@ngxtm/devkit 3.18.0 → 3.20.0

package/skills/incident-response-incident-response/SKILL.md

@@ -45,19 +45,19 @@ Orchestrate multi-agent incident response with modern SRE practices for rapid re
 ## Phase 1: Detection & Triage
 
 ### 1. Incident Detection and Classification
-- Use Task tool with subagent_type="incident-responder"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "URGENT: Detect and classify incident: $ARGUMENTS. Analyze alerts from PagerDuty/Opsgenie/monitoring. Determine: 1) Incident severity (P0-P3), 2) Affected services and dependencies, 3) User impact and business risk, 4) Initial incident command structure needed. Check error budgets and SLO violations."
 - Output: Severity classification, impact assessment, incident command assignments, SLO status
 - Context: Initial alerts, monitoring dashboards, recent changes
 
 ### 2. Observability Analysis
-- Use Task tool with subagent_type="observability-monitoring::observability-engineer"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Perform rapid observability sweep for incident: $ARGUMENTS. Query: 1) Distributed tracing (OpenTelemetry/Jaeger), 2) Metrics correlation (Prometheus/Grafana/DataDog), 3) Log aggregation (ELK/Splunk), 4) APM data, 5) Real User Monitoring. Identify anomalies, error patterns, and service degradation points."
 - Output: Observability findings, anomaly detection, service health matrix, trace analysis
 - Context: Severity level from step 1, affected services
 
 ### 3. Initial Mitigation
-- Use Task tool with subagent_type="incident-responder"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Implement immediate mitigation for P$SEVERITY incident: $ARGUMENTS. Actions: 1) Traffic throttling/rerouting if needed, 2) Feature flag disabling for affected features, 3) Circuit breaker activation, 4) Rollback assessment for recent deployments, 5) Scale resources if capacity-related. Prioritize user experience restoration."
 - Output: Mitigation actions taken, temporary fixes applied, rollback decisions
 - Context: Observability findings, severity classification
@@ -65,19 +65,19 @@ Orchestrate multi-agent incident response with modern SRE practices for rapid re
 ## Phase 2: Investigation & Root Cause Analysis
 
 ### 4. Deep System Debugging
-- Use Task tool with subagent_type="error-debugging::debugger"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Conduct deep debugging for incident: $ARGUMENTS using observability data. Investigate: 1) Stack traces and error logs, 2) Database query performance and locks, 3) Network latency and timeouts, 4) Memory leaks and CPU spikes, 5) Dependency failures and cascading errors. Apply Five Whys analysis."
 - Output: Root cause identification, contributing factors, dependency impact map
 - Context: Observability analysis, mitigation status
 
 ### 5. Security Assessment
-- Use Task tool with subagent_type="security-scanning::security-auditor"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Assess security implications of incident: $ARGUMENTS. Check: 1) DDoS attack indicators, 2) Authentication/authorization failures, 3) Data exposure risks, 4) Certificate issues, 5) Suspicious access patterns. Review WAF logs, security groups, and audit trails."
 - Output: Security assessment, breach analysis, vulnerability identification
 - Context: Root cause findings, system logs
 
 ### 6. Performance Engineering Analysis
-- Use Task tool with subagent_type="application-performance::performance-engineer"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Analyze performance aspects of incident: $ARGUMENTS. Examine: 1) Resource utilization patterns, 2) Query optimization opportunities, 3) Caching effectiveness, 4) Load balancer health, 5) CDN performance, 6) Autoscaling triggers. Identify bottlenecks and capacity issues."
 - Output: Performance bottlenecks, resource recommendations, optimization opportunities
 - Context: Debug findings, current mitigation state
@@ -85,13 +85,13 @@ Orchestrate multi-agent incident response with modern SRE practices for rapid re
 ## Phase 3: Resolution & Recovery
 
 ### 7. Fix Implementation
-- Use Task tool with subagent_type="backend-development::backend-architect"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Design and implement production fix for incident: $ARGUMENTS based on root cause. Requirements: 1) Minimal viable fix for rapid deployment, 2) Risk assessment and rollback capability, 3) Staged rollout plan with monitoring, 4) Validation criteria and health checks. Consider both immediate fix and long-term solution."
 - Output: Fix implementation, deployment strategy, validation plan, rollback procedures
 - Context: Root cause analysis, performance findings, security assessment
 
 ### 8. Deployment and Validation
-- Use Task tool with subagent_type="deployment-strategies::deployment-engineer"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Execute emergency deployment for incident fix: $ARGUMENTS. Process: 1) Blue-green or canary deployment, 2) Progressive rollout with monitoring, 3) Health check validation at each stage, 4) Rollback triggers configured, 5) Real-time monitoring during deployment. Coordinate with incident command."
 - Output: Deployment status, validation results, monitoring dashboard, rollback readiness
 - Context: Fix implementation, current system state
@@ -99,13 +99,13 @@ Orchestrate multi-agent incident response with modern SRE practices for rapid re
 ## Phase 4: Communication & Coordination
 
