oh-my-customcode 0.72.1 → 0.73.0

package/README.md

@@ -13,7 +13,7 @@
 
 **[한국어 문서 (Korean)](./README_ko.md)**
 
-46 agents. 99 skills. 21 rules. One command.
+46 agents. 100 skills. 21 rules. One command.
 
 ```bash
 npm install -g oh-my-customcode && cd your-project && omcustom init
@@ -146,7 +146,7 @@ Each agent declares its tools, model, memory scope, and limitations in YAML fron
 
 ---
 
-### Skills (99)
+### Skills (100)
 
 | Category | Count | Includes |
 |----------|-------|----------|
@@ -282,7 +282,7 @@ your-project/
 ├── CLAUDE.md                   # Entry point
 ├── .claude/
 │   ├── agents/                 # 46 agent definitions
-│   ├── skills/                 # 99 skill modules
+│   ├── skills/                 # 100 skill modules
 │   ├── rules/                  # 21 governance rules (R000-R021)
 │   ├── hooks/                  # 15 lifecycle hook scripts
 │   ├── schemas/                # Tool input validation schemas

package/dist/cli/index.js

@@ -9325,7 +9325,7 @@ var init_package = __esm(() => {
     workspaces: [
       "packages/*"
     ],
-    version: "0.72.1",
+    version: "0.73.0",
     description: "Batteries-included agent harness for Claude Code",
     type: "module",
     bin: {

package/dist/index.js

@@ -1820,7 +1820,7 @@ var package_default = {
   workspaces: [
     "packages/*"
   ],
-  version: "0.72.1",
+  version: "0.73.0",
   description: "Batteries-included agent harness for Claude Code",
   type: "module",
   bin: {

package/package.json

@@ -3,7 +3,7 @@
   "workspaces": [
     "packages/*"
   ],
-  "version": "0.72.1",
+  "version": "0.73.0",
   "description": "Batteries-included agent harness for Claude Code",
   "type": "module",
   "bin": {

package/templates/.claude/agents/sys-memory-keeper.md

@@ -94,6 +94,29 @@ After updating memory entries, aggregate agent performance:
    - New agent: append row
 5. Enforce 20-row budget: prune lowest-usage rows
 
+### User Model Extraction (Session-End)
+
+After metrics aggregation, extract user model data:
+
+1. **Skill Preferences**: Parse conversation for Skill tool invocations
+   - Count each skill's invocations in this session
+   - Merge with existing `## User Model > ### Skill Preferences` table
+   - Keep top 10 by cumulative invocation count
+2. **Correction Patterns**: Scan for R016 violation corrections
+   - User says "no", "don't", "stop doing X" → potential correction
+   - Match to rule ID if possible (R007, R010, etc.)
+   - Update or create entry in Correction Patterns
+3. **Expertise Profile**: Analyze file access patterns
+   - Count file extensions accessed (*.ts, *.py, *.go, etc.)
+   - Map to domain: .ts→TypeScript, .py→Python, .go→Go, etc.
+   - Update primary domains list (top 3 by file access count)
+4. **Override Decisions**: Detect explicit user overrides
+   - User changes agent routing, overrides verdict, rejects suggestion
+   - Record with date and context (max 5 most recent)
+5. Write `## User Model` section to MEMORY.md (max 30 lines)
+   - New entries start at `[confidence: low]`
+   - Existing entries seen again → promote confidence
+
 ### Failure Handling
 
 - MEMORY.md update failure → report error to orchestrator

package/templates/.claude/hooks/hooks.json

@@ -438,6 +438,16 @@
         ],
         "description": "Auto-extract failure patterns from session outcomes (advisory, exit 0)"
       },
+      {
+        "matcher": "*",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash .claude/hooks/scripts/skill-extractor-analyzer.sh"
+          }
+        ],
+        "description": "Detect reusable skill candidates from session task outcomes (advisory, exit 0)"
+      },
       {
         "matcher": "*",
         "hooks": [

package/templates/.claude/hooks/scripts/skill-extractor-analyzer.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+# skill-extractor-analyzer.sh — Stop hook for skill candidate detection
+# Advisory-only: exit 0 always. Emits stderr message if candidates found.
+
+set -euo pipefail
+
+OUTCOMES_FILE="/tmp/.claude-task-outcomes-${PPID}"
+PROPOSALS_FILE="/tmp/.claude-skill-proposals-${PPID}"
+
+# Early exit if no outcomes
+if [ ! -f "$OUTCOMES_FILE" ] || [ ! -s "$OUTCOMES_FILE" ]; then
+  exit 0
+fi
+
+# Count qualifying patterns (3+ successes with 80%+ rate)
+# Group by agent_type+skill, count successes
+CANDIDATES=0
+
+if command -v jq &>/dev/null; then
+  # Parse JSONL and group by agent_type+skill
+  CANDIDATES=$(cat "$OUTCOMES_FILE" | \
+    jq -s '
+      group_by(.agent_type + "|" + (.skill // "none"))
+      | map({
