kodelyth-ecc 1.5.10 → 1.7.1

package/wiki/Platform-Support.md

@@ -1,31 +1,32 @@
 # Platform Support
 
-Kodelyth ECC works across 6 AI coding platforms. Feature availability varies by platform.
+Kodelyth ECC installs on 11 AI coding platforms via 13 install targets. Feature depth varies — hooks use Claude Code's native settings format and have no equivalent elsewhere; Cursor's agent system is incompatible with ECC's markdown agent format.
 
 ---
 
 ## Capability Matrix
 
-| Feature | Claude Code | Windsurf | Cursor | Codex CLI | Antigravity | OpenCode |
-|---|---|---|---|---|---|---|
-| 62 Agents | Yes | Yes | Yes | Yes | Yes | Yes |
-| 188 Skills | Yes | Yes | Yes | Yes | Yes | Yes |
-| 80 Commands | Yes | Yes | Yes | Yes | Yes | Yes |
-| Hooks (20+) | Yes | No | No | No | No | No |
-| Memory inject | Yes | Manual | Manual | Manual | Manual | Manual |
-| Compound learning | Yes (auto) | Yes (manual) | Yes (manual) | Yes (manual) | Yes (manual) | Yes (manual) |
-| Semantic routing | Yes | Yes | Yes | Yes | Yes | Yes |
-| `/project-launch` | Yes | Yes | Yes | Yes | Yes | Yes |
-| `/team-review` | Yes | Yes | Yes | Yes | Yes | Yes |
-| `/lessons` | Yes | Yes | Yes | Yes | Yes | Yes |
+| Feature | Claude Code | Roo Code | Codex CLI | Aider | Kimi | Windsurf | Antigravity | Gemini CLI | Cursor | Cline | OpenCode |
+|---|---|---|---|---|---|---|---|---|---|---|---|
+| 70 Agents | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | — | ✓* | — |
+| 194 Skills | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | partial | ✓ | ✓ | — | — |
+| 97 Commands | ✓ | ✓ | ✓ | ✓ | ✓ | — | ✓ | — | — | ✓* | — |
+| 22+ Hooks | ✓ | — | — | — | — | — | — | — | — | — | — |
+| Rules | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
+| Auto lessons | ✓ | — | — | — | — | — | — | — | — | — | — |
+| Manual `/lessons` | — | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
 
-**Hooks** are a Claude Code feature — they require the Claude Code hook system (`~/.claude/settings.json`). All other features work on every platform through CLAUDE.md or project-local config files.
+\* Cline flattens agents and commands into `.clinerules/` as markdown rules — they provide context but not structured invocation.
+
+**Hooks** require Claude Code's JSON hook system (`~/.claude/settings.json`). No other platform has an equivalent.
+
+**MCP Server** and **Dashboard** (`npx kodelyth-ecc mcp` / `npx kodelyth-ecc dashboard`) are standalone CLI tools that work on every platform.
 
 ---
 
 ## Claude Code
 
-The full ECC experience. All features active.
+Full feature set. All hooks auto-wired. Memory injected automatically at session start.
 