 ### 9. Stakeholder Communication
-- Use Task tool with subagent_type="content-marketing::content-marketer"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Manage incident communication for: $ARGUMENTS. Create: 1) Status page updates (public-facing), 2) Internal engineering updates (technical details), 3) Executive summary (business impact/ETA), 4) Customer support briefing (talking points), 5) Timeline documentation with key decisions. Update every 15-30 minutes based on severity."
 - Output: Communication artifacts, status updates, stakeholder briefings, timeline log
 - Context: All previous phases, current resolution status
 
 ### 10. Customer Impact Assessment
-- Use Task tool with subagent_type="incident-responder"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Assess and document customer impact for incident: $ARGUMENTS. Analyze: 1) Affected user segments and geography, 2) Failed transactions or data loss, 3) SLA violations and contractual implications, 4) Customer support ticket volume, 5) Revenue impact estimation. Prepare proactive customer outreach list."
 - Output: Customer impact report, SLA analysis, outreach recommendations
 - Context: Resolution progress, communication status
@@ -113,19 +113,19 @@ Orchestrate multi-agent incident response with modern SRE practices for rapid re
 ## Phase 5: Postmortem & Prevention
 
 ### 11. Blameless Postmortem
-- Use Task tool with subagent_type="documentation-generation::docs-architect"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Conduct blameless postmortem for incident: $ARGUMENTS. Document: 1) Complete incident timeline with decisions, 2) Root cause and contributing factors (systems focus), 3) What went well in response, 4) What could improve, 5) Action items with owners and deadlines, 6) Lessons learned for team education. Follow SRE postmortem best practices."
 - Output: Postmortem document, action items list, process improvements, training needs
 - Context: Complete incident history, all agent outputs
 
 ### 12. Monitoring and Alert Enhancement
-- Use Task tool with subagent_type="observability-monitoring::observability-engineer"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Enhance monitoring to prevent recurrence of: $ARGUMENTS. Implement: 1) New alerts for early detection, 2) SLI/SLO adjustments if needed, 3) Dashboard improvements for visibility, 4) Runbook automation opportunities, 5) Chaos engineering scenarios for testing. Ensure alerts are actionable and reduce noise."
 - Output: New monitoring configuration, alert rules, dashboard updates, runbook automation
 - Context: Postmortem findings, root cause analysis
 
 ### 13. System Hardening
-- Use Task tool with subagent_type="backend-development::backend-architect"
+- Use Task tool with subagent_type="general-purpose"
 - Prompt: "Design system improvements to prevent incident: $ARGUMENTS. Propose: 1) Architecture changes for resilience (circuit breakers, bulkheads), 2) Graceful degradation strategies, 3) Capacity planning adjustments, 4) Technical debt prioritization, 5) Dependency reduction opportunities. Create implementation roadmap."
 - Output: Architecture improvements, resilience patterns, technical debt items, roadmap
 - Context: Postmortem action items, performance analysis

package/skills/incident-response-smart-fix/resources/implementation-playbook.md

@@ -8,7 +8,7 @@ This file contains detailed patterns, checklists, and code samples referenced by
 
 ## Phase 1: Issue Analysis - Error Detection and Context Gathering
 
-Use Task tool with subagent_type="error-debugging::error-detective" followed by subagent_type="error-debugging::debugger":
+Use Task tool with subagent_type="general-purpose" followed by subagent_type="general-purpose":
 
 **First: Error-Detective Analysis**
 
@@ -93,7 +93,7 @@ TESTING_REQUIREMENTS: [scenarios that must be covered]
 
 ## Phase 2: Root Cause Investigation - Deep Code Analysis
 
-Use Task tool with subagent_type="error-debugging::debugger" and subagent_type="comprehensive-review::code-reviewer" for systematic investigation:
+Use Task tool with subagent_type="general-purpose" and subagent_type="general-purpose" for systematic investigation:
 
 **First: Debugger Code Analysis**
 
@@ -185,13 +185,13 @@ ARCHITECTURAL_CONCERNS: {if systemic issues exist}
 Based on Phase 2 output, route to appropriate domain agent using Task tool:
 
 **Routing Logic:**
-- Python issues → subagent_type="python-development::python-pro"
-- TypeScript/JavaScript → subagent_type="javascript-typescript::typescript-pro"
-- Go → subagent_type="systems-programming::golang-pro"
-- Rust → subagent_type="systems-programming::rust-pro"
-- SQL/Database → subagent_type="database-cloud-optimization::database-optimizer"
-- Performance → subagent_type="application-performance::performance-engineer"
-- Security → subagent_type="security-scanning::security-auditor"
+- Python issues → subagent_type="general-purpose"
+- TypeScript/JavaScript → subagent_type="general-purpose"
+- Go → subagent_type="general-purpose"
+- Rust → subagent_type="general-purpose"
+- SQL/Database → subagent_type="general-purpose"
+- Performance → subagent_type="general-purpose"
+- Security → subagent_type="general-purpose"
 
