Class: KairosMcp::KnowledgeProvider

Inherits:
Object
  • Object
show all
Defined in:
lib/kairos_mcp/knowledge_provider.rb

Overview

KnowledgeProvider: Manages L1 (knowledge layer) skills in Anthropic format

L1 characteristics:

  • Project-specific universal knowledge
  • Hash-only blockchain recording
  • Lightweight modification constraints
  • Folder-based archiving (.archived/ directory)

Storage:

  • Content (*.md files): Always stored in files for human readability
  • Metadata: Stored in files (default) or SQLite (when sqlite backend enabled)
  • Blockchain: Uses the configured storage backend

Constant Summary collapse

ARCHIVED_DIR =
'.archived'
ARCHIVE_META_FILE =
'.archive_meta.yml'
BACKUP_DIR_PATTERN =

Backup directories created by upgrade flow (.bak.<timestamp>). Loader must skip these — they may contain old/broken frontmatter.

/(?:^|\.)bak(?:\.|$)/.freeze

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(knowledge_dir = nil, vector_search_enabled: true, storage_backend: nil, user_context: nil, include_skillset_knowledge: true) ⇒ KnowledgeProvider

Initialize the KnowledgeProvider

Parameters:

  • knowledge_dir (String) (defaults to: nil)

    Path to knowledge directory

  • vector_search_enabled (Boolean) (defaults to: true)

    Enable vector search

  • storage_backend (Storage::Backend, nil) (defaults to: nil)

    Storage backend to use

  • include_skillset_knowledge (Boolean) (defaults to: true)

    Register knowledge dirs declared by enabled SkillSets. Default true; pass false to build a provider scoped strictly to the main knowledge dir.



46
47
48
49
50
51
52
53
54
55
56
57
58
# File 'lib/kairos_mcp/knowledge_provider.rb', line 46

def initialize(knowledge_dir = nil, vector_search_enabled: true, storage_backend: nil,
               user_context: nil, include_skillset_knowledge: true)
  knowledge_dir ||= KairosMcp.knowledge_dir(user_context: user_context)
  @knowledge_dir = knowledge_dir
  @user_context = user_context
  @vector_search_enabled = vector_search_enabled
  @storage_backend = storage_backend
  @vector_search = nil
  @index_built = false
  @external_dirs = []
  FileUtils.mkdir_p(@knowledge_dir)
  register_skillset_knowledge_dirs if include_skillset_knowledge
end

Instance Attribute Details

#knowledge_dirObject (readonly)

Main knowledge directory (constitutively-recorded L1). Exposed so callers can distinguish main-dir knowledge from read-only external SkillSet knowledge, e.g. to scope INV-A correspondence checks to recorded artifacts.



30
31
32
# File 'lib/kairos_mcp/knowledge_provider.rb', line 30

def knowledge_dir
  @knowledge_dir
end

Instance Method Details

#add_external_dir(dir, source:, layer: :L1, index: true, only: nil) ⇒ Object

Register an external knowledge directory (e.g. from a SkillSet) Knowledge is read-only from external dirs; no merge into the main dir.

Idempotent by directory: registering the same dir twice (e.g. once from the SkillSet manifest at construction, once from a SkillSet that also registers itself) keeps the first registration rather than duplicating list entries.

Parameters:

  • dir (String)

    Absolute path to the container directory holding entry subdirectories (e.g. <skillset>/knowledge), not an entry directory itself

  • source (String)

    Identifier for the source (e.g. "skillset:mmp")

  • layer (Symbol) (defaults to: :L1)

    Layer governance (:L0, :L1, :L2)

  • index (Boolean) (defaults to: true)

    Whether to include in vector search index

  • only (Array<String>, nil) (defaults to: nil)

    Entry names to expose. nil exposes every subdirectory. Passing the declared set keeps undeclared knowledge shipped inside a SkillSet from becoming visible as L1 by proximity alone.