+          key: .[0].agent_type + "|" + (.[0].skill // "none"),
+          total: length,
+          successes: [.[] | select(.outcome == "success")] | length
+        })
+      | map(select(.successes >= 3 and (.successes / .total) >= 0.8))
+      | length
+    ' 2>/dev/null || echo "0")
+fi
+
+if [ "$CANDIDATES" -gt 0 ] 2>/dev/null; then
+  echo "[skill-extractor] ${CANDIDATES} skill candidate(s) detected from session outcomes" >&2
+  echo "[skill-extractor] Run /skill-extractor to review and create" >&2
+
+  # Save proposal count for Stop prompt hook to pick up
+  echo "{\"candidates\": $CANDIDATES, \"timestamp\": \"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"}" > "$PROPOSALS_FILE"
+fi
+
+exit 0

package/templates/.claude/rules/SHOULD-memory-integration.md

@@ -213,6 +213,70 @@ The Metrics section shares the 200-line MEMORY.md budget:
 
 -->
 
+<!-- DETAIL: User Model (sys-memory-keeper reference)
+
+MEMORY.md supports an optional `## User Model` section (DISTINCT from `## Behaviors`) for tracking structured user interaction patterns.
+
+### User Model vs Behaviors
+
+| Aspect | Behaviors | User Model |
+|--------|-----------|------------|
+| Focus | Communication/workflow preferences | Correction patterns, expertise, skill usage |
+| Source | Conversation style observations | R016 violations, tool invocations, override decisions |
+| Update | Session-end extraction | Session-end aggregation from task outcomes |
+
+### User Model Section Format
+
+    ## User Model [auto-updated by sys-memory-keeper]
+
+    ### Correction Patterns [confidence: medium]
+    - R010 direct-write attempts: 3 times → now delegates consistently
+    - Prefers explicit agent selection over auto-routing
+
+    ### Skill Preferences [confidence: high]
+    | Skill | Invocations | Last Used |
+    |-------|------------|-----------|
+    | /research | 12 | 2026-04-03 |
+    | /pipeline auto-dev | 8 | 2026-04-03 |
+
+    ### Expertise Profile [confidence: medium]
+    - Primary domains: TypeScript, Python, Go
+    - Focus: AI agent orchestration, CLI tooling
+
+    ### Override Decisions [confidence: low]
+    - Overrode /scout SKIP verdict → INTEGRATE for RTK (#756)
+
+### Categories
+
+| Category | Source | Description |
+|----------|--------|-------------|
+| Correction Patterns | R016 violation history | Rules the user corrected most |
+| Skill Preferences | Skill tool invocation count | Most-invoked skills ranked |
+| Expertise Profile | File patterns + routing history | User's domain expertise areas |
+| Override Decisions | Explicit user overrides | When user disagreed with agent recommendation |
+
+### Budget Management
+
+User Model shares the 200-line MEMORY.md budget:
+- Max 30 lines for User Model section
+- Prune low-confidence entries first
+- Skill Preferences table: max 10 rows (top by invocation count)
+- Override Decisions: max 5 entries (most recent)
+
+### Extraction Guidelines
+
+sys-memory-keeper extracts user model data at session end:
+1. Parse task outcomes for Skill invocations → update Skill Preferences
+2. Scan conversation for R016 violations → update Correction Patterns
+3. Analyze file patterns and routing decisions → update Expertise Profile
+4. Detect explicit user overrides (verdict changes, agent redirects) → update Override Decisions
+
+### Precedence
+
+User Model data feeds into intent-detection (R015) and routing skill confidence scoring. Higher expertise in a domain → higher confidence for auto-routing to that domain's agent.
+
+-->
+
 ## Session-End Auto-Save
 
 ### Trigger

package/templates/.claude/skills/skill-extractor/SKILL.md

@@ -0,0 +1,155 @@
+---
+name: skill-extractor
+description: Analyze task trajectories to propose reusable SKILL.md candidates from successful patterns
+scope: core
+user-invocable: true
+argument-hint: "[--threshold <n>] [--dry-run]"
+version: 1.0.0
+---
+
+# Skill Extractor
+
+Analyze completed task outcomes to identify reusable patterns and propose new SKILL.md candidates. Inspired by Hermes Agent's self-learning skill extraction — adapted for oh-my-customcode's compilation metaphor.
+
+## Philosophy
+
+In the compilation metaphor: task trajectories are runtime traces, and extracted skills are new source code. This skill turns successful execution patterns into reusable knowledge artifacts.
+
+```
+Runtime traces (task outcomes) → Pattern analysis → SKILL.md proposal → User approval → mgr-creator
+```
+
+## Usage
+
+```
+/skill-extractor                    # Analyze current session outcomes