 **Prompt Template (adapt for language):**
 ```
@@ -269,7 +269,7 @@ BACKWARD_COMPATIBILITY: {maintained | breaking with mitigation}
 
 ## Phase 4: Verification - Automated Testing and Performance Validation
 
-Use Task tool with subagent_type="unit-testing::test-automator" and subagent_type="application-performance::performance-engineer":
+Use Task tool with subagent_type="general-purpose" and subagent_type="general-purpose":
 
 **First: Test-Automator Regression Suite**
 
@@ -497,7 +497,7 @@ FINAL_VERDICT: {
 
 ## Phase 5: Documentation and Prevention - Long-term Resilience
 
-Use Task tool with subagent_type="comprehensive-review::code-reviewer" for prevention strategies:
+Use Task tool with subagent_type="general-purpose" for prevention strategies:
 
 **Prompt:**
 ```
@@ -613,13 +613,13 @@ For issues spanning multiple domains, orchestrate specialized agents sequentiall
 
 **Sequence:**
 1. **Phase 1-2**: error-detective + debugger identify slow database queries
-2. **Phase 3a**: Task(subagent_type="database-cloud-optimization::database-optimizer")
+2. **Phase 3a**: Task(subagent_type="general-purpose")
    - Optimize query with proper indexes
    - Context: "Query execution taking 5s, missing index on user_id column, N+1 query pattern detected"
-3. **Phase 3b**: Task(subagent_type="application-performance::performance-engineer")
+3. **Phase 3b**: Task(subagent_type="general-purpose")
    - Add caching layer for frequently accessed data
    - Context: "Database query optimized from 5s to 50ms by adding index on user_id column. Application still experiencing 2s response times due to N+1 query pattern loading 100+ user records per request. Add Redis caching with 5-minute TTL for user profiles."
-4. **Phase 3c**: Task(subagent_type="incident-response::devops-troubleshooter")
+4. **Phase 3c**: Task(subagent_type="general-purpose")
    - Configure monitoring for query performance and cache hit rates
    - Context: "Cache layer added with Redis. Need monitoring for: query p95 latency (threshold: 100ms), cache hit rate (threshold: >80%), cache memory usage (alert at 80%)."
 
@@ -630,11 +630,11 @@ For issues spanning multiple domains, orchestrate specialized agents sequentiall
    - Context: "TypeError: Cannot read property 'map' of undefined, 500+ occurrences in last hour, affects Safari users on iOS 14"
 2. **Phase 2**: debugger + code-reviewer investigate
    - Context: "API response sometimes returns null instead of empty array when no results. Frontend assumes array."
-3. **Phase 3a**: Task(subagent_type="javascript-typescript::typescript-pro")
+3. **Phase 3a**: Task(subagent_type="general-purpose")
    - Fix frontend with proper null checks
    - Add type guards
    - Context: "Backend API /api/users endpoint returning null instead of [] when no results. Fix frontend to handle both. Add TypeScript strict null checks."
-4. **Phase 3b**: Task(subagent_type="backend-development::backend-architect")
+4. **Phase 3b**: Task(subagent_type="general-purpose")
    - Fix backend to always return array
    - Update API contract
    - Context: "Frontend now handles null, but API should follow contract and return [] not null. Update OpenAPI spec to document this."
@@ -648,7 +648,7 @@ For issues spanning multiple domains, orchestrate specialized agents sequentiall
    - Context: "SQL injection vulnerability in login endpoint, Snyk severity: HIGH"
 2. **Phase 2**: debugger + security-auditor investigate
    - Context: "User input not sanitized in SQL WHERE clause, allows authentication bypass"
-3. **Phase 3**: Task(subagent_type="security-scanning::security-auditor")
+3. **Phase 3**: Task(subagent_type="general-purpose")
    - Implement parameterized queries
    - Add input validation
    - Add rate limiting

package/skills/learn/SKILL.md

@@ -1,16 +1,21 @@
 ---
 name: learn
 description: Guided project building — you code, AI mentors. Build your own product step-by-step with best practices and deep understanding.
-argument-hint: [topic]
+argument-hint: [topic] [--plan <path>]
 ---
 
-# Learn Mode v3.1
+# Learn Mode v3.2
 
 > Build your product. Design the architecture. Write every line. Understand every decision.
 
 ## Activation
 
 `/learn "topic"` — e.g., `/learn "JWT auth in Express"`, `/learn "build real-time chat"`
+`/learn --plan <path>` — Learn by following an existing plan (topic auto-extracted from plan title)
+`/learn "topic" --plan <path>` — Learn with custom topic + existing plan
+
+e.g., `/learn --plan plans/skill-sync-rewrite/plan.md`
+e.g., `/learn "upstream sync" --plan plans/skill-sync-rewrite/plan.md`
 
 ---
 
@@ -18,7 +23,17 @@ argument-hint: [topic]
 
 1. **Resume check**: Look in `learn/` for existing file matching topic. If found, read its YAML frontmatter and offer to resume from last checkpoint via `AskUserQuestion`.
 
