Tool Reference

skill

Manage skills. Actions: list (list or get by ID), create, update (including approve/reject/deprecate via status), generate (from memories), versions (history), rollback, audit_log, analyze (pattern detection).

Aliases

Old tool names list_skills, skill_versions, get_audit_log, and analyze_patterns still work as aliases and route to the corresponding action on this tool.


Parameters

NameTypeRequiredDescription
actionstringYesOne of list, create, update, generate, versions, rollback, audit_log, analyze.
idstringSkill ID. Used with list (by ID), update, versions, rollback.
namestringSkill name. Used with create.
descriptionstringSkill description.
domainstring or string[]Domains. Array for create/generate. Single string for list/analyze filter.
instructionsstringSkill instructions body.
triggersstring[]Phrases that should trigger this skill.
templatesstring[]Output templates.
statusstringOne of draft, active, rejected, deprecated. Filter on list, set on update.
successor_idstringReplacement skill ID. Used when deprecating.
review_notesstringNotes when approving or rejecting.
reasonstringReason for status change.
memory_idsstring[]Memory IDs to generate a skill from. Used with generate.
check_conflictsbooleanCheck for conflicting skills. Used with generate.
skill_idstringSkill ID. Used with versions, audit_log.
rollback_tonumberVersion number. Used with rollback.
limitnumberMaximum number of results.
daysnumberLookback days. Used with analyze.
include_conflictsbooleanDetect skill conflicts. Used with analyze.

list

List skills with optional filters, or retrieve a single skill by ID.

Parameters: id, domain, status, limit

{
  "name": "skill",
  "arguments": {
    "action": "list",
    "domain": "client-reporting",
    "status": "active"
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "2 active skills in client-reporting:\n\n1. Weekly Client Report (skill_abc123)\n   3-step report workflow: pull data, format template, send to [email protected]\n   Triggers: 'client report', 'weekly report'\n   Success rate: 87.5% (8 uses)\n\n2. Metrics Dashboard Export (skill_def456)\n   Export and format Plausible analytics for client decks\n   Triggers: 'export metrics', 'dashboard data'\n   Success rate: 100% (3 uses)"
    }
  ]
}

create

Create a new skill in draft status. Skills require approval before becoming active.

Parameters: name (required), description, domain, instructions, triggers, templates

{
  "name": "skill",
  "arguments": {
    "action": "create",
    "name": "PR Review Checklist",
    "description": "Standard code review checklist for the frontend team",
    "domain": ["engineering", "code-review"],
    "instructions": "1. Check for TypeScript errors\n2. Verify test coverage above 80%\n3. Review accessibility attributes\n4. Confirm no console.log statements\n5. Validate responsive breakpoints",
    "triggers": ["review PR", "code review", "check this PR"]
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Skill created: PR Review Checklist (skill_gh789)\nStatus: draft — approve to activate.\nDomains: engineering, code-review\nTriggers: review PR, code review, check this PR"
    }
  ]
}

update

Update a skill's content or change its status (approve, reject, deprecate).

Parameters: id (required), plus any fields to update: name, description, instructions, triggers, templates, status, review_notes, reason, successor_id

{
  "name": "skill",
  "arguments": {
    "action": "update",
    "id": "skill_gh789",
    "status": "active",
    "review_notes": "Reviewed, checklist covers our current standards"
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Skill skill_gh789 (PR Review Checklist) updated: status → active. Version 2 created."
    }
  ]
}

generate

Generate a skill from a set of related memories. Kyew synthesizes the memories into structured instructions.

Parameters: memory_ids (required), domain, check_conflicts

{
  "name": "skill",
  "arguments": {
    "action": "generate",
    "memory_ids": ["mem_a1b2c3", "mem_d4e5f6", "mem_g7h8i9"],
    "domain": ["infrastructure"],
    "check_conflicts": true
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Skill generated: Deploy Checklist (skill_x1y2z3)\nStatus: draft\nBased on 3 memories about deployment failures and fixes.\n\nInstructions:\n1. Run npm run build:worker to catch type errors\n2. Run npm run typecheck\n3. Apply database migrations with db:migrate:remote before deploying worker\n4. Deploy with --env production flag\n\nNo conflicts detected with existing skills."
    }
  ]
}

versions

View the version history of a skill.

Parameters: skill_id (required), limit

{
  "name": "skill",
  "arguments": {
    "action": "versions",
    "skill_id": "skill_abc123"
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Version history for skill_abc123 (Weekly Client Report):\n\nv3 (current) — 2026-03-10\n  Updated metrics source from Google Analytics to Plausible\n\nv2 — 2026-02-15\n  Added NPS score step\n\nv1 — 2026-01-08\n  Initial version generated from 4 memories"
    }
  ]
}

rollback

Roll a skill back to a previous version.

Parameters: id (required), rollback_to (required)

{
  "name": "skill",
  "arguments": {
    "action": "rollback",
    "id": "skill_abc123",
    "rollback_to": 2
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Skill skill_abc123 (Weekly Client Report) rolled back to version 2. Current version is now v4 (copy of v2)."
    }
  ]
}

audit_log

View the audit log for a skill, showing all status changes and updates.

Parameters: skill_id (required), limit

{
  "name": "skill",
  "arguments": {
    "action": "audit_log",
    "skill_id": "skill_abc123"
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Audit log for skill_abc123 (Weekly Client Report):\n\n2026-03-10 — updated (v2 → v3): Changed metrics source\n2026-03-01 — reflect: success, 'Report format confirmed by client'\n2026-02-15 — updated (v1 → v2): Added NPS score\n2026-01-08 — created: Generated from 4 memories\n2026-01-08 — status: draft → active, notes: 'Reviewed, matches workflow'"
    }
  ]
}

analyze

Detect patterns across recent memories that could become new skills. Optionally detect conflicts between existing skills.

Parameters: domain, days, limit, include_conflicts

{
  "name": "skill",
  "arguments": {
    "action": "analyze",
    "domain": "engineering",
    "days": 30,
    "include_conflicts": true
  }
}
{
  "content": [
    {
      "type": "text",
      "text": "Analysis for engineering (last 30 days):\n\nEmerging patterns:\n1. PR Review Process — 4 related memories, confidence 0.82\n   Memories: mem_a1, mem_b2, mem_c3, mem_d4\n   Suggestion: Generate skill with action=generate\n\n2. Incident Response — 3 related memories, confidence 0.71\n   Memories: mem_e5, mem_f6, mem_g7\n   Suggestion: 1 more observation before generation\n\nConflicts:\n- 'Deploy Checklist' (skill_x1y2z3) and 'Release Process' (skill_h8i9j0) have overlapping triggers"
    }
  ]
}
Previous
session
Next
task