 **Install:**
 ```bash
@@ -35,125 +36,231 @@ npx kodelyth-ecc
 **File layout:**
 ```
 ~/.claude/
-├── agents/          → 62 specialist agents
-├── skills/          → 188 skills
-├── commands/        → 80 slash commands
-├── rules/           → 17 always-on rules
-└── settings.json    → hooks wired here
+├── agents/          → 70 specialist agents
+├── skills/          → 194 skills
+├── commands/        → 97 slash commands
+├── rules/           → always-on coding standards + routing
+└── hooks/           → hooks.json (wired into settings.json)
 ```
 
 **Hooks active:**
-- Session start: `read-lessons.js` + `memory inject`
-- Every prompt: `auto-recall` (BM25 search)
-- After every edit: `quality-gate` + `test-reminder`
-- Before every commit: `secret-scan` + `console-check`
-- Before every push: `branch-check`
-- Session end: `capture-correction` + `memory-capture` + `desktop-notify` + `cost-tracker`
+- Session start: `read-lessons.js` + memory inject
+- Every prompt: auto-recall (BM25 search)
+- After edits: quality-gate + test-reminder
+- Before commit: secret-scan + console-check
+- Before push: branch-check
+- Session end: capture-correction + memory-capture + cost-tracker
 
-**Compound learning:** fully automatic — corrections are captured at session end and injected at session start without any user action.
+**Safety hooks (v1.7.0):**
+- `prompt-injection-guard` — blocks prompt injection attempts
+- `token-budget-enforcer` — warns when approaching token limit
 
 ---
 
-## Windsurf
+## Roo Code
 
-Agents, skills, and commands work. No hooks.
+Agents, skills, and commands. No hooks.
 
 **Install:**
 ```bash
-npx kodelyth-ecc --target windsurf-project
+npx kodelyth-ecc --target roocode
 ```
 
 **File layout:**
 ```
-.windsurf/
+.roo/
 ├── agents/
 ├── skills/
 ├── commands/
 └── rules/
 ```
 
-**Lessons (manual):** Run `/lessons` at the start of each session to load `tasks/lessons.md`. Without hooks, lessons are not auto-injected.
+**Lessons (manual):** Run `/lessons` at session start to load `tasks/lessons.md`.
 
 ---
 
-## Cursor
+## Codex CLI
 
-Agents, skills, and commands work. No hooks.
+Agents, skills, and commands. `image-architect` uses DALL-E 3 natively.
 
 **Install:**
 ```bash
-npx kodelyth-ecc --target cursor-project
+npx kodelyth-ecc --target codex-home
 ```
 
 **File layout:**
 ```
-.cursor/
+~/.codex/
 ├── agents/
 ├── skills/
 ├── commands/
 └── rules/
 ```
 
-**Lessons (manual):** Run `/lessons` at the start of each session.
+**Image generation:** `image-architect` routes to DALL-E 3 automatically on Codex CLI (no API key required).
 
 ---
 
-## Codex CLI
+## Aider
 
-Full agents and skills. `image-architect` uses DALL-E 3 natively (no API key required).
+Agents, skills, and commands. `CONVENTIONS.md` auto-generated at project root.
 
 **Install:**
 ```bash
-npx kodelyth-ecc --target codex-home
+npx kodelyth-ecc --target aider
 ```
 
 **File layout:**
 ```
-~/.codex/
+.aider-ecc/
 ├── agents/
 ├── skills/
 ├── commands/
 └── rules/
+CONVENTIONS.md     → auto-generated, points aider at .aider-ecc/
 ```
 
-**Image generation:** `image-architect` on Codex CLI routes to DALL-E 3 automatically.
+**Usage:**
+```bash
+aider --read .aider-ecc/rules/agent-intent-routing.md \
+      --read .aider-ecc/agents/debug-detective.md
+```
 
 ---
 
-## Google Antigravity
+## Kimi
 
-Full agents and skills. `image-architect` uses Gemini Imagen 3 natively (no API key required).
+Agents, skills, and commands.
 
