LeanCTX — Context Runtime for AI Agents

LeanCTX optimizes LLM context through 58 MCP tools, 95+ shell compression patterns, and tree-sitter AST parsing for 18 languages. It provides adaptive file reading, cross-session memory (CCP), task-conditioned relevance scoring, and a feedback loop for learning optimal compression.

Setup (run first)

Before using lean-ctx, verify it is installed:

which lean-ctx || bash scripts/install.sh

If the install script is not available locally, install manually:

curl -fsSL https://raw.githubusercontent.com/yvgude/lean-ctx/main/skills/lean-ctx/scripts/install.sh | bash

After installation, run the one-command setup (installs shell hook + editor wiring + rules + skills):

lean-ctx setup

lean-ctx supports two integration styles:

• CLI-redirect (preferred when shell access exists): no MCP tool schema overhead.
• MCP (required for some IDE extensions): cached reads + tools via MCP.

When to use lean-ctx

Always prefer lean-ctx -c <command> over running commands directly when:

• The command produces verbose output (build logs, git diffs, dependency trees, test results)
• You are reading files and only need the structure or API surface
• You want to check token savings for the current session

Shell commands (use instead of raw exec)

lean-ctx -c git status          # Compressed git output
lean-ctx -c git diff            # Only meaningful diff lines
lean-ctx -c git log --oneline -10
lean-ctx -c npm install         # Strips progress bars, noise
lean-ctx -c cargo build
lean-ctx -c cargo test
lean-ctx -c docker ps
lean-ctx -c kubectl get pods
lean-ctx -c aws ec2 describe-instances
lean-ctx -c helm list
lean-ctx -c prisma migrate dev
lean-ctx -c curl -s <url>       # JSON schema extraction
lean-ctx -c ls -la <dir>        # Grouped directory listing

Supported: git, npm, pnpm, yarn, bun, deno, cargo, docker, kubectl, helm, gh, pip, ruff, go, eslint, prettier, tsc, aws, psql, mysql, prisma, swift, zig, cmake, ansible, composer, mix, bazel, systemd, terraform, make, maven, dotnet, flutter, poetry, rubocop, playwright, curl, wget, and more.

File reading (compressed modes)

lean-ctx read <file>                    # Full content with structured header
lean-ctx read <file> -m map             # Dependency graph + exports + API (~5-15% tokens)
lean-ctx read <file> -m signatures      # Function/class signatures only (~10-20% tokens)
lean-ctx read <file> -m aggressive      # Syntax-stripped (~30-50% tokens)
lean-ctx read <file> -m entropy         # Shannon entropy filtered (~20-40% tokens)
lean-ctx read <file> -m diff            # Only changed lines since last read

Use map mode when you need to understand what a file does without reading every line. Use signatures mode when you need the API surface of a module (tree-sitter for 18 languages). Use full mode only when you will edit the file.

AI Tool Integration

lean-ctx init --global                             # Install shell aliases
lean-ctx init --agent cursor --mode cli-redirect   # CLI-first (no MCP schema overhead)
lean-ctx init --agent claude --mode cli-redirect   # CLI-first (Claude Code)
lean-ctx init --agent codex --mode cli-redirect    # CLI-first (Codex)
lean-ctx init --agent opencode --mode cli-redirect # CLI-first (OpenCode)

lean-ctx init --agent copilot                      # MCP (VS Code / Copilot)
lean-ctx init --agent jetbrains                    # MCP (JetBrains)
lean-ctx init --agent windsurf                     # MCP/Hybrid (Windsurf)

Multi-Agent & Knowledge (v2.7.0+)

CLI (works in CLI-redirect and MCP setups):

lean-ctx knowledge remember "value" --category <c> --key <k>
lean-ctx knowledge recall "query"
lean-ctx knowledge search "query"
lean-ctx knowledge export [--format json|jsonl|simple] [--output <path>]
lean-ctx knowledge import <path> [--merge replace|append|skip-existing] [--dry-run]
lean-ctx knowledge remove --category <c> --key <k>

lean-ctx session task "what you're doing"
lean-ctx session finding "what you found"
lean-ctx session decision "what you decided"
lean-ctx session save

If MCP is enabled for your IDE, the same capabilities are also available as MCP tools (ctx_knowledge, ctx_session, ctx_agent, ...).

Additional Intelligence Tools (v2.19.0)

• ctx_edit(path, old_string, new_string) — search-and-replace file editing without native Read/Edit
• ctx_overview(task) — task-relevant project map at session start
• ctx_preload(task) — proactive context loader, caches task-relevant files
• ctx_semantic_search(query) — BM25 code search by meaning across the project
• ctx_intent now supports multi-intent detection and complexity classification
• Semantic cache: TF-IDF + cosine similarity for finding similar files across reads

Session Continuity (CCP)

lean-ctx sessions list          # List all CCP sessions
lean-ctx sessions show          # Show latest session state
lean-ctx wrapped                # Weekly savings report card
lean-ctx wrapped --month        # Monthly savings report card
lean-ctx benchmark run          # Real project benchmark (terminal output)
lean-ctx benchmark run --json   # Machine-readable JSON output
lean-ctx benchmark report       # Shareable Markdown report

MCP tools for CCP:

• ctx_session status — show current session state (~400 tokens)
• ctx_session load — restore previous session (cross-chat memory)
• ctx_session task "description" — set current task
• ctx_session finding "file:line — summary" — record key finding
• ctx_session decision "summary" — record architectural decision
• ctx_session save — force persist session to disk
• ctx_wrapped — generate savings report card in chat

Analytics

lean-ctx gain                   # Visual token savings dashboard
lean-ctx dashboard              # Web dashboard at localhost:3333
lean-ctx session                # Adoption statistics
lean-ctx discover               # Find uncompressed commands in shell history

Tips

• The output suffix [lean-ctx: 5029→197 tok, -96%] shows original vs compressed token count
• For large outputs, lean-ctx automatically truncates while preserving relevant context
• JSON responses from curl/wget are reduced to schema outlines
• Build errors are grouped by type with counts
• Test results show only failures with summary counts
• Cached re-reads cost only ~13 tokens