75
76
77
78
79
80
81
82
83
84
85
86
# File 'lib/kairos_mcp/knowledge_provider.rb', line 75

def add_external_dir(dir, source:, layer: :L1, index: true, only: nil)
  return unless File.directory?(dir)

  absolute = File.expand_path(dir)
  return if @external_dirs.any? { |ext| ext[:dir] == absolute }

  @external_dirs << {
    dir: absolute, source: source, layer: layer, index: index,
    only: only && Array(only).map { |n| File.basename(n.to_s) }
  }
  @index_built = false if index # Invalidate index when new indexed dir added
end

#archive(name, reason:, superseded_by: nil) ⇒ Hash

Archive a knowledge skill (move to .archived/ directory)

Parameters:

  • name (String)

    Skill name

  • reason (String)

    Reason for archiving

  • superseded_by (String, nil) (defaults to: nil)

    Name of the knowledge that supersedes this one

Returns:

  • (Hash)

    Result with success status



385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
# File 'lib/kairos_mcp/knowledge_provider.rb', line 385

def archive(name, reason:, superseded_by: nil)
  skill = get(name)
  unless skill
    return { success: false, error: "Knowledge '#{name}' not found" }
  end

  # Check if already archived
  if archived?(name)
    return { success: false, error: "Knowledge '#{name}' is already archived" }
  end

  # Create archive directory
  archived_dir = File.join(@knowledge_dir, ARCHIVED_DIR)
  FileUtils.mkdir_p(archived_dir)

  # Calculate hash before moving
  content = File.read(skill.md_file_path)
  content_hash = Digest::SHA256.hexdigest(content)

  # Move to archive
  dest_path = File.join(archived_dir, name)
  FileUtils.mv(skill.base_path, dest_path)

  # Create archive metadata file
  meta = {
    'archived_at' => Time.now.iso8601,
    'archived_reason' => reason,
    'superseded_by' => superseded_by,
    'original_path' => skill.base_path,
    'content_hash' => content_hash
  }
  File.write(File.join(dest_path, ARCHIVE_META_FILE), meta.to_yaml)

  # Record to blockchain
  record_hash_reference(
    name: name,
    action: 'archive',
    prev_hash: content_hash,
    next_hash: nil,
    reason: reason
  )

  # Remove from vector search index
  remove_from_vector_index(name)

  # Track pending change for state commit (archive = demotion)
  track_pending_change(layer: 'L1', action: 'archive', skill_id: name, reason: reason)

  { success: true, archived: name, path: dest_path, hash: content_hash }
rescue StandardError => e
  { success: false, error: "Archive failed: #{e.message}" }
end

#archived?(name) ⇒ Boolean

Check if a knowledge skill is archived

Parameters:

  • name (String)

    Skill name

Returns:

  • (Boolean)

    True if archived



541
542
543
544
# File 'lib/kairos_mcp/knowledge_provider.rb', line 541

def archived?(name)
  archived_path = File.join(@knowledge_dir, ARCHIVED_DIR, name)
  File.directory?(archived_path)
end

#create(name, content, reason: nil, create_subdirs: false) ⇒ Hash

Create a new knowledge skill

Parameters:

  • name (String)

    Skill name

  • content (String)

    Full content including YAML frontmatter

  • reason (String) (defaults to: nil)

    Reason for creation

  • create_subdirs (Boolean) (defaults to: false)

    Whether to create scripts/assets/references

Returns:

  • (Hash)

    Result with success status and skill info



163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
# File 'lib/kairos_mcp/knowledge_provider.rb', line 163

def create(name, content, reason: nil, create_subdirs: false)
  skill_dir = File.join(@knowledge_dir, name)
  
  if File.exist?(skill_dir)
    return { success: false, error: "Knowledge '#{name}' already exists" }
  end

  skill = AnthropicSkillParser.create(@knowledge_dir, name, content, create_subdirs: create_subdirs)
  
  # Record hash reference to blockchain
  content_hash = Digest::SHA256.hexdigest(content)
  record_hash_reference(
    name: name,
    action: 'create',
    prev_hash: nil,
    next_hash: content_hash,
    reason: reason || "Create knowledge: #{name}"
  )

  # Update vector search index
  update_vector_index(name, content, skill)

  # Track pending change for state commit
  track_pending_change(layer: 'L1', action: 'create', skill_id: name, reason: reason)

  { success: true, skill: skill.to_h, hash: content_hash, next_hash: content_hash }
end