-2. **Language detection**: Scan project for config files to identify primary language.
+2. **Plan import check**: If arguments contain `--plan <path>`:
+   a. Read the plan file at `<path>`. Validate it exists and has content.
+   b. **Topic resolution**: If no topic provided in arguments, extract from plan's YAML frontmatter `title` field.
+   c. Detect plan structure:
+      - **Single-file plan**: no `phase-XX` files referenced → tasks = plan's task list
+      - **Multi-phase plan**: has `## Phases` table with phase-XX links → each phase = potential learn module
+   d. If multi-phase: ask user via `AskUserQuestion` which phase(s) to learn.
+   e. Read selected phase file(s) for detailed tasks.
+   f. Store parsed steps for REVIEW phase.
+
+3. **Language detection** (skip if `--plan` already implies language from plan context): Scan project for config files to identify primary language.
 
 | Language | Config Files | Verify: Syntax | Verify: Run/Test |
 |----------|-------------|----------------|------------------|
@@ -70,6 +85,8 @@ step: 0
 total_steps: 0
 difficulty: {deep|standard|quick}
 teaching: {guided|scaffolded|demonstrated}
+plan_source: "{path or none}"
+plan_type: {single|multi-phase|none}
 started: {ISO timestamp}
 updated: {ISO timestamp}
 ---
@@ -93,9 +110,32 @@ Update frontmatter: `phase: LEARN`
 
 ---
 
+## Phase 3-ALT: REVIEW (only when --plan provided, replaces Phase 3 + 4)
+
+> Understand the plan before building. Light touch — not redesign.
+
+1. **Summarize**: Present plan overview to user:
+   > "This plan proposes: {overview}. It has {N} steps targeting {files}."
+   > Key decisions: {list key decisions from plan}
+
+2. **Socratic check** (skip in Quick difficulty): Ask 1-2 questions via `AskUserQuestion`:
+   > "Before we start — why do you think {first step} comes before {later step}?"
+   > OR "What problem does {key decision} solve?"
+   Build on user's answer. Correct misconceptions if any.
+
+3. **Adapt**: Ask via `AskUserQuestion`:
+   > "Want to reorder, skip, or add any steps? Or proceed as-is?"
+   Adjust step list based on user feedback.
+
+4. **Write to tutorial file**: Record plan source, overview, and adapted steps.
+
+Update frontmatter: `phase: REVIEW`, `total_steps: {N}`
+
+---
+
 ## Phase 3: DESIGN (Socratic architecture thinking)
 
-> User thinks first, AI guides — not the other way around.
+> **Skip this phase entirely if `--plan` was provided.** Go to Phase 3-ALT: REVIEW instead.
 
 1. **Frame the problem**: AI presents the high-level problem to solve:
    > "We need to build {topic}. Before I suggest anything — how would YOU approach this? What components or pieces do you think we need?"
@@ -134,6 +174,8 @@ Update frontmatter: `phase: DESIGN`
 
 ## Phase 4: PLAN (concrete implementation steps)
 
+> **Skip this phase entirely if `--plan` was provided.** Steps come from REVIEW phase instead.
+
 1. **Break down the chosen design** into 3-7 concrete, verifiable steps. Each step should:
    - Have a clear goal (what's done when this step is complete)
    - Build on previous steps (incremental, testable progress)
@@ -155,6 +197,10 @@ Update frontmatter: `phase: PLAN`, `total_steps: {N}`
 
 ## Phase 5: BUILD (core phase)
 
+> If `--plan` was provided, steps come from REVIEW phase (imported plan).
+> If no `--plan`, steps come from Phase 4 (PLAN) as usual.
+> Everything else in BUILD works identically for both paths.
+
 1. **For each step from the PLAN phase, follow the teaching mode**:
 
 ### Guided Mode (user codes everything)
@@ -267,6 +313,7 @@ Display: `Tutorial saved: learn/{filename}.md`
 
 ## Version History
 
+- **3.2.0** - Added --plan flag: import existing plan files, REVIEW phase replaces DESIGN+PLAN for plan-driven learning. Flow with plan: INIT → LEARN → REVIEW → BUILD → WRAP-UP
 - **3.1.0** - Added DESIGN phase (Socratic architecture) and PLAN phase (concrete steps). Full flow: INIT → LEARN → DESIGN → PLAN → BUILD → WRAP-UP
 - **3.0.0** - Teaching modes (guided/scaffolded/demonstrated), best-practice review, explain-back checkpoints, user-codes-first philosophy
 - **2.0.0** - Rewrite: adaptive difficulty via codingLevel, 4 phases, WebSearch, Socratic method, resume support, tiered verify, 17 languages, codebase-aware

package/skills/linear-claude-skill/SKILL.md

@@ -387,13 +387,13 @@ For detailed helper script usage, see **troubleshooting.md**.
 
