Google Antigravity

Add this to ~/.gemini/antigravity/mcp_config.json:

{
  "mcpServers": {
    "codesift": {
      "command": "/Users/you/.npm-global/bin/codesift-mcp",
      "args": []
    }
  }
}

All platforms at once

codesift setup all

This configures Codex, Claude Code, Cursor, Gemini CLI, and Antigravity in one command. Safe to run multiple times — skips platforms that are already configured.

Semantic search

Semantic search uses embeddings to answer concept queries like "how does authentication work?" that keyword search misses.

Setup

Zero config — semantic search works out of the box. CodeSift defaults to local on-device embeddings (nomic-ai/nomic-embed-text-v1.5 via @huggingface/transformers v3, INT8 ONNX, ~140MB downloaded on first use, cached after). No API key, no internet after first run, no data leaves your machine. The provider applies the model's task-aware prefixes (search_document: / search_query:) automatically, so retrieval quality matches remote providers.

To opt into a remote provider for higher quality, set one of these:

To disable local embeddings entirely (BM25-only), set CODESIFT_DISABLE_LOCAL_EMBEDDINGS=true. To pin a different local model, set CODESIFT_LOCAL_MODEL=<owner>/<model> (e.g. Xenova/bge-small-en-v1.5).

Usage

# Pure semantic search
codesift retrieve local/my-project \
  --queries '[{"type":"semantic","query":"error handling and retry logic","top_k":10}]'

# Hybrid search (semantic + BM25 text, RRF-merged)
codesift retrieve local/my-project \
  --queries '[{"type":"hybrid","query":"caching strategy","top_k":10}]'

Semantic and hybrid queries exclude test files by default to maximize token efficiency. To include test files, set "exclude_tests": false in the sub-query or pass --exclude-tests=false on the CLI.

Configuration

All configuration is via environment variables.

How it works

1. Indexing -- Tree-sitter WASM grammars parse source files into ASTs. Symbol extraction produces functions, classes, methods, types, constants, etc. with signatures, docstrings, and source code. Filesystem mtime is stored per file for incremental skip on reindex.

2. BM25F search -- Symbols are tokenized (camelCase/snake_case splitting) and indexed with field-weighted BM25 scoring. Name matches rank 5x higher than body matches. Symbols in frequently-imported files get a log-scaled centrality bonus as tiebreaker.

3. Semantic search (optional) -- Source code is chunked and embedded via the configured provider. Queries are embedded at search time and ranked by cosine similarity. Multi-sub-query decomposition with Reciprocal Rank Fusion (RRF, k=60).

4. Hybrid search -- Combines semantic embedding similarity with BM25 text matches via RRF, getting the best of both keyword and concept search.

5. File watcher -- chokidar watches indexed folders for changes. Modified files are re-parsed and the index is updated incrementally.

6. Response guards -- Multiple layers prevent token waste: progressive cascade (>15K tok → compact, >25K → counts, >30K → truncate), response dedup cache (30s), in-flight request coalescing, H1-H9 sequential hints, and source truncation.

7. Agent onboarding -- MCP instructions field sends ~800 tokens of guidance (tool discovery, hints, ALWAYS/NEVER rules) to every client automatically. codesift setup installs full rules files per platform + Claude Code hooks for enforcement.

8. LSP bridge (optional) -- When a language server is installed (typescript-language-server, pylsp, gopls, rust-analyzer, kotlin-language-server, solargraph, intelephense), CodeSift uses it for type-safe find_references, precise go_to_definition, get_type_info via hover, and cross-file rename_symbol. Falls back to tree-sitter/grep when LSP is unavailable. Lazy start + 5 min idle kill — zero overhead when not used.

Glob pattern support

File pattern parameters (file_pattern) support full glob syntax via picomatch:

• *.ts — match by extension at any depth

• *.{ts,tsx} — brace expansion

• src/**/*.service.ts — directory globbing

• [!.]*.ts — character classes

• service — plain substring match (no glob chars)

React workflow with CodeSift

CodeSift auto-loads 6 React tools when a React project is detected (package.json with react + .tsx/.jsx files). Zero config.

Day 1 — new React codebase (1 command, ~5s)

react_quickstart

One call returns: component/hook counts, stack (state mgmt, routing, UI lib, form lib, build tool), critical pattern violations, top hooks used, and suggested next queries. Replaces 5+ manual exploration calls.