 **Install:**
 ```bash
-npx kodelyth-ecc --target antigravity
+npx kodelyth-ecc --target kimi
 ```
 
 **File layout:**
 ```
-~/.antigravity/
+.kimi/
 ├── agents/
 ├── skills/
 ├── commands/
 └── rules/
 ```
 
-**Image generation:** `image-architect` on Antigravity routes to Gemini Imagen 3 automatically — the most capable image model in the ECC stack.
+---
 
-**Recommended models (Antigravity-specific):**
+## Windsurf
 
-| Task | Model |
-|---|---|
-| Most coding work | Gemini 3.1 Pro (Low) |
-| Architecture, security audits, hard bugs | Gemini 3.1 Pro (High) |
-| Claude agent behavior + extended thinking | Claude Sonnet 4.6 (Thinking) |
-| Hardest problems, last resort | Claude Opus 4.6 (Thinking) |
+Agents, skills, and rules. No commands (Windsurf workflows use a different format). A `.windsurfrules` file is generated from the common rules.
+
+**Install (project):**
+```bash
+npx kodelyth-ecc --target windsurf-project
+```
+
+**Install (global):**
+```bash
+npx kodelyth-ecc --target windsurf-home
+```
+
+**File layout:**
+```
+.windsurf/
+├── agents/
+├── skills/
+└── rules/
+.windsurfrules     → auto-generated from common rules
+```
+
+---
+
+## Antigravity
+
+Agents (installed as `.agent/skills/`), commands (as `.agent/workflows/`), rules. Four core ECC skills are copied into rules as always-on context. The full 194-skill directory is not installed (Antigravity has no separate skills system).
+
+**Install:**
+```bash
+npx kodelyth-ecc --target antigravity
+```
+
+**File layout:**
+```
+.agent/
+├── skills/          → 70 agents (mapped to Antigravity's skills system)
+├── workflows/       → 97 commands (mapped to Antigravity's workflows)
+└── rules/           → rules + 4 featured ECC skills as context
+```
+
+**Image generation:** `image-architect` routes to Gemini Imagen 3 natively.
+
+---
+
+## Gemini CLI
+
+Agents, skills, and rules. No commands. A `GEMINI.md` is generated pointing to the installed content.
+
+**Install (project):**
+```bash
+npx kodelyth-ecc --target gemini-project
+```
+
+**Install (global):**
+```bash
+npx kodelyth-ecc --target gemini-home
+```
+
+**File layout:**
+```
+.gemini/
+├── agents/
+├── skills/
+├── rules/
+└── GEMINI.md        → auto-generated context file
+```
+
+---
+
+## Cursor
+
+Rules and skills only. Cursor's agent system uses a different YAML format that is not compatible with ECC's markdown agents.
+
+**Install:**
+```bash
+npx kodelyth-ecc --target cursor-project
+```
+
+**File layout:**
+```
+.cursor/
+├── skills/
+└── rules/           → flattened from all rule directories
+```
+
+---
+
+## Cline
+
+Rules, agents, and commands all flattened into `.clinerules/` as markdown context files. No separate skills directory.
+
+**Install:**
+```bash
+npx kodelyth-ecc --target cline
+```
+
+**File layout:**
+```
+.clinerules/
+├── (all rule .md files)
+├── (all agent .md files — flattened)
+└── (all command .md files — flattened)
+```
+
+Cline reads everything in `.clinerules/` as persistent context, so agents and commands are available as reference material even though they aren't structured invocations.
 
 ---
 
 ## OpenCode
 
-Agents, skills, and commands work.
+Rules only. OpenCode has no agent or skill directory concept.
 
 **Install:**
 ```bash
@@ -163,36 +270,32 @@ npx kodelyth-ecc --target opencode
 **File layout:**
 ```
 .opencode/
-├── agents/
-├── skills/
-├── commands/
-└── rules/
+└── rules/           → flattened from all rule directories
 ```
 
