@dzhechkov/keysarium 1.2.0 → 1.4.0

package/templates/.claude/skills/knowledge-extractor/references/artifact-categories.md

@@ -0,0 +1,100 @@
+# Artifact Categories Reference
+
+Seven mutually exclusive categories for classifying extracted findings.
+Every finding must be assigned exactly one category before passing through the quality gates.
+If a finding spans multiple categories, split it or assign the primary category.
+
+## Category Taxonomy
+
+| Category | Destination | Description | Examples | Agent Lens |
+|----------|-------------|-------------|----------|------------|
+| skills | .claude/skills/<name>/ | Complete skill bundles (SKILL.md + modules) | New analysis methodology, reusable research approach | All lenses (cross-cutting) |
+| commands | .claude/commands/<name>.md | Slash commands for Claude Code | Pipeline phases, utility commands | extractor-commands |
+| hooks | .claude/hooks/ or TOOLKIT_HARVEST.md | Event-driven automations, pre/post hooks | Pre-commit checks, post-phase triggers | extractor-commands |
+| rules | .claude/rules/<name>.md | Constraints, quality gates, anti-patterns | Domain restrictions, naming conventions | extractor-rules |
+| templates | TOOLKIT_HARVEST.md or lib/ | Document/config structure templates | ADR template, mermaid diagram template | extractor-templates |
+| patterns | TOOLKIT_HARVEST.md or lib/ | Architecture and design patterns | Agent swarm topology, event-driven pipeline | extractor-patterns |
+| snippets | TOOLKIT_HARVEST.md | Reusable code fragments | React components, utility functions, bash one-liners | extractor-snippets |
+
+## Category Descriptions
+
+### skills
+A skill is a self-contained capability bundle that can be loaded by a Claude Code session.
+It requires at minimum a SKILL.md with trigger conditions, usage instructions, and output specification.
+Extract as a skill when the finding represents a repeatable methodology that warrants its own invocation pattern.
+Destination: create a new directory under `.claude/skills/<name>/` with SKILL.md as the entry point.
+
+### commands
+A command is a slash command definition — a markdown file in `.claude/commands/` that Claude Code exposes as `/command-name`.
+Extract as a command when the finding defines a distinct user-facing workflow step or utility.
+Destination: `.claude/commands/<name>.md`
+
+### hooks
+A hook is an event-driven automation that fires on a specific lifecycle event (session start, phase completion, file save).
+Extract as a hook when the finding describes behavior that should trigger automatically, not on user request.
+Destination: `.claude/hooks/` directory if the project uses the hooks directory convention; otherwise document in TOOLKIT_HARVEST.md under the Hooks section.
+
+### rules
+A rule is a constraint, convention, or quality gate that governs agent behavior.
+Extract as a rule when the finding specifies what agents must or must not do, regardless of context.
+Destination: `.claude/rules/<name>.md`
+
+### templates
+A template is a document or configuration scaffold with placeholders that gets filled in for each use.
+Extract as a template when the finding defines a repeatable document structure (not the content itself).
+Destination: `lib/` if it will be referenced by multiple commands; otherwise TOOLKIT_HARVEST.md under Templates section.
+
+### patterns
+A pattern is an architectural or design solution to a recurring structural problem.
+Extract as a pattern when the finding describes how to compose components or agents, not what they contain.
+Patterns are higher-level than snippets and lower-level than skills.
+Destination: `lib/` if it has widespread applicability; otherwise TOOLKIT_HARVEST.md under Patterns section.
+
+### snippets
+A snippet is a concrete, copy-paste-ready code or configuration fragment.
+Extract as a snippet when the finding is language/framework-specific code that solves one small problem.
+Destination: TOOLKIT_HARVEST.md under Snippets section, with language and framework tags.
+
+## Scope Filtering Mapping (category to agent lens for /harvest only X)
+
+When the user invokes `/harvest only <scope>`, only the corresponding extractor agent lens runs.
+All other lenses are suppressed to reduce cost and noise.
+
+| User Scope Argument | Active Agent Lenses | Categories Produced |
+|--------------------|---------------------|---------------------|
+| `skills` | ALL 5 lenses (cross-cutting) | skills |
+| `commands` | extractor-commands | commands, hooks |
+| `rules` | extractor-rules | rules |
+| `templates` | extractor-templates | templates |
+| `patterns` | extractor-patterns | patterns |
+| `snippets` | extractor-snippets | snippets |
+| `code` | extractor-snippets | snippets |
+| `all` (default) | All 5 lenses | All 7 categories |
+
+Note: hooks are produced by extractor-commands because hooks are closely related to command lifecycle automation.
+Note: skills scope spawns ALL 5 lenses because skills are cross-cutting — any lens may discover patterns worthy of promotion to a full skill bundle.
+
+## Deduplication Rules (per-category)
+
+Deduplication is applied during gate G6 (Not Duplicate). The similarity threshold is 0.85.
+Similarity is assessed by the orchestrator using structural comparison (not LLM) unless specified.
+
+| Category | Deduplication Key | Comparison Method |
+|----------|------------------|-------------------|
+| skills | Skill name + trigger conditions | Exact name match OR semantic trigger overlap |
+| commands | Command name (slug) | Exact slash-command name match |
+| hooks | Event type + trigger condition | Exact event type + condition string match |
+| rules | Rule title + first sentence of constraint | Normalized string similarity |
+| templates | Template name + section headings list | Section heading set overlap |
+| patterns | Pattern name + problem statement | Normalized string similarity on problem statement |
+| snippets | Language + first 5 non-whitespace lines | Normalized code similarity |
+
+When a near-duplicate is found (similarity 0.70–0.85), the extractor must decide:
+- If the new finding improves the existing entry (adds a section, fixes an error): propose a MERGE, not a new entry.
+- If the new finding is a different context of the same idea: add as a "variant" note under the existing entry.
+- If similarity >= 0.85: treat as duplicate, reject via G6.
+
+Merge proposals are surfaced in the harvest output as:
+```
+MERGE PROPOSAL: Finding #{X} improves existing entry "{existing_name}" — {reason}
+```