Daily development

analyze_renders("MyComponent")          # re-render risk for a specific component
trace_component_tree("App")             # JSX composition hierarchy
analyze_hooks(component_name="Foo")     # hook inventory + Rule of Hooks check
trace_call_chain("useAuth", filter_react_hooks=true)  # hook dependency graph, stdlib filtered
find_references("UserContext")          # where this context is consumed
analyze_context_graph                   # all createContext → Provider → useContext flows

PR review

review_diff                             # 10-check composite (React patterns auto-skipped on non-.tsx diffs)
changed_symbols(since="HEAD~3")         # what changed structurally
search_patterns("hook-in-condition")    # Rule of Hooks violations in changed files
impact_analysis(since="HEAD~3")         # blast radius of your changes

CI gates (via audit_scan REACT gate + audit_compiler_readiness)

audit_scan                              # includes REACT gate: hook-in-condition, useEffect-async,
                                        # dangerously-set-html, index-as-key, nested-component-def
audit_compiler_readiness                # React Compiler (v1.0) adoption score — flags bailout
                                        # patterns before migration, counts redundant memo to remove

Set CI to fail on: any dangerously-set-html, any Rule of Hooks violation, any useEffect-missing-cleanup in new code.

Common queries — "how do I..."

Supported languages

TypeScript, JavaScript (JSX/TSX), Python, Go, Rust, Kotlin, Java, Ruby, PHP, Markdown, CSS, Prisma, Astro.

React/JSX/TSX has first-class support across 8 waves: component and hook SymbolKind values, JSX-aware call graph (all graph tools see <Component> usage as call edges), 43 React anti-patterns with engine-level comment/string preprocessing. Tier 8 (May 2026) added preprocess: "strip-comments-strings" declarative field on BUILTIN_PATTERNS entries — single-pass 7-state-machine source stripper at src/utils/source-stripper.ts strips comments, string/template/regex literals before regex match (closes the false-positive class where comment-embedded mentions spoofed detection). Tier 7 (May 2026) fixed 3 pre-existing CRITICAL bugs (useOptimistic lookahead trivial bypass, useEffect-setstate-loop array-arg false positive, react19-server-action-not-async missing arrow/default-export forms) and added cross-file findSuspenseAncestor + findLazyComponentsWithoutSuspense walkers reusing reverse JSX adjacency from Tier 5. Tier 6 (May 2026) — derived-state-reducer (useReducer sync action), derived-state-custom-setter (custom setter naming), stale-closure-toggle (setX(!X)), stale-closure-broken-functional (setX(prev => X+1) wrong reference), context-provider-value-via-variable (intermediate-var inline), context-provider-value-inline-destructured ({Provider} form), react-lazy-no-suspense-same-file (single-file heuristic), rsc-non-serializable-prop-deep (Map/Set/Class across RSC boundary), error-boundary-incomplete (partial lifecycle) + full severity migration on all 29 prior patterns; multiline hook-in-condition, bug-free nested-component-def, and Tier 5 (May 2026) patterns — derived-state (useState(props.X) + useEffect sync), stale-closure-setstate (setX(X+1) non-functional update), context-provider-value-inline (inline object/array forces consumer re-render), jsx-no-target-blank (tabnabbing security with postFilter validator), button-no-type (implicit submit foot-gun, lookahead-bounded for HTML only). trace_component_tree (BFS JSX composition tree), analyze_hooks (hook inventory + Rule of Hooks violation detection), analyze_renders (re-render risk + prop_chain_depth render-tree depth metric with explicit "NOT prop-drilling depth" disclaimer in suggestion text — semantic prop-flow tracking is Tier 6 scope), buildContextGraph (createContext → Provider → useContext consumer mapping), React complexity metrics, enriched get_context_bundle, filter_react_hooks option on trace_call_chain, audit_scan REACT gate, React-aware review_diff, generate_report React section, route entry point detection, shadcn/ui + Tailwind + form library detection, @/ alias resolution, RSC boundary detection, build tool detection (Vite, CRA, webpack, Parcel, esbuild, Rspack, Rsbuild, Turbopack), severity-aware react_quickstart bucketing into critical_issues / warnings / style_issues, and declarative postFilter field on BUILTIN_PATTERNS entries. Auto-loaded on React projects (package.json + .tsx files).

Astro has deep framework intelligence — the first and only static code intelligence for Astro in the MCP ecosystem. 4 dedicated tools: astro_analyze_islands (detect all client:*/server:defer directives, group by framework, track server islands), astro_hydration_audit (12 anti-pattern detectors AH01-AH12 with A/B/C/D scoring — catches client:load on Astro components, islands in loops, missing framework hints, below-fold eager hydration, and more), astro_route_map (file-based routing analysis with dynamic params, route conflicts, rendering mode per page, endpoint method detection), astro_config_analyze (tree-sitter AST walker for astro.config.mjs — extracts output mode, adapter, integrations, i18n, redirects with config_resolution honesty field). Also: 6 Astro anti-patterns in search_patterns, Astro-aware trace_route, analyze_project returns full astro_conventions, .astro extension normalization in import graph, framework detection for dead-code analysis, .mdx file indexing. Template parser (parseAstroTemplate) extracts islands, slots, component usages, and directives from HTML template section with balanced-brace tracking, conditional/loop detection, and landmark section awareness.