+/skill-extractor --threshold 2      # Lower success threshold (default: 3)
+/skill-extractor --dry-run          # Preview proposals without writing
+```
+
+## Options
+
+```
+--threshold, -t   Minimum success count for pattern qualification (default: 3)
+--dry-run, -d     Preview proposals to stdout only, no file writes
+--all             Include all sessions (not just current, requires task outcome history)
+```
+
+## Workflow
+
+### Phase 1: Collect Task Outcomes
+
+Read task outcome data from the session:
+
+```bash
+# Current session outcomes (from task-outcome-recorder hook)
+OUTCOMES_FILE="/tmp/.claude-task-outcomes-${PPID}"
+```
+
+If file doesn't exist or is empty: report "No task outcomes recorded in this session." and stop.
+
+Parse JSONL entries. Each entry has:
+```json
+{"agent_type": "lang-typescript-expert", "skill": "typescript-best-practices", "description": "Fix type error in auth module", "outcome": "success", "model": "sonnet", "timestamp": "2026-04-05T09:30:00Z", "duration_ms": 15000}
+```
+
+### Phase 2: Pattern Detection
+
+Group outcomes by `(agent_type, skill)` tuple:
+
+```
+Pattern: (lang-typescript-expert, typescript-best-practices)
+  → success: 5, failure: 1, total: 6
+  → success_rate: 0.83
+  → descriptions: ["Fix type error...", "Refactor module...", ...]
+```
+
+Filter qualifying patterns:
+- `success_count >= threshold` (default: 3)
+- `success_rate >= 0.8`
+- Not already an existing skill (check `.claude/skills/*/SKILL.md`)
+
+### Phase 3: Generate Proposals
+
+For each qualifying pattern, generate a SKILL.md proposal:
+
+```markdown
+## Proposal: {proposed-skill-name}
+
+**Source Pattern**: {agent_type} + {skill} ({success_count} successes, {success_rate}% rate)
+**Confidence**: {low|medium|high} (based on count and rate)
+
+### Proposed SKILL.md
+
+name: {proposed-name}
+description: {inferred from common description patterns}
+scope: core
+user-invocable: false
+
+### Rationale
+{Why this pattern should be extracted as a skill — based on frequency and success rate}
+
+### Overlap Check
+{List any existing skills with >50% keyword overlap}
+```
+
+**Confidence scoring**:
+| Successes | Rate | Confidence |
+|-----------|------|------------|
+| 3-5 | >= 0.8 | low |
+| 6-10 | >= 0.85 | medium |
+| 10+ | >= 0.9 | high |
+
+### Phase 4: Present to User
+
+Display proposals in ranked order (highest confidence first):
+
+```
+[skill-extractor] {N} skill candidates detected
+
+  1. [high] proposed-skill-name
+     Source: {agent_type} + {skill} (12 successes, 92%)
+     Description: {inferred description}
+
+  2. [medium] another-skill-name
+     Source: {agent_type} + {skill} (7 successes, 86%)
+     Description: {inferred description}
+
+Select [1-N] to create, "all" to create all, or "skip" to cancel:
+```
+
+### Phase 5: Create Skill (on approval)
+
+Delegate to mgr-creator with the proposal context:
+- Proposed name and description
+- Source pattern data
+- Confidence level
+- Any overlap warnings
+
+mgr-creator handles: SKILL.md creation, template sync, ontology registration.
+
+## Integration
+
+| System | How |
+|--------|-----|
+| task-outcome-recorder | Reads JSONL outcomes as input data |
+| feedback-collector | Complementary: feedback-collector extracts failure patterns, skill-extractor extracts success patterns |
+| mgr-creator | Delegated skill creation on user approval |
+| skills-sh-search | Check agentskills.io for existing equivalent before creating |
+| R011 (memory) | User Model tracks extraction decisions in Override Decisions |
+
+## Hook Integration
+
+The `skill-extractor-analyzer.sh` Stop hook provides a lightweight pre-analysis:
+- Reads task outcomes file
+- Counts qualifying patterns
+- Emits advisory stderr message if candidates found
+- Does NOT create skills (that requires user approval via the skill)
+
+## Safety
+
+- **User approval required**: Never auto-creates skills
+- **Overlap check**: Prevents duplicating existing skills
+- **Dry-run mode**: Preview without side effects
+- **Advisory hook**: Stop hook is advisory-only (exit 0)
+- **Confidence transparency**: All proposals show confidence scores

package/templates/.claude/skills/skills-sh-search/SKILL.md

@@ -23,6 +23,7 @@ Search the [skills.sh](https://skills.sh/) marketplace for reusable AI agent ski
 --global, -g     Install to ~/.claude/skills/ instead of project .claude/skills/
 --list, -l       List currently installed skills.sh skills
 --check, -c      Check for updates on installed skills.sh skills
+--source, -s     Search source: "skills-sh" (default) | "agentskills" | "all"
 ```
 