 ### Parallel Agent Execution
 
-For bulk operations or background execution, use the `Linear-specialist` subagent:
+For bulk operations or background execution, use the `Linear-specialist` Task agent:
 
 ```javascript
 Task({
   description: "Update Linear issues",
   prompt: "Mark ENG-101, ENG-102, ENG-103 as Done",
-  subagent_type: "Linear-specialist"
+  subagent_type: "general-purpose"
 })
 ```
 

package/skills/loki-mode/ACKNOWLEDGEMENTS.md

@@ -21,7 +21,7 @@ Loki Mode is built for Claude and incorporates Anthropic's cutting-edge research
 | [Computer Use Safety](https://www.anthropic.com/news/3-5-models-and-computer-use) | Safe autonomous operation patterns |
 | [Sabotage Evaluations](https://www.anthropic.com/research/sabotage-evaluations-for-frontier-models) | Safety evaluation methodology |
 | [Effective Harnesses for Long-Running Agents](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents) | One-feature-at-a-time pattern, Playwright MCP for E2E |
-| [Claude Agent SDK Overview](https://platform.claude.com/docs/en/agent-sdk/overview) | Task tool, subagents, resume parameter, hooks |
+| [Claude Agent SDK Overview](https://platform.claude.com/docs/en/agent-sdk/overview) | Task tool, Task agents, resume parameter, hooks |
 
 ### Google DeepMind
 
@@ -101,7 +101,7 @@ AWS Bedrock's multi-agent collaboration patterns inform Loki Mode's routing and
 | Resource | Contribution |
 |----------|--------------|
 | [NVIDIA ToolOrchestra](https://github.com/NVlabs/ToolOrchestra) | Efficiency metrics, three-reward signal framework, dynamic agent selection |
-| [LerianStudio/ring](https://github.com/LerianStudio/ring) | Subagent-driven-development pattern |
+| [LerianStudio/ring](https://github.com/LerianStudio/ring) | Task-agent-driven-development pattern |
 | [Awesome Agentic Patterns](https://github.com/nibzard/awesome-agentic-patterns) | 105+ production patterns catalog |
 
 ### Best Practices Guides
@@ -124,7 +124,7 @@ Battle-tested insights from practitioners deploying agents in production.
 |--------|-------------|
 | [What Actually Works in Production for Autonomous Agents](https://news.ycombinator.com/item?id=44623207) | "Zero companies without human in the loop" |
 | [Coding with LLMs in Summer 2025](https://news.ycombinator.com/item?id=44623953) | Context curation beats automatic RAG |
-| [Superpowers: How I'm Using Coding Agents](https://news.ycombinator.com/item?id=45547344) | Sub-agents for context isolation (Simon Willison) |
+| [Superpowers: How I'm Using Coding Agents](https://news.ycombinator.com/item?id=45547344) | Task agents for context isolation (Simon Willison) |
 | [Claude Code Experience After Two Weeks](https://news.ycombinator.com/item?id=44596472) | Fresh contexts yield better results |
 | [AI Agent Benchmarks Are Broken](https://news.ycombinator.com/item?id=44531697) | LLM-as-judge has shared blind spots |
 | [How to Orchestrate Multi-Agent Workflows](https://news.ycombinator.com/item?id=45955997) | Event-driven, decoupled coordination |
@@ -150,7 +150,7 @@ Special thanks to thought leaders whose patterns and insights shaped Loki Mode:
 | **Boris Cherny** (Creator of Claude Code) | Self-verification loop (2-3x quality improvement), extended thinking mode, "Less prompting, more systems" philosophy |
 | **Ivan Steshov** | Centralized constitution, agent lineage tracking, structured artifacts as contracts |
 | **Addy Osmani** | Git checkpoint system, specification-first approach, visual aids (Mermaid diagrams) |
-| **Simon Willison** | Sub-agents for context isolation, skills system, context curation patterns |
+| **Simon Willison** | Task agents for context isolation, skills system, context curation patterns |
 
 ---
 

package/skills/loki-mode/CHANGELOG.md

@@ -158,7 +158,7 @@ Battle-tested patterns from practitioners:
 - **Confidence-Based Routing**: Auto-approve high confidence, escalate low
 - **Deterministic Outer Loops**: Rule-based validation, not LLM-judged
 - **Context Curation**: Manual selection beats automatic RAG
-- **Sub-Agents for Context Isolation**: Prevent token waste
+- **Task Agents for Context Isolation**: Prevent token waste
 - **Event-Driven Orchestration**: Async, decoupled coordination
 - **Policy-First Enforcement**: Runtime governance
 
