loki-mode 5.49.2 → 5.49.3

package/README.md

@@ -11,7 +11,7 @@
 [![Agent Types](https://img.shields.io/badge/Agent%20Types-41-blue)]()
 [![Benchmarks](https://img.shields.io/badge/Benchmarks-Infrastructure%20Ready-blue)](benchmarks/)
 
-**Current Version: v5.49.2**
+**Current Version: v5.49.3**
 
 **[Autonomi](https://www.autonomi.dev/)** | **[Documentation](https://www.autonomi.dev/docs)** | **[GitHub](https://github.com/asklokesh/loki-mode)**
 
@@ -150,6 +150,16 @@ Loki Mode is powerful but not magic. Be aware of these honest limitations:
 - Human oversight is expected for: deployment credentials, domain setup, API keys, and critical business decisions
 - The system is as good as the underlying AI model -- it can make mistakes, especially on novel or complex problems
 
+## What To Expect
+
+| Project Type | Examples | Autonomy Level | Typical Experience |
+|---|---|---|---|
+| Simple | Landing page, todo app, static site, single API | High | Completes with minimal retries. Human reviews output. |
+| Standard | CRUD app with auth, REST API + React frontend | Medium | Completes most features. Complex components may need guidance. |
+| Complex | Microservices, real-time systems, ML pipelines | Guided | Use as accelerator. Human reviews between phases. |
+
+"Autonomous" means the system runs RARV cycles without prompting. It does NOT mean zero oversight.
+
 ---
 
 ## Why Loki Mode?

package/SKILL.md

@@ -3,7 +3,7 @@ name: loki-mode
 description: Multi-agent autonomous startup system. Triggers on "Loki Mode". Takes PRD to deployed product with minimal human intervention. Requires --dangerously-skip-permissions flag.
 ---
 
-# Loki Mode v5.49.2
+# Loki Mode v5.49.3
 
 **You are an autonomous agent. You make decisions. You do not ask questions. You do not stop.**
 
@@ -263,4 +263,4 @@ The following features are documented in skill modules but not yet fully automat
 | Quality gates 3-reviewer system | Implemented (v5.35.0) | 5 specialist reviewers in `skills/quality-gates.md`; execution in run.sh |
 | Benchmarks (HumanEval, SWE-bench) | Infrastructure only | Runner scripts and datasets exist in `benchmarks/`; no published results |
 
-**v5.49.2 | [Autonomi](https://www.autonomi.dev/) flagship product | ~260 lines core**
+**v5.49.3 | [Autonomi](https://www.autonomi.dev/) flagship product | ~260 lines core**

package/VERSION

@@ -1 +1 @@
-5.49.2
+5.49.3

package/dashboard/__init__.py

@@ -7,7 +7,7 @@ Modules:
     control: Session control API (start/stop/pause/resume)
 """
 
-__version__ = "5.49.2"
+__version__ = "5.49.3"
 
 # Expose the control app for easy import
 try:

package/docs/COMPETITIVE-ANALYSIS.md

@@ -85,7 +85,7 @@ GSD is the closest competitor -- a context engineering system that spawns fresh
 
 **Strengths:**
 - 85.9-87.7% Pass@1 on HumanEval
-- 100% task completion rate in evaluations
+- High task completion rate in evaluations (100% reported by MetaGPT authors; not independently verified)
 - Standard Operating Procedures (SOPs) reduce hallucinations
 - Assembly line paradigm with role specialization
 - Low cost: ~$1.09 per project completion

package/docs/INSTALLATION.md

@@ -2,7 +2,7 @@
 
 The flagship product of [Autonomi](https://www.autonomi.dev/). Complete installation instructions for all platforms and use cases.
 
-**Version:** v5.49.2
+**Version:** v5.49.3
 
 ---
 
@@ -36,9 +36,7 @@ The flagship product of [Autonomi](https://www.autonomi.dev/). Complete installa
 
 - [Quick Install (Recommended)](#quick-install-recommended)
 - [VS Code Extension](#vs-code-extension)
-- [npm (Node.js)](#npm-nodejs)
-- [Homebrew (macOS/Linux)](#homebrew-macoslinux)
-- [Docker](#docker)
+- [Alternative Methods](#alternative-methods)
 - [Sandbox Mode](#sandbox-mode)
 - [Multi-Provider Support](#multi-provider-support)
 - [Claude Code (CLI)](#claude-code-cli)
@@ -53,23 +51,19 @@ The flagship product of [Autonomi](https://www.autonomi.dev/). Complete installa
 