 ## Workflow
@@ -123,6 +124,44 @@ Suggestions:
   - Consider creating a custom skill with /create-agent
 ```
 
+## Alternative Sources
+
+### agentskills.io (opt-in)
+
+Search the [agentskills.io](https://agentskills.io/) community skill registry as an alternative source.
+
+**Search workflow:**
+```
+1. Try: npx --yes @agentskill.sh/cli search "<query>"
+2. If CLI unavailable: WebSearch "site:agentskills.io <query>"
+3. Present results with source attribution
+```
+
+**Install workflow:**
+```
+1. Run: npx --yes @agentskill.sh/cli install <slug>
+2. Verify installation in .claude/skills/
+3. Add source metadata:
+   ├── source-type: agentskills-io
+   └── source-origin: <slug>
+```
+
+**Usage:**
+```bash
+# Search agentskills.io only
+/skills-sh-search "memory management" --source agentskills
+
+# Search both sources
+/skills-sh-search "testing patterns" --source all
+```
+
+**Fallback chain:**
+| Step | Tool | Condition |
+|------|------|-----------|
+| 1 | `@agentskill.sh/cli search` | Primary — if CLI available |
+| 2 | `WebSearch site:agentskills.io` | CLI unavailable or no results |
+| 3 | Report no results | Both failed |
+
 ## Examples
 
 ```bash
@@ -140,6 +179,12 @@ Suggestions:
 
 # Check for updates
 /skills-sh-search --check
+
+# Search agentskills.io
+/skills-sh-search "agent memory" --source agentskills
+
+# Search all sources
+/skills-sh-search "code review" --source all --install
 ```
 
 ## Integration
@@ -153,6 +198,9 @@ Installed skills.sh skills are tracked with `source-type: skills-sh` metadata, e
 ### With create-agent
 If a skills.sh skill provides domain knowledge, `create-agent` can reference it when building a new agent for that domain.
 
+### With agentskills.io
+Installed agentskills.io skills are tracked with `source-type: agentskills-io` metadata, enabling `update-external` to check for updates. Default source remains skills.sh; agentskills.io is opt-in via `--source` flag.
+
 ## Safety
 
 - **Read-only by default**: Search does not modify anything

package/templates/manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "0.72.1",
+  "version": "0.73.0",
   "lastUpdated": "2026-03-24T00:00:00.000Z",
   "components": [
     {
@@ -18,7 +18,7 @@
       "name": "skills",
       "path": ".claude/skills",
       "description": "Reusable skill modules (includes slash commands)",
-      "files": 99
+      "files": 100
     },
     {
       "name": "guides",