#delete(name, reason: nil) ⇒ Hash

Delete a knowledge skill

Parameters:

  • name (String)

    Skill name

  • reason (String) (defaults to: nil)

    Reason for deletion

Returns:

  • (Hash)

    Result with success status



238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
# File 'lib/kairos_mcp/knowledge_provider.rb', line 238

def delete(name, reason: nil)
  skill = get(name)
  unless skill
    return { success: false, error: "Knowledge '#{name}' not found" }
  end

  # Calculate hash before deletion
  prev_content = File.read(skill.md_file_path)
  prev_hash = Digest::SHA256.hexdigest(prev_content)

  # Delete the directory
  FileUtils.rm_rf(skill.base_path)

  # Record hash reference to blockchain
  record_hash_reference(
    name: name,
    action: 'delete',
    prev_hash: prev_hash,
    next_hash: nil,
    reason: reason || "Delete knowledge: #{name}"
  )

  # Remove from vector search index
  remove_from_vector_index(name)

  # Track pending change for state commit
  track_pending_change(layer: 'L1', action: 'delete', skill_id: name, reason: reason)

  { success: true, deleted: name, prev_hash: prev_hash }
end

#get(name) ⇒ AnthropicSkillParser::SkillEntry?

Get a specific knowledge skill by name Searches main knowledge dir first, then external SkillSet dirs

Parameters:

  • name (String)

    Skill name

Returns:



141
142
143
144
145
146
147
148
149
150
151
152
153
154
# File 'lib/kairos_mcp/knowledge_provider.rb', line 141

def get(name)
  skill_dir = File.join(@knowledge_dir, name)
  return AnthropicSkillParser.parse(skill_dir) if File.directory?(skill_dir)

  # Search external directories
  @external_dirs.each do |ext|
    next if ext[:only] && !ext[:only].include?(name)

    ext_skill_dir = File.join(ext[:dir], name)
    return AnthropicSkillParser.parse(ext_skill_dir) if File.directory?(ext_skill_dir)
  end

  nil
end

#get_archived(name) ⇒ Hash?

Get a specific archived knowledge skill

Parameters:

  • name (String)

    Skill name

Returns:

  • (Hash, nil)

    Archived skill info or nil



518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
# File 'lib/kairos_mcp/knowledge_provider.rb', line 518

def get_archived(name)
  archived_path = File.join(@knowledge_dir, ARCHIVED_DIR, name)
  return nil unless File.directory?(archived_path)

  skill = AnthropicSkillParser.parse(archived_path)
  return nil unless skill

  meta_file = File.join(archived_path, ARCHIVE_META_FILE)
  meta = File.exist?(meta_file) ? YAML.safe_load(File.read(meta_file)) : {}

  {
    skill: skill.to_h,
    archived_at: meta['archived_at'],
    archived_reason: meta['archived_reason'],
    superseded_by: meta['superseded_by'],
    content_hash: meta['content_hash']
  }
end

#listArray<Hash>

List all knowledge skills (including those from external SkillSet dirs)

Returns:

  • (Array<Hash>)

    List of knowledge skill summaries



97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
# File 'lib/kairos_mcp/knowledge_provider.rb', line 97

def list
  results = skill_dirs.map do |dir|
    skill = AnthropicSkillParser.parse(dir)
    next unless skill

    {
      name: skill.name,
      description: skill.description,
      version: skill.version,
      tags: skill.tags,
      has_scripts: skill.has_scripts?,
      has_assets: skill.has_assets?,
      has_references: skill.has_references?
    }
  end.compact

  # Include knowledge from external directories (SkillSets)
  @external_dirs.each do |ext|
    external_skill_dirs(ext[:dir], only: ext[:only]).each do |dir|
      skill = AnthropicSkillParser.parse(dir)
      next unless skill

      results << {
        name: skill.name,
        description: skill.description,
        version: skill.version,
        tags: skill.tags,
        has_scripts: skill.has_scripts?,
        has_assets: skill.has_assets?,
        has_references: skill.has_references?,
        source: ext[:source],
        layer: ext[:layer]
      }
    end
  end

  results
end

#list_archivedArray<Hash>

List all archived knowledge skills

Returns:

  • (Array<Hash>)

    List of archived knowledge summaries



494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
# File 'lib/kairos_mcp/knowledge_provider.rb', line 494

def list_archived
  archived_dir = File.join(@knowledge_dir, ARCHIVED_DIR)
  return [] unless File.directory?(archived_dir)

  Dir[File.join(archived_dir, '*')].select { |f| File.directory?(f) }.map do |dir|
    skill = AnthropicSkillParser.parse(dir)
    meta_file = File.join(dir, ARCHIVE_META_FILE)
    meta = File.exist?(meta_file) ? YAML.safe_load(File.read(meta_file)) : {}

    {
      name: skill&.name || File.basename(dir),
      description: skill&.description,
      archived_at: meta['archived_at'],
      archived_reason: meta['archived_reason'],
      superseded_by: meta['superseded_by'],
      content_hash: meta['content_hash']
    }
  end
end

#list_assets(name) ⇒ Array<Hash>

List assets in a knowledge skill

Parameters:

  • name (String)

    Skill name

Returns:

  • (Array<Hash>)

    List of asset info



284
285
286
287
288
289
# File 'lib/kairos_mcp/knowledge_provider.rb', line 284

def list_assets(name)
  skill = get(name)
  return [] unless skill

  AnthropicSkillParser.list_assets(skill)
end

#list_references(name) ⇒ Array<Hash>

List references in a knowledge skill

Parameters:

  • name (String)

    Skill name

Returns:

  • (Array<Hash>)

    List of reference info



295
296
297
298
299
300
# File 'lib/kairos_mcp/knowledge_provider.rb', line 295

def list_references(name)
  skill = get(name)
  return [] unless skill

  AnthropicSkillParser.list_references(skill)
end

#list_scripts(name) ⇒ Array<Hash>

List scripts in a knowledge skill

Parameters:

  • name (String)

    Skill name

Returns:

  • (Array<Hash>)

    List of script info



273
274
275
276
277
278
# File 'lib/kairos_mcp/knowledge_provider.rb', line 273

def list_scripts(name)
  skill = get(name)
  return [] unless skill

  AnthropicSkillParser.list_scripts(skill)
end

#rebuild_indexBoolean

Rebuild the vector search index (includes indexed external dirs)

Returns:

  • (Boolean)

    Success status



333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
# File 'lib/kairos_mcp/knowledge_provider.rb', line 333

def rebuild_index
  documents = skill_dirs.filter_map do |dir|
    skill = AnthropicSkillParser.parse(dir)
    next unless skill

    content = File.read(skill.md_file_path) rescue ''
    {
      id: skill.name,
      text: build_searchable_text(skill, content),
      metadata: {
        description: skill.description,
        tags: skill.tags,
        version: skill.version
      }
    }
  end

  # Include external dirs that have indexing enabled
  @external_dirs.select { |ext| ext[:index] }.each do |ext|
    external_skill_dirs(ext[:dir], only: ext[:only]).each do |dir|
      skill = AnthropicSkillParser.parse(dir)
      next unless skill

      content = File.read(skill.md_file_path) rescue ''
      documents << {
        id: "#{ext[:source]}:#{skill.name}",
        text: build_searchable_text(skill, content),
        metadata: {
          description: skill.description,
          tags: skill.tags,
          version: skill.version,
          source: ext[:source]
        }
      }
    end
  end

  result = vector_search.rebuild(documents)
  @index_built = result
  result
end

#search(query, max_results = 5, semantic: nil) ⇒ Array<Hash>

Search knowledge skills by query

Parameters:

  • query (String)

    Search query

  • max_results (Integer) (defaults to: 5)

    Maximum number of results

  • semantic (Boolean) (defaults to: nil)

    Force semantic search if available

Returns:

  • (Array<Hash>)

    Matching skills



308
309
310
311
312
313
314
315
316
# File 'lib/kairos_mcp/knowledge_provider.rb', line 308

def search(query, max_results = 5, semantic: nil)
  use_semantic = semantic.nil? ? @vector_search_enabled : semantic
  
  if use_semantic && vector_search.semantic?
    semantic_search(query, max_results)
  else
    regex_search(query, max_results)
  end
end