 ## Quick Install (Recommended)
 
-Choose your preferred method:
-
 ```bash
-# Option A: npm (easiest)
-npm install -g loki-mode
+git clone https://github.com/asklokesh/loki-mode.git ~/.claude/skills/loki-mode
+```
 
-# Option B: Homebrew (macOS/Linux)
-brew tap asklokesh/tap && brew install loki-mode
+That's it. Claude Code auto-discovers skills in `~/.claude/skills/`.
 
-# Option C: Docker
-docker pull asklokesh/loki-mode:latest
+**Update:** `cd ~/.claude/skills/loki-mode && git pull`
 
-# Option D: Git clone
-git clone https://github.com/asklokesh/loki-mode.git ~/.claude/skills/loki-mode
-```
+Skip to [Verify Installation](#verify-installation) to confirm it's working.
 
-**Done!** Skip to [Verify Installation](#verify-installation).
+### Alternative Installation Methods
+
+Also available via npm, Homebrew, Docker, VS Code Extension, and GitHub Action. Each has trade-offs -- see [docs/alternative-installations.md](alternative-installations.md) for details, limitations, and current status of each method.
 
 ---
 
@@ -145,153 +139,42 @@ The extension will automatically connect when it detects the server is running a
 
 ---
 
-## npm (Node.js)
+## Alternative Methods
 
-Install via npm for the easiest setup with automatic PATH configuration.
+The following installation methods are available but each has limitations. Git clone (above) is the recommended primary method.
 
-### Prerequisites
+For full details, troubleshooting, and current status of each method, see [alternative-installations.md](alternative-installations.md).
 
-- Node.js 16.0.0 or later
+### npm
 
-### Installation
+**Status:** Published to npm registry. Verify current version: `npm view loki-mode version`
 
 ```bash
-# Global installation
 npm install -g loki-mode
-
-# The skill is automatically installed to ~/.claude/skills/loki-mode
-
-# Opt out of anonymous install telemetry:
-# LOKI_TELEMETRY_DISABLED=true npm install -g loki-mode
-# Or set DO_NOT_TRACK=1
-```
-
-### Usage
-
-```bash
-# Use the CLI
-loki start ./my-prd.md
-loki status
-loki dashboard
-
-# Or invoke in Claude Code
-claude --dangerously-skip-permissions
-> Loki Mode with PRD at ./my-prd.md
-```
-
-### Updating
-
-```bash
-npm update -g loki-mode
-```
-
-### Uninstalling
-
-```bash
-npm uninstall -g loki-mode
-rm -rf ~/.claude/skills/loki-mode
 ```
 
----
-
-## Homebrew (macOS/Linux)
+Requires Node.js 16+. Provides the `loki` CLI and auto-installs the skill to `~/.claude/skills/loki-mode`.
 
-Install via Homebrew with automatic dependency management.
+### Homebrew
 
-### Prerequisites
-
-- Homebrew (https://brew.sh)
-
-### Installation
+**Status:** Available via tap. Verify formula: `brew info asklokesh/tap/loki-mode`
 
 ```bash
-# Add the tap
-brew tap asklokesh/tap
-
-# Install Loki Mode
-brew install loki-mode
-
-# Set up Claude Code skill integration (manual symlink required)
+brew tap asklokesh/tap && brew install loki-mode
+# Manual symlink required for Claude Code:
 ln -sf "$(brew --prefix)/opt/loki-mode/libexec" ~/.claude/skills/loki-mode
 ```
 
-### Dependencies
-
-Homebrew automatically installs:
-- bash 4.0+ (for associative arrays)
-- jq (JSON processing)
-- gh (GitHub CLI for integration)
-
-### Usage
-
-```bash
-# Use the CLI
-loki start ./my-prd.md
-loki status
-loki --help
-```
-
-### Updating
-
-```bash
-brew upgrade loki-mode
-```
-
-### Uninstalling
+### Docker
 
-```bash
-brew uninstall loki-mode
-rm -rf ~/.claude/skills/loki-mode
-```
-
----
-
-## Docker
-
-Run Loki Mode in a container for isolated execution.
-
-### Prerequisites
-
-- Docker installed and running
-
-### Installation
+**Status:** Published to Docker Hub.
 