@@ -172,7 +172,7 @@ Battle-tested patterns from practitioners:
 - Confidence-Based Routing thresholds
 - Deterministic Outer Loops workflow
 - Context Engineering principles
-- Sub-Agents for Context Isolation
+- Task Agents for Context Isolation
 
 ### Key Practitioner Insights
 
@@ -347,7 +347,7 @@ Comprehensive guide covering:
 - **Haiku**: Unit tests, monitoring, and simple tasks
 
 ### Changed
-- SKILL.md: Added Single-Responsibility Principle to subagent guidance
+- SKILL.md: Added Single-Responsibility Principle to Task agent guidance
 - SKILL.md: Clarified model selection (Opus=planning, Sonnet=dev, Haiku=tests)
 - SKILL.md: Dynamic Agent Selection table now shows Planning/Development/Testing columns
 - tool-orchestration.md: Added correlation_id, tool_reliability_rate to schema
@@ -939,7 +939,7 @@ The previous "37 agents" messaging was misleading because:
   - **Sonnet 4.5** is now clearly marked as **DEFAULT** for all standard implementation work
   - **Haiku 4.5** changed to **OPTIMIZATION ONLY** for simple/parallelizable tasks
   - **Opus 4.5** changed to **COMPLEX ONLY** for architecture & security
-  - Previous documentation incorrectly suggested Haiku as default for most subagents
+  - Previous documentation incorrectly suggested Haiku as default for most Task agents
   - Aligns with best practices: Sonnet for quality, Haiku for speed optimization only
 
 - **run.sh Implementation Gap** - RARV cycle now implemented in runner script (run.sh:870-871, 908-916):
@@ -1150,7 +1150,7 @@ For existing `.loki/` projects:
 
 ### Fixed
 - Removed duplicate CONTINUITY.md template from bootstrap script (was lines 2436-2470)
-- Removed duplicate Task Completion Report from subagent dispatch section (was lines 1731-1764)
+- Removed duplicate Task Completion Report from Task agent dispatch section (was lines 1731-1764)
 - Consolidated severity matrices (removed duplicates, kept one authoritative version)
 
 ## [2.14.0] - 2026-01-02
@@ -1353,8 +1353,8 @@ Loki Mode now combines the best practices from GitHub's ecosystem:
   - Secret scanning to prevent credential leaks
   - 5 quality gate categories with blocking rules
 
-  **Principle 2: Structured Prompting for Subagents**
-  - All subagent dispatches must include: GOAL, CONSTRAINTS, CONTEXT, OUTPUT FORMAT
+  **Principle 2: Structured Prompting for Task Agents**
+  - All Task agent dispatches must include: GOAL, CONSTRAINTS, CONTEXT, OUTPUT FORMAT
   - Goals explain "what success looks like" (not just actions)
   - Constraints define boundaries (dependencies, compatibility, performance)
   - Context includes CONTINUITY.md, ledgers, learnings, architecture decisions
@@ -1390,10 +1390,10 @@ Loki Mode now combines the best practices from GitHub's ecosystem:
 
 ### Changed
 - CODE_REVIEW phase now requires static analysis before AI reviewers
-- Subagent dispatch template updated with GOAL/CONSTRAINTS/CONTEXT/OUTPUT
+- Task agent dispatch template updated with GOAL/CONSTRAINTS/CONTEXT/OUTPUT
 - Task completion requires decision documentation (not just code output)
 - Quality gates now include static analysis tools (CodeQL, linters, security scanners)
-- Context-Aware Subagent Dispatch section rewritten for structured prompting
+- Context-Aware Task Agent Dispatch section rewritten for structured prompting
 
 ### Philosophy
 "Speed and control aren't trade-offs. They reinforce each other." - GitHub

package/skills/loki-mode/CONTEXT-EXPORT.md

@@ -13,7 +13,7 @@
 ### Key Features
 - 37 specialized agent types across 6 swarms (Engineering, Operations, Business, Data, Product, Growth)
 - Dynamic agent scaling based on project complexity
-- Task tool for subagent dispatch with fresh context
+- Task tool for Task agent dispatch with fresh context
 - Distributed task queue (pending, in-progress, completed, failed, dead-letter)
 - Circuit breakers for per-agent failure handling
 - Timeout/stuck agent detection with heartbeat monitoring

package/skills/loki-mode/README.md

@@ -522,12 +522,12 @@ Loki Mode incorporates research and patterns from leading AI labs and practition
 ### Practitioner Insights
 
 - **Boris Cherny** (Claude Code creator) - Self-verification loop, extended thinking
-- **Simon Willison** - Sub-agents for context isolation, skills system
+- **Simon Willison** - Task agents for context isolation, skills system
 - **Hacker News Community** - [Production patterns](https://news.ycombinator.com/item?id=44623207) from real deployments
 