 ---
 
 ## Lessons on Non-Claude-Code Platforms
 
-On all platforms except Claude Code, the compound learning system requires one manual step per session:
+On all platforms except Claude Code, load project lessons manually at session start:
 
-1. Start your session
-2. Type `/lessons`
-3. ECC reads `tasks/lessons.md` and activates all project rules for the session
+```
+/lessons
+```
 
-Claude Code users: this happens automatically via the `read-lessons.js` SessionStart hook.
+This reads `tasks/lessons.md` and activates all project rules for the session. On Claude Code, this happens automatically via the `read-lessons.js` SessionStart hook.
 
 ---
 
 ## image-architect Platform Routing
 
-| Platform | Image model used |
+| Platform | Image model |
 |---|---|
 | Antigravity | Gemini Imagen 3 (native, no key) |
 | Codex CLI | DALL-E 3 (native, no key) |
 | Claude Code | fal.ai MCP → SVG fallback |
-| Windsurf | Native model gen → fal.ai → SVG fallback |
-| Cursor | Native model gen → fal.ai → SVG fallback |
-| OpenCode | SVG fallback |
+| Windsurf / Cursor | native model gen → fal.ai → SVG fallback |
+| Others | SVG fallback |
 
 ---
 
@@ -200,8 +303,22 @@ Claude Code users: this happens automatically via the `read-lessons.js` SessionS
 
 | OS | Supported | Install method |
 |---|---|---|
-| macOS | Yes | `npx` or `./install.sh` |
-| Linux | Yes | `npx` or `./install.sh` |
-| Windows | Yes | `npx` or `./install.ps1` |
+| macOS | ✓ | `npx` or `./install.sh` |
+| Linux | ✓ | `npx` or `./install.sh` |
+| Windows | ✓ | `npx` or `.\install.ps1` |
+
+The `desktop-notify` hook (macOS system notification) is macOS-only and silently skips on other platforms.
+
+---
+
+## Choosing a Platform
 
-All Node.js-based scripts are cross-platform. The `desktop-notify` hook (macOS notification) is macOS-only and silently skips on other platforms.
+| Workflow | Best platform |
+|---|---|
+| Maximum feature set | Claude Code |
+| Git-native automation | Aider |
+| VS Code + Cline extension | Cline or Roo Code |
+| OpenAI stack | Codex CLI |
+| Deep reasoning, complex architecture | Antigravity |
+| Windsurf IDE | Windsurf |
+| Gemini stack | Gemini CLI |

package/wiki/Skill-Reference.md

@@ -1,6 +1,6 @@
 # Skill Reference
 
-Skills are domain knowledge and workflow definitions loaded into your session with a slash command. 188 skills total.
+Skills are domain knowledge and workflow definitions loaded into your session with a slash command. 194 skills total.
 
 Skills provide deep, actionable guidance — patterns, checklists, code templates, anti-patterns — that agents draw on when they are active. Rules define *what* to do; skills define *how* to do it.
 
@@ -39,7 +39,8 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | Skill | Command | What it provides |
 |---|---|---|
 | Memory Protocol | `/memory` | BM25 recall workflow, capture triggers, review-pending flow |
-| Lessons | `/lessons` | Load/save/manage `tasks/lessons.md` on all 6 platforms |
+| Memory Evolve | `/memory-evolve` | Self-improving memory analysis, pattern extraction (v1.7.0) |
+| Lessons | `/lessons` | Load/save/manage `tasks/lessons.md` on all 11 platforms |
 | Configure ECC | `/configure-ecc` | Customize agents, hooks, and rules for your project |
 | Quickstart | `/kodelyth-quickstart` | 5-minute tour of the entire toolkit |
 
@@ -53,6 +54,8 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 |---|---|
 | Coding standards | `/coding-standards` |
 | Bun runtime | `/bun-runtime` |
+| Next.js patterns | `/nextjs-patterns` |
+| React patterns | `/react-patterns` |
 
 ### Rust
 
@@ -60,6 +63,7 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 |---|---|
 | Rust patterns | `/rust-patterns` |
 | Rust testing | `/rust-testing` |
+| Ownership & lifetimes | `/rust-ownership` |
 
 ### Java / Spring Boot
 
@@ -80,6 +84,22 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | Swift protocol + DI testing | `/swift-protocol-di-testing` |
 | Compose multiplatform | `/compose-multiplatform-patterns` |
 
+### Python
+
+| Skill | Command |
+|---|---|
+| Python async patterns | `/python-async-patterns` |
+| FastAPI patterns | `/fastapi-patterns` |
+| Django TDD | `/django-tdd` |
+
+### Go
+
+| Skill | Command |
+|---|---|
+| Go idiomatic patterns | `/go-patterns` |
+| Go error handling | `/go-error-handling` |
+| Go concurrency | `/go-concurrency` |
+
 ---
 