 ```bash
-# Pull the image
 docker pull asklokesh/loki-mode:latest
-
-# Or use docker-compose
-curl -o docker-compose.yml https://raw.githubusercontent.com/asklokesh/loki-mode/main/docker-compose.yml
-```
-
-### Usage
-
-```bash
-# Run with a PRD file
 docker run -v $(pwd):/workspace -w /workspace asklokesh/loki-mode:latest start ./my-prd.md
-
-# Interactive mode
-docker run -it -v $(pwd):/workspace -w /workspace asklokesh/loki-mode:latest
-
-# Using docker-compose
-docker-compose run loki start ./my-prd.md
 ```
 
-### Environment Variables
-
-Pass your configuration via environment variables:
-
-```bash
-docker run -e LOKI_MAX_RETRIES=100 -e LOKI_BASE_WAIT=120 \
-  -v $(pwd):/workspace -w /workspace \
-  asklokesh/loki-mode:latest start ./my-prd.md
-```
-
-### Updating
-
-```bash
-docker pull asklokesh/loki-mode:latest
-```
+**Limitation:** Docker cannot run Claude Code interactively (Claude Code is a terminal-based CLI requiring TTY access). Docker is suitable for CI/CD pipelines, API-only modes, and sandbox execution -- not for the primary interactive workflow.
 
 ---
 

package/mcp/__init__.py

@@ -21,4 +21,4 @@ try:
 except ImportError:
     __all__ = ['mcp']
 
-__version__ = '5.49.2'
+__version__ = '5.49.3'

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "loki-mode",
-  "version": "5.49.2",
+  "version": "5.49.3",
   "description": "Loki Mode by Autonomi - Multi-agent autonomous startup system for Claude Code, Codex CLI, and Gemini CLI",
   "keywords": [
     "autonomi",

package/references/quality-control.md

@@ -165,6 +165,16 @@ IMPLEMENT -> BLIND REVIEW (parallel) -> DEBATE (if disagreement) -> AGGREGATE ->
 - NEVER dispatch reviewers sequentially (always parallel - 3x faster)
 - NEVER aggregate before all 3 reviewers complete
 
+### Test Quality Review (Apply to Every Review)
+
+Before approving, verify:
+- Are tests using real implementations or excessive mocks of internal code?
+- Were any assertion expected values changed in the same commit as implementation? (This is the top sign an agent cheated.)
+- Do tests verify meaningful behavior or just "runs without throwing"?
+- Could all tests pass while the feature is completely broken?
+
+Assertion manipulation in the same commit as implementation = CRITICAL finding = automatic REJECT.
+
 ### Anti-Sycophancy Protocol (CONSENSAGENT Research)
 
 **Problem:** Reviewers may reinforce each other's findings instead of critically engaging.

package/skills/quality-gates.md

@@ -14,6 +14,24 @@
 8. **Mock Detector** - Classifies internal vs external mocks; flags tests that never import source code, tautological assertions, and high internal mock ratios
 9. **Test Mutation Detector** - Detects assertion value changes alongside implementation changes (test fitting), low assertion density, and missing pass/fail tracking
 
+## Gate 8 and 9: Automated Test Integrity
+
+Gates 8 (Mock Detector) and 9 (Test Mutation Detector) run during the VERIFY phase and are enabled by default.
+
+**How they run:**
+- Gate 8 runs `tests/detect-mock-problems.sh` against all test files in the project
+- Gate 9 runs `tests/detect-test-mutations.sh` against recent commits (default: last 5, or use `--commit HASH` for targeted checks)
+- Both produce findings at HIGH/MEDIUM/LOW severity levels
+- HIGH findings = automatic FAIL (same as other blocking gates)
+
+**Disabling (not recommended):**
+```bash
+LOKI_GATE_MOCK_DETECTOR=false    # Disable gate 8
+LOKI_GATE_MUTATION_DETECTOR=false # Disable gate 9
+```
+
+---
+
 ## Guardrails Execution Modes
 
 - **Blocking**: Guardrail completes before agent starts (use for expensive operations)

package/skills/testing.md

@@ -1,5 +1,20 @@
 # Testing
 
+## Mandatory Testing Rules
+
+1. Write tests FIRST. Commit the test before writing implementation.
+2. Tests must call REAL functions with REAL inputs and assert REAL outputs.
+3. Mock ONLY external dependencies: HTTP APIs, databases, file system, third-party services.
+4. NEVER mock internal modules, utility functions, or any code that is part of this project.
+5. NEVER change a test's expected value to make it pass. If a test fails, the implementation is wrong. Fix the code, not the test.
+6. If you believe a test expectation is incorrect, document WHY and flag for council review. Do not silently change it.
+7. Every test file must have at least one assertion per tested function.
+
+Gate 8 (mock detector) and Gate 9 (mutation detector) enforce rules 3-5 automatically.
+Violations result in automatic FAIL during VERIFY phase.
+
+---
+
 ## E2E Testing with Playwright MCP
 
 **Use Playwright MCP for browser-based testing.**