Kotlin support includes full tree-sitter parsing with a dedicated extractor for functions, classes (data/sealed/enum/abstract/annotation), interfaces, objects (singleton + companion), properties (val/var/const), type aliases, extension functions, suspend functions, generics, KDoc comments, and JUnit test detection (@Test, @BeforeEach, @AfterEach, @BeforeAll, @AfterAll). Route tracing supports Ktor DSL and Spring Boot Kotlin. Six Kotlin anti-patterns are built-in. | PHP/Yii2 support | src/parser/extractors/php.ts (+ PHPDoc @property/@method synthesis), src/tools/php-tools.ts (6 tools: resolve_php_namespace, trace_php_event, find_php_views, resolve_php_service, php_security_scan, php_project_audit), src/tools/project-tools.ts (Yii2Conventions), src/tools/route-tools.ts (findYii2Handlers, findLaravelHandlers), src/tools/pattern-tools.ts (8 PHP anti-patterns), src/tools/graph-tools.ts (PHP method call detection), src/utils/import-graph.ts (PHP require/include + PSR-4 cross-file edges via resolvePhpNamespace), src/utils/walk.ts (BACKUP_FILE_PATTERNS auto-exclusion), src/parser/parser-manager.ts (error recovery try/catch), src/lsp/lsp-servers.ts (Intelephense), scripts/download-wasm.ts (tree-sitter-php@0.23.12) |

Development

git clone https://github.com/greglas75/codesift.git
cd codesift-mcp
npm install
npm run download-wasm   # Download tree-sitter WASM grammars
npm run build           # TypeScript compilation
npm test                # Run tests (Vitest, 2900+ tests)
npm run test:coverage   # Coverage report
npm run lint            # Type check (tsc --noEmit)

Publishing a new version

After making changes, follow these steps to publish to npm:

# 1. Ensure clean working tree
git status              # No uncommitted changes

# 2. Build and verify
npm run build           # Must succeed with 0 errors
npm test                # Must pass (flaky ast-query tests may fail in full suite — OK if they pass individually)

# 3. Bump version (choose one)
npm version patch       # 0.2.0 → 0.2.1 (bug fixes)
npm version minor       # 0.2.0 → 0.3.0 (new features)
npm version major       # 0.2.0 → 1.0.0 (breaking changes)
# This creates a git commit + tag automatically

# 4. Publish to npm
npm publish --ignore-scripts
# npm will open browser for WebAuthn/Keychain authentication
# Press Enter, confirm in browser, done

# 5. Push to GitHub (commit + tag)
git push && git push --tags

What gets published

The files field in package.json controls what ships:

• dist/ — compiled JavaScript

• rules/ — platform-specific agent rules (codesift.md, codesift.mdc, codex.md, gemini.md)

• src/parser/languages/ — tree-sitter WASM grammars

• README.md, LICENSE

After publishing

Users update with:

npm update -g codesift-mcp        # Update package
codesift setup all                 # Update rules files to latest version

codesift setup all rewrites old npx / dist/server.js MCP entries to the current absolute codesift-mcp bin path.

Checklist before publishing

• npm run build — 0 TypeScript errors

• npm test — 2900+ tests pass

• rules/codesift.md updated if hints or tools changed

• src/instructions.ts updated if rules changed (compact version)

• README.md updated if features added

• CLAUDE.md updated if architecture changed

• Version bumped via npm version

• Changes committed and pushed to GitHub

License

BSL-1.1

Low-memory / multi-session (teams, 16–24GB machines)

Each editor session (Claude Code/Cursor/Codex window) can spawn its own stdio codesift server process, and each loads per-repo embeddings into its own RAM. On big repos (GB-scale embedding files) several windows can exhaust memory on smaller machines. Controls:

# Prefer one shared daemon for many editor windows.
codesift serve
codesift setup all --http

# Clean up stale helper MCP processes left by closed sessions.
codesift cleanup-processes
codesift cleanup-processes --dry-run

Set these in the env block of your MCP client config (e.g. mcpServers.codesift.env). The shared HTTP daemon (codesift serve) loads embeddings once for all windows and avoids per-session duplication.