 ## Architecture & Design Skills
@@ -93,6 +113,8 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | AI-First Engineering | `/ai-first-engineering` | Designing systems with AI in the loop |
 | Agentic Engineering | `/agentic-engineering` | Patterns for multi-agent orchestration |
 | Backend Patterns | `/backend-patterns` | Service architecture, data layer, API patterns |
+| Microservices | `/microservices` | Distributed system patterns, service mesh, async messaging |
+| Domain-Driven Design | `/ddd` | Bounded contexts, aggregates, ubiquitous language |
 
 ---
 
@@ -105,6 +127,8 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | AI Regression Testing | `/ai-regression-testing` | Preventing AI-introduced regressions |
 | Benchmark | `/benchmark` | Performance benchmarking patterns |
 | Verification Loop | `/verification-loop` | Multi-step verification before marking work done |
+| Load Testing | `/load-testing` | k6, Locust, Artillery setup and patterns |
+| E2E Testing Strategy | `/e2e-strategy` | End-to-end test planning and execution |
 
 ---
 
@@ -117,6 +141,9 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | Security Bounty Hunter | `/security-bounty-hunter` | Structured vuln discovery process |
 | Safety Guard | `/safety-guard` | Guardrail patterns for AI-generated code |
 | Spring Boot Security | `/springboot-security` | Spring-specific auth, CSRF, secrets |
+| Secrets Management | `/secrets-management` | `.env`, vaults, rotation, credential handling |
+| Supply Chain Security | `/supply-chain-security` | SBOM, SLSA, artifact integrity (v1.7.0) |
+| Zero Trust Architecture | `/zero-trust` | Implementation patterns for zero-trust systems |
 
 ---
 
@@ -128,13 +155,18 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | Canary Watch | `/canary-watch` | Canary deployment monitoring patterns |
 | Automation Audit Ops | `/automation-audit-ops` | Audit trail for automated operations |
 | Continuous Agent Loop | `/continuous-agent-loop` | Long-running agent patterns |
+| Docker & Containers | `/docker-patterns` | Docker, Kubernetes, container security |
+| CI/CD Pipelines | `/ci-cd-patterns` | GitHub Actions, GitLab CI, CircleCI setup |
+| Infrastructure as Code | `/iac-patterns` | Terraform, CloudFormation, Pulumi |
+| Observability | `/observability` | Logging, metrics, tracing, SLOs, OpenTelemetry |
 
 ---
 
-## Agent Orchestration Skills
+## Swarm & Orchestration Skills
 
 | Skill | Command | What it provides |
 |---|---|---|
+| Swarm Orchestrator | `/swarm-orchestrator` | Multi-agent parallel execution and coordination (v1.7.0) |
 | Agent Eval | `/agent-eval` | Evaluating agent output quality |
 | Agent Handoff | `/agent-handoff` | Multi-agent chain handoff protocols |
 | Agent Harness Construction | `/agent-harness-construction` | Building agent test harnesses |
@@ -154,6 +186,19 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | API Connector Builder | `/api-connector-builder` | Building typed API connectors |
 | Claude API | `/claude-api` | Anthropic API patterns, tool use, streaming |
 | Agent Payment (x402) | `/agent-payment-x402` | Monetizing agents with x402 protocol |
+| GraphQL Patterns | `/graphql-patterns` | Schema design, resolver patterns, federation |
+| WebSocket Patterns | `/websocket-patterns` | Real-time communication, state sync |
+
+---
+
+## Cost & Model Routing (v1.7.0)
+
+| Skill | Command | What it provides |
+|---|---|---|
+| Cost-Aware Routing | `/cost-aware-routing` | Match tasks to optimal model (Haiku/Sonnet/Opus) |
+| Context Budget | `/context-budget` | Managing context window efficiently |
+| Token Budget Advisor | `/token-budget-advisor` | Cost optimization guidance |
+| Model Selection | `/model-selection` | When to use which Claude or Gemini model |
 