 ### Inspirations
 
-- [LerianStudio/ring](https://github.com/LerianStudio/ring) - Subagent-driven-development pattern
+- [LerianStudio/ring](https://github.com/LerianStudio/ring) - Task-agent-driven-development pattern
 - [Awesome Agentic Patterns](https://github.com/nibzard/awesome-agentic-patterns) - 105+ production patterns
 
 **[Full Acknowledgements](ACKNOWLEDGEMENTS.md)** - Complete list of 50+ research papers, articles, and resources

package/skills/loki-mode/SKILL.md

@@ -137,7 +137,7 @@ If bugs are found in these files, document them in `.loki/CONTINUITY.md` under "
 | - Identify highest priority unblocked task                        |
 +-------------------------------------------------------------------+
 | ACT: Execute the task                                             |
-| - Dispatch subagent via Task tool OR execute directly             |
+| - Dispatch Task agent via Task tool OR execute directly             |
 | - Write code, run tests, fix issues                               |
 | - Commit changes atomically (git checkpoint)                      |
 +-------------------------------------------------------------------+
@@ -175,7 +175,7 @@ If bugs are found in these files, document them in `.loki/CONTINUITY.md` under "
 ### Task Tool Model Parameter
 ```python
 # Opus for planning/architecture ONLY
-Task(subagent_type="Plan", model="opus", description="Design system architecture", prompt="...")
+Task(subagent_type="general-purpose", model="opus", description="Design system architecture", prompt="...")
 
 # Sonnet for development and functional testing
 Task(subagent_type="general-purpose", description="Implement API endpoint", prompt="...")
@@ -286,7 +286,7 @@ Task(description="Refactor database layer for performance", prompt="...")     #
 - **Direct Routing:** Minimal context - just the task and relevant file(s)
 - **Supervisor Mode:** Full context - CONTINUITY.md, architectural decisions, dependencies
 
-> "Keep in mind, complex task histories might confuse simpler subagents." - AWS Best Practices
+> "Keep in mind, complex task histories might confuse simpler Task agents." - AWS Best Practices
 
 ### E2E Testing with Playwright MCP (Anthropic Harness Pattern)
 
@@ -325,7 +325,7 @@ mcp_servers = {
 | Metric | What to Track | Store In |
 |--------|---------------|----------|
 | Wall time | Seconds from start to completion | `.loki/metrics/efficiency/` |
-| Agent count | Number of subagents spawned | `.loki/metrics/efficiency/` |
+| Agent count | Number of Task agents spawned | `.loki/metrics/efficiency/` |
 | Retry count | Attempts before success | `.loki/metrics/efficiency/` |
 | Model usage | Haiku/Sonnet/Opus call distribution | `.loki/metrics/efficiency/` |
 
@@ -351,12 +351,12 @@ See `references/tool-orchestration.md` for full implementation details.
 
 ---
 
-## Structured Prompting for Subagents
+## Structured Prompting for Task Agents
 
 **Single-Responsibility Principle:** Each agent should have ONE clear goal and narrow scope.
 ([UiPath Best Practices](https://www.uipath.com/blog/ai/agent-builder-best-practices))
 
-**Every subagent dispatch MUST include:**
+**Every Task agent dispatch MUST include:**
 
 ```markdown
 ## GOAL (What success looks like)
@@ -446,7 +446,7 @@ See `references/agent-types.md` for complete definitions and capabilities.
 - **NEVER** skip code review between tasks
 - **NEVER** proceed with unfixed Critical/High/Medium issues
 - **NEVER** dispatch reviewers sequentially (always parallel - 3x faster)
-- **NEVER** dispatch multiple implementation subagents in parallel (conflicts)
+- **NEVER** dispatch multiple implementation Task agents in parallel (conflicts)
 - **NEVER** implement without reading task requirements first
 
 ### Review Anti-Patterns
@@ -466,7 +466,7 @@ See `references/agent-types.md` for complete definitions and capabilities.
 - **ALWAYS** wait for all reviewers before aggregating
 - **ALWAYS** fix Critical/High/Medium immediately
 - **ALWAYS** re-run ALL 3 reviewers after fixes
-- **ALWAYS** checkpoint state before spawning subagents
+- **ALWAYS** checkpoint state before spawning Task agents
 
 ---
 
@@ -618,14 +618,14 @@ context_budget:
   reserve: "90% for model reasoning"
 ```
 
-### Sub-Agents for Context Isolation
+### Task Agents for Context Isolation
 
-**Use sub-agents to prevent token waste on noisy subtasks:**
+**Use Task agents to prevent token waste on noisy subtasks:**
 