package/templates/.claude/skills/knowledge-extractor/references/maturity-model.md

@@ -0,0 +1,94 @@
+# Maturity Model Reference
+
+Three maturity levels are used to communicate how well-tested and battle-proven an extracted finding is.
+Every finding must be assigned a maturity level at extraction time. The level is surfaced in the artifact card
+and in TOOLKIT_HARVEST.md entries so consumers know how much trust to place in the finding.
+
+## Maturity Levels
+
+| Level | Label | Criteria | Badge |
+|-------|-------|----------|-------|
+| Alpha | Alpha | First extraction, untested. Used in exactly 1 project. No validation. | 🔴 |
+| Beta | Beta | Used in 2+ projects. Basic validation. May have rough edges. | 🟡 |
+| Stable | Stable | Used in 3+ projects. Proven track record. Passed /bto-test or equivalent. | 🟢 |
+
+## Promotion Protocol
+
+Promotion moves a finding from one maturity level to the next. Promotion requires evidence — it is not
+self-declared. The `/harvest` command can suggest promotion but not enforce it; the user confirms.
+
+### Alpha to Beta
+
+Required evidence (any one of the following):
+- The same pattern, rule, template, or snippet was independently observed in a second project's harvest.
+- A team member explicitly applied the finding in a second project and confirmed it worked.
+- The finding was referenced in a command or skill that was used in a second research session.
+
+Promotion action:
+1. Update the `maturity` field in the finding's TOOLKIT_HARVEST.md entry from `alpha` to `beta`.
+2. Add a `used_in` list to the entry noting the two (or more) projects.
+3. Log the promotion: `PROMOTED: "{name}" alpha -> beta — evidence: {source}`.
+
+### Beta to Stable
+
+Required evidence (ALL of the following):
+- Used in 3 or more distinct projects (verified by `used_in` list).
+- At least one of the following validation signals exists:
+  - Passed `/bto-test` Layer 2 evaluation with score >= 7.0.
+  - Has a documented test case in a `references/` or `examples/` directory.
+  - Was reviewed and approved by a second human reviewer in a harvest session.
+
+Promotion action:
+1. Update the `maturity` field from `beta` to `stable`.
+2. Record the validation signal: BTO score, test case path, or reviewer name.
+3. Add the finding to the summary table in TOOLKIT_HARVEST.md under the Stable Findings section.
+4. Log the promotion: `PROMOTED: "{name}" beta -> stable — validation: {signal}`.
+
+### Demotion Protocol
+
+A stable or beta finding may be demoted if evidence of unreliability emerges:
+- A project reports the finding caused errors or required significant modification.
+- A `/bto-test` run scores the finding below 5.0.
+
+Demotion action:
+1. Downgrade maturity by one level (stable -> beta, beta -> alpha).
+2. Add a `demotion_reason` note to the TOOLKIT_HARVEST.md entry.
+3. Log the demotion: `DEMOTED: "{name}" {from} -> {to} — reason: {reason}`.
+
+## Initial Assignment Rules
+
+1. The first extraction of any finding is ALWAYS assigned `alpha` maturity.
+2. Exception: if the finding describes a pattern that was already in use across multiple projects
+   prior to being formally extracted, it may be assigned `beta` at extraction time — but only if
+   the extractor can cite at least 2 distinct prior uses in the finding's `used_in` field.
+3. Exception: if the extracted finding is a verbatim copy of content that already exists in the
+   toolkit at beta or stable maturity (e.g., an updated version of a known pattern), inherit the
+   existing maturity level and propose a MERGE rather than a new entry.
+4. No finding may be assigned `stable` at first extraction, regardless of prior use.
+
+## Maturity in Artifact Card
+
+Every finding written to TOOLKIT_HARVEST.md includes a maturity badge in its header card.
+
+### Format
+
+```
+## [Category] — {finding_name}
+**Maturity:** [ALPHA | BETA | STABLE]
+**Confidence:** {confidence_score}
+**Used in:** {project_slug_1}, {project_slug_2}, ...
+**Validated by:** {bto_score or test_path or reviewer or "not yet validated"}
+```
+
+### Display Rules
+
+- Display maturity label in uppercase.
+- If `used_in` has only one entry, display it as `{slug} (source project)`.
+- If `validated_by` is empty, display `not yet validated`.
+- Include the maturity badge on the same line as the finding name when embedding in summary tables:
+
+```
+| {finding_name} [ALPHA] | {category} