shynlee04/hivemind-plugin/hm-l2-lineage-router

Claude Code and agent skill by shynlee04/hivemind-plugin

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Route tasks to correct hm-* skills based on intent classification. Maps task categories to skill loading bundles. Use when an agent needs to determine which hm-* skills to load for a given task. Triggers: "route task", "classify intent", "which skill to load", "task routing", "lineage routing", "skill bundle for task", "load skills for", "what skills do I need", "skill loading order", "determine skill lineage", "map task to skills", "skill selection for workflow". NOT for executing skills — only for determining which skills to load. Max 5 skills per bundle.

Install

npx vskill@latest install shynlee04/hivemind-plugin/hm-l2-lineage-router

SKILL.md

The Iron Law

Every task has a lineage. Route to the bundle that matches the task's category. Max 5 skills per bundle.

Lineage Router

Overview

Given a task intent, determine which hm-* skills should be loaded. This skill maps task categories to skill loading bundles. It does NOT execute skills — it only determines which skills to load and in what order.

Six task categories, six skill bundles:

CategorySkillsMax
Researchhm-detective + hm-deep-research + hm-tech-stack-ingest3
Planninghm-spec-driven-authoring + hm-planning-persistence2
Executionhm-phase-execution + hm-cross-cutting-change2
Qualityhm-test-driven-execution + hm-gate-orchestrator2
Debughm-debug + hm-completion-looping2
Reviewhm-production-readiness + hm-gate-orchestrator2

On Load

  1. Read references/routing-map.md — the complete routing map with priority ordering and dependency chains
  2. Identify the task intent from the current context
  3. Classify the task into one of the six categories

Trigger Phrases

  • "route task" / "route this task"
  • "classify intent" / "classify task intent"
  • "which skill to load" / "what skills do I need"
  • "task routing" / "lineage routing"
  • "skill bundle for task" / "load skills for"
  • "skill loading order" / "determine skill lineage"
  • "map task to skills" / "skill selection for workflow"

Routing Map

Category 1: Research Tasks

Intent signals: "investigate", "research", "find out", "analyze", "look into", "explore"

PrioritySkillRole
1hm-detectiveCodebase investigation (SCAN/READ/DEEP modes)
2hm-deep-researchVersion-matched deep research with citations
3hm-tech-stack-ingestCache third-party docs for offline reference

Loading order: hm-tech-stack-ingest (if stack cache needed) → hm-detective → hm-deep-research

Category 2: Planning Tasks

Intent signals: "plan", "spec", "requirements", "design", "architect"

PrioritySkillRole
1hm-spec-driven-authoringFalsifiable requirements + acceptance criteria
2hm-planning-persistence3-file external memory for multi-session planning

Loading order: hm-planning-persistence (setup) → hm-spec-driven-authoring (content)

Category 3: Execution Tasks

Intent signals: "implement", "build", "execute", "run the phase", "code"

PrioritySkillRole
1hm-phase-executionWave-based parallelization + checkpoint recovery
2hm-cross-cutting-changeCross-pane modification safety for breaking changes

Loading order: hm-phase-execution (primary) → hm-cross-cutting-change (when touching multiple layers)

Category 4: Quality Tasks

Intent signals: "test", "verify", "quality", "validate", "TDD", "red-green-refactor"

PrioritySkillRole
1hm-test-driven-executionRED/GREEN/REFACTOR cycles with runtime truth
2hm-gate-orchestratorQuality gate triad pipeline (lifecycle → spec → evidence)

Loading order: hm-test-driven-execution (implementation) → hm-gate-orchestrator (validation)

Category 5: Debug Tasks

Intent signals: "debug", "fix", "broken", "failing", "error", "investigate issue"

PrioritySkillRole
1hm-debugSystematic debugging with persistent state
2hm-completion-loopingGuardrail against premature completion claims

Loading order: hm-debug (investigation) → hm-completion-looping (verification)

Category 6: Review Tasks

Intent signals: "review", "audit", "readiness", "deploy check", "ship ready"