 ```
-Main agent (focused) --> Sub-agent (file search)
-                     --> Sub-agent (test running)
-                     --> Sub-agent (linting)
+Main agent (focused) --> Task agent (file search)
+                     --> Task agent (test running)
+                     --> Task agent (linting)
 ```
 
 See `references/production-patterns.md` for full practitioner patterns.
@@ -690,7 +690,7 @@ Loki Mode with PRD at path/to/prd   # Start with PRD
 |-------|-------|
 | Trigger | "Loki Mode" or "Loki Mode with PRD at [path]" |
 | Skip When | Need human approval, want to review plan first, single small task |
-| Related Skills | subagent-driven-development, executing-plans |
+| Related Skills | task-agent-driven-development, executing-plans |
 
 ---
 

package/skills/loki-mode/autonomy/run.sh

@@ -1473,7 +1473,7 @@ build_prompt() {
     local analysis_instruction="CODEBASE_ANALYSIS_MODE: No PRD. FIRST: Analyze codebase - scan structure, read package.json/requirements.txt, examine README. THEN: Generate PRD at .loki/generated-prd.md. FINALLY: Execute SDLC phases."
 
     # Context Memory Instructions
-    local memory_instruction="CONTEXT MEMORY: Save state to .loki/memory/ledgers/LEDGER-orchestrator.md before complex operations. Create handoffs at .loki/memory/handoffs/ when passing work to subagents. Extract learnings to .loki/memory/learnings/ after completing tasks. Check .loki/rules/ for established patterns. If context feels heavy, create .loki/signals/CONTEXT_CLEAR_REQUESTED and the wrapper will reset context with your ledger preserved."
+    local memory_instruction="CONTEXT MEMORY: Save state to .loki/memory/ledgers/LEDGER-orchestrator.md before complex operations. Create handoffs at .loki/memory/handoffs/ when passing work to Task agents. Extract learnings to .loki/memory/learnings/ after completing tasks. Check .loki/rules/ for established patterns. If context feels heavy, create .loki/signals/CONTEXT_CLEAR_REQUESTED and the wrapper will reset context with your ledger preserved."
 
     # Load existing context if resuming
     local context_injection=""

package/skills/loki-mode/integrations/vibe-kanban.md

@@ -10,7 +10,7 @@ Loki Mode can optionally integrate with [Vibe Kanban](https://github.com/BloopAI
 | Progress monitoring | Log files | Real-time dashboard |
 | Manual intervention | Edit queue files | Drag-and-drop tasks |
 | Code review | Automated 3-reviewer | + Visual diff review |
-| Parallel agents | Background subagents | Isolated git worktrees |
+| Parallel agents | Background Task agents | Isolated git worktrees |
 
 ## Setup
 

package/skills/loki-mode/references/core-workflow.md

@@ -34,7 +34,7 @@ Every iteration follows this cycle:
 | - Determine exact steps to complete it                            |
 +-------------------------------------------------------------------+
 | ACT: Execute the task                                             |
-| - Dispatch subagent via Task tool OR execute directly             |
+| - Dispatch Task agent via Task tool OR execute directly             |
 | - Write code, run tests, fix issues                               |
 | - Commit changes atomically (git checkpoint)                      |
 | - Update queue files (.loki/queue/*.json)                         |
@@ -266,10 +266,10 @@ move_tasks_to_pending after_task=$last_good_task_id
 
 ---
 
-## If Subagent Fails
+## If Task Agent Fails
 
 1. Do NOT try to fix manually (context pollution)
-2. Dispatch fix subagent with specific error context
-3. If fix subagent fails 3x, move to dead letter queue
+2. Dispatch fix Task agent with specific error context
+3. If fix Task agent fails 3x, move to dead letter queue
 4. Open circuit breaker for that agent type
 5. Alert orchestrator for human review

package/skills/loki-mode/references/production-patterns.md

@@ -283,20 +283,20 @@ skills_architecture:
     - Skills can be updated independently
 ```
 
-### Sub-Agents for Context Isolation
+### Task Agents for Context Isolation
 
 **Key Insight:** Prevent massive token waste by isolating context-noisy subtasks.
 
 ```python
 async def context_isolated_search(query, codebase_path):
     """
-    Use sub-agent for grep/search to prevent context pollution.
+    Use Task agent for grep/search to prevent context pollution.
     Based on Simon Willison's patterns.
     """
     # Main agent stays focused
-    # Sub-agent handles noisy file searching
+    # Task agent handles noisy file searching
 
-    search_agent = spawn_subagent(
+    search_agent = spawn_task_agent(
         role="codebase-searcher",
         context_limit="10k tokens",
         permissions=["read-only"]
@@ -538,11 +538,11 @@ cost_patterns:
     max_tier: "$200/month often needs 2-3 subscriptions"
 
   token_economics:
-    sub_agents_multiply_cost: "Each duplicates context"
+    task_agents_multiply_cost: "Each duplicates context"
     example: "5-task parallel job = 50,000+ tokens per subtask"
 
   optimization:
-    context_isolation: "Use sub-agents for noisy tasks"
+    context_isolation: "Use Task agents for noisy tasks"
     information_abstraction: "Summarize, don't dump"
     fresh_conversations: "Reset after major tasks"
     skill_on_demand: "Load only when needed"