clawpowers 1.1.4 → 2.0.0

package/.opencode/INSTALL.md

@@ -1,52 +0,0 @@
-# ClawPowers for OpenCode — Installation
-
-Fetch and follow these instructions to install ClawPowers skills in your OpenCode session.
-
-## Quick Install
-
-Paste this into your OpenCode session:
-
-```
-Fetch and follow instructions from https://raw.githubusercontent.com/up2itnow0822/clawpowers/main/.opencode/INSTALL.md
-```
-
-## What Gets Installed
-
-ClawPowers injects 20 skills into your agent context:
-
-- **14 Core Development Skills** — TDD, subagent orchestration, systematic debugging, code review, git worktrees, planning, and more
-- **6 Extended Skills** — Agent payments (x402), security auditing, content pipeline, market intelligence, lead prospecting, and metacognitive learning
-
-## Session Hook Setup
-
-To auto-inject ClawPowers on every OpenCode session start, add to your OpenCode configuration:
-
-```json
-{
-  "hooks": {
-    "session_start": "bash /path/to/clawpowers/hooks/session-start"
-  }
-}
-```
-
-## Manual Context Injection
-
-Copy the contents of `skills/using-clawpowers/SKILL.md` and prepend it to your OpenCode session context. This gives you all skill triggers without the runtime layer.
-
-## Runtime Layer (Optional)
-
-For persistent memory, outcome tracking, and self-improvement:
-
-```bash
-npx clawpowers init
-```
-
-Creates `~/.clawpowers/` with:
-- `state/` — cross-session key-value store
-- `metrics/` — JSONL outcome logs per skill
-- `checkpoints/` — resumable workflow state
-- `feedback/` — RSI analysis outputs
-
-## Skills Reference
-
-Skills activate on pattern recognition. See `skills/using-clawpowers/SKILL.md` for the full list of 20 skills and their trigger conditions.

package/ARCHITECTURE.md

@@ -1,69 +0,0 @@
-# ClawPowers Architecture
-
-## Design Principles
-
-1. **Skills execute, not just instruct.** Every skill can invoke tools, persist state, and measure outcomes.
-2. **Simplicity is a feature.** Complex functionality, simple interface. One SKILL.md per skill. Zero mandatory dependencies.
-3. **Progressive enhancement.** Works as static markdown (like competitors). Gains runtime powers when the persistence layer is available.
-4. **Platform-agnostic.** Same skills, same format, every platform: Claude Code, Cursor, Codex, OpenCode, Gemini CLI.
-5. **Self-improving.** The system gets better at helping you the more you use it.
-
-## Skill Format
-
-Every skill is a directory containing a `SKILL.md` with YAML frontmatter:
-
-```yaml
----
-name: skill-name
-description: When to trigger this skill
-version: 1.0.0
-requires:
-  tools: []          # Optional: CLI tools needed (e.g., git, npm, trivy)
-  runtime: false     # Optional: true if skill needs persistence layer
-metrics:
-  tracks: []         # What outcomes this skill measures
-  improves: []       # What parameters RSI can adjust
----
-```
-
-## Runtime Layer (Optional, Additive)
-
-The runtime layer lives at `~/.clawpowers/` and provides three services:
-
-### 1. Persistence (`runtime/persistence/`)
-- File-based key-value store for cross-session state
-- Checkpoint system for resumable workflows
-- Schema: flat file key-value store under `state/`
-
-### 2. Metrics (`runtime/metrics/`)
-- Outcome tracking per skill execution
-- Time-to-completion, success rate, error patterns
-- Statistical process control for anomaly detection
-- Export to JSON for dashboard consumption
-
-### 3. Feedback (`runtime/feedback/`)
-- RSI engine: measure → analyze → adapt cycle
-- Parameter adjustment recommendations
-- Skill effectiveness scoring
-- Anti-pattern detection from execution history
-
-## Session Hook
-
-The session-start hook (`hooks/session-start`) injects the `using-clawpowers` skill content into the agent context. This is the only mandatory injection — all other skills load on-demand via trigger matching.
-
-The hook:
-1. Detects platform (Claude Code, Cursor, Codex, OpenCode, Gemini)
-2. Reads `skills/using-clawpowers/SKILL.md`
-3. Outputs platform-appropriate JSON for context injection
-4. Optionally initializes persistence layer if not present
-
-## Graceful Degradation
-
-If the runtime layer is not installed:
-- Skills still function as static methodology guides (same as competitors)
-- Persistence features degrade to session-only memory
-- Metrics are not collected
-- RSI feedback is not available
-- All extended skills that require tool execution note the limitation
-
-This means ClawPowers works everywhere Superpowers works, and does more everywhere the runtime is available.