#storage_typeSymbol

Get the storage backend type

Returns:

  • (Symbol)

    :file or :sqlite



90
91
92
# File 'lib/kairos_mcp/knowledge_provider.rb', line 90

def storage_type
  storage_backend.backend_type
end

#unarchive(name, reason:) ⇒ Hash

Unarchive a knowledge skill (restore from .archived/ directory)

Parameters:

  • name (String)

    Skill name

  • reason (String)

    Reason for unarchiving

Returns:

  • (Hash)

    Result with success status



443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
# File 'lib/kairos_mcp/knowledge_provider.rb', line 443

def unarchive(name, reason:)
  archived_path = File.join(@knowledge_dir, ARCHIVED_DIR, name)

  unless File.directory?(archived_path)
    return { success: false, error: "Archived knowledge '#{name}' not found" }
  end

  # Check if active knowledge with same name exists
  active_path = File.join(@knowledge_dir, name)
  if File.directory?(active_path)
    return { success: false, error: "Active knowledge '#{name}' already exists. Rename or delete it first." }
  end

  # Read archive metadata
  meta_file = File.join(archived_path, ARCHIVE_META_FILE)
  meta = File.exist?(meta_file) ? YAML.safe_load(File.read(meta_file)) : {}

  # Move back to active
  FileUtils.mv(archived_path, active_path)

  # Remove archive metadata file
  FileUtils.rm_f(File.join(active_path, ARCHIVE_META_FILE))

  # Parse the restored skill
  skill = AnthropicSkillParser.parse(active_path)
  content = File.read(skill.md_file_path)
  content_hash = Digest::SHA256.hexdigest(content)

  # Record to blockchain
  record_hash_reference(
    name: name,
    action: 'unarchive',
    prev_hash: meta['content_hash'],
    next_hash: content_hash,
    reason: reason
  )

  # Update vector search index
  update_vector_index(name, content, skill)

  # Track pending change for state commit
  track_pending_change(layer: 'L1', action: 'unarchive', skill_id: name, reason: reason)

  { success: true, unarchived: name, path: active_path, hash: content_hash }
rescue StandardError => e
  { success: false, error: "Unarchive failed: #{e.message}" }
end

#update(name, new_content, reason: nil) ⇒ Hash

Update an existing knowledge skill

Parameters:

  • name (String)

    Skill name

  • new_content (String)

    New content including YAML frontmatter

  • reason (String) (defaults to: nil)

    Reason for update

Returns:

  • (Hash)

    Result with success status



197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
# File 'lib/kairos_mcp/knowledge_provider.rb', line 197

def update(name, new_content, reason: nil)
  skill = get(name)
  unless skill
    return { success: false, error: "Knowledge '#{name}' not found" }
  end

  # Calculate hashes
  prev_content = File.read(skill.md_file_path)
  prev_hash = Digest::SHA256.hexdigest(prev_content)
  next_hash = Digest::SHA256.hexdigest(new_content)

  if prev_hash == next_hash
    return { success: false, error: "No changes detected" }
  end

  # Update the file
  updated_skill = AnthropicSkillParser.update(skill.base_path, new_content)

  # Record hash reference to blockchain
  record_hash_reference(
    name: name,
    action: 'update',
    prev_hash: prev_hash,
    next_hash: next_hash,
    reason: reason || "Update knowledge: #{name}"
  )

  # Update vector search index
  update_vector_index(name, new_content, updated_skill)

  # Track pending change for state commit
  track_pending_change(layer: 'L1', action: 'update', skill_id: name, reason: reason)

  { success: true, skill: updated_skill.to_h, prev_hash: prev_hash, next_hash: next_hash }
end

#vector_search_statusHash

Get vector search status

Returns:

  • (Hash)

    Status information



321
322
323
324
325
326
327
328
# File 'lib/kairos_mcp/knowledge_provider.rb', line 321

def vector_search_status
  {
    enabled: @vector_search_enabled,
    semantic_available: VectorSearch.available?,
    index_built: @index_built,
    document_count: vector_search.count
  }
end