 ---
 
@@ -162,26 +207,45 @@ Type the skill name with a `/` prefix. Skills load instantly and activate for th
 | Skill | Command | Domain |
 |---|---|---|
 | ClickHouse IO | `/clickhouse-io` | Analytics database patterns |
-| Context Budget | `/context-budget` | Managing context window efficiently |
 | Research Ops | `/research-ops` | Structured research workflows |
 | Rules Distill | `/rules-distill` | Extracting rules from codebase patterns |
 | Skill Stocktake | `/skill-stocktake` | Auditing which skills are active |
 | Strategic Compact | `/strategic-compact` | Condensing context before large tasks |
-| Token Budget Advisor | `/token-budget-advisor` | Cost optimization guidance |
 | Search First | `/search-first` | Always-search-before-implement workflow |
 | SEO | `/seo` | Technical SEO, meta, schema, Core Web Vitals |
 | Remotion Video | `/remotion-video-creation` | Programmatic video patterns |
 
 ---
 
+## Devil-Mode Security Skills (v1.7.0)
+
+| Skill | Command | What it provides |
+|---|---|---|
+| Prompt Injection Hunting | `/prompt-injection-hunting` | Finding AI injection vectors |
+| Supply Chain Auditing | `/supply-chain-auditing` | Dependency vulnerability analysis |
+| Secret Hunting | `/secret-hunting` | Hardcoded credential detection |
+| License Compliance | `/license-compliance` | GPL and license violation hunting |
+| Jailbreak Testing | `/jailbreak-testing` | Safety bypass testing techniques |
+| Code Theft Detection | `/code-theft-detection` | Exfiltration and unauthorized access paths |
+| Backdoor Hunting | `/backdoor-hunting` | Identifying inserted malicious code |
+| Chaos Engineering | `/chaos-engineering` | Failure mode and resilience testing |
+
+---
+
 ## Parallel Commands That Fire Multiple Skills
 
 | Command | Skills activated |
 |---|---|
 | `/project-launch` | architect + pair-programmer + security-reviewer + tdd-guide + ux-reviewer |
 | `/team-review` | code-reviewer + security-reviewer + performance-optimizer + api-guardian |
+| `/security-audit` | security-reviewer + dependency-doctor + api-guardian |
+| `/debug-blitz` | debug-detective + silent-failure-hunter + env-debugger |
+| `/refactor-sprint` | refactor-cleaner + code-simplifier + type-design-analyzer + tdd-guide |
+| `/pre-release` | release-captain + security-reviewer + code-reviewer |
+| `/onboard` | code-explorer + architect + doc-updater |
+| `/devil-mode` | prompt-injection-hunter + supply-chain-auditor + secret-hunter + license-violation-finder (or all 8 with `--all`) |
 
-These are not skill loads — they fire 4-5 specialist agents in parallel and return all results at once.
+These are not skill loads — they fire 4-8 specialist agents in parallel and return all results at once.
 
 ---
 
@@ -193,3 +257,34 @@ If you are unsure what command to use, just describe the problem. The semantic r
 You:  "How should I approach debugging this?"
 AI:   → Loading /smart-debug (debugging workflow signal)
 ```
+
+---
+
+## Skill Categories
+
+**Quick overview by category:**
+
+- **9 workflow skills** — core debugging, testing, review, APIs, security
+- **27 language/framework skills** — TypeScript, Rust, Java, Python, Go, Swift, Kotlin, and more
+- **9 architecture skills** — design patterns, ADRs, microservices, DDD
+- **7 testing skills** — TDD, E2E, benchmarking, flake prevention
+- **8 security skills** — OWASP, bounty hunting, supply chain, zero trust
+- **8 DevOps skills** — containers, CI/CD, IaC, observability
+- **9 swarm/orchestration skills** — multi-agent coordination, autonomy
+- **6 API skills** — REST, GraphQL, WebSockets, payment
+- **4 cost/routing skills** — model selection, token budgets (v1.7.0)
+- **8 specialized skills** — ClickHouse, SEO, video, research
+- **8 devil-mode security skills** — red-team, penetration testing (v1.7.0)
+
+**Total: 194 skills**
+
+---
+
+## Installing All Skills
+
+```bash
+# Download and cache all 194 skills locally
+npx kodelyth-ecc --profile full
+```
+
+Or load skills on-demand as needed.