PrioritySkillRole
1hm-production-readinessDeployment safety verification (8 dimensions)
2hm-gate-orchestratorQuality gate triad for final approval

Loading order: hm-production-readiness (evidence collection) → hm-gate-orchestrator (gate verification)

Loading Rules

  1. Max 5 skills per bundle. If a task needs more than 5, split the task.
  2. Load in priority order. Lower-numbered skills load first.
  3. Respect dependencies. If Skill B depends on Skill A's output, load A first.
  4. Classify before loading. Determine the category before loading any skills.
  5. Multi-category tasks. If a task spans 2 categories, load the primary category's bundle first, then add 1-2 skills from the secondary category (staying within the 5-skill limit).
Multi-Category Decision Tree
Task spans 2 categories?
→ YES: Primary category gets full bundle (2-3 skills)
Secondary category gets 1-2 skills from its bundle
Total must be ≤ 5
→ NO: Load single bundle (2-3 skills)
Task spans 3+ categories?
→ SPLIT THE TASK. No single task should span 3+ categories.

Self-Correction

Anti-Pattern 1: Overloading

Detection: More than 5 skills being loaded for a single task. Correction: Split the task into subtasks. Each subtask gets its own bundle. Signal: "Task too broad — split into [A] and [B], then route each independently."

Anti-Pattern 2: Wrong Lineage

Detection: Loading hf- skills for an hm- task, or vice versa. Correction: This router handles hm-* lineage only. If the task is a meta-builder task (creating skills, agents, commands), route to hf-meta-builder instead. If the task is a product-dev task, use this router.

Anti-Pattern 3: Missing Input

Detection: Task intent cannot be classified into any of the 6 categories. Correction: If the task is genuinely new (e.g., "monitor production"), route to the closest category and flag: "No exact match. Routed to [closest category]. Verify the bundle covers the task's needs."

Anti-Pattern 4: Stale Bundles

Detection: A loaded skill references another skill not in the current bundle, and the referenced skill is critical for the task. Correction: Add the missing skill to the bundle if the total stays ≤ 5. If adding would exceed 5, remove the lowest-priority skill from the current bundle and document the trade-off.

Quality Contract (HMQUAL Compliance)

HMQUALComplianceEvidence
HMQUAL-01Trigger phrases ≥7 in description12 trigger phrases in description
HMQUAL-02Self-correction with 4 anti-patternsOverloading, Wrong Lineage, Missing Input, Stale Bundles
HMQUAL-03Cross-references to sibling skills13 skill references across 6 bundles + 4 consumer skills
HMQUAL-04Progressive disclosureSKILL.md (routing map) + references/routing-map.md (detailed routing)
HMQUAL-05Evals with 3 scenariosevals/evals.json — 3 trigger scenarios
HMQUAL-06Metrics scorecardmetrics/rich-gate-scorecard.md
HMQUAL-07Iron law enforcement"Max 5 skills per bundle."
HMQUAL-08Honest RICH scoringScored in metrics/rich-gate-scorecard.md

Cross-References

Related SkillBoundary
hm-coordinating-loopConsumer — uses this router to determine skill bundles for delegated tasks
hm-phase-executionConsumer — uses this router to determine execution-time skill loading
hm-phase-loopConsumer — uses this router for iterative phase skill selection
hm-subagent-delegation-patternsConsumer — uses this router to attach skill bundles to delegation packets
hm-gate-orchestratorReferenced in Quality and Review bundles — the gate triad orchestrator

Security

  • Security scanPassed, 91 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
51
Repository forks
17
Repository npm downloads, 7 days
48
7-day trend score
30

Stars and forks are for the whole GitHub repository, not this skill alone. Downloads are for the repository's npm package. Refreshed 17 Jun 2026, 4 months ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/shynlee04/hivemind-plugin/hm-l2-lineage-router/badge)](https://verified-skill.com/skills/shynlee04/hivemind-plugin/hm-l2-lineage-router)

Something wrong with this skill or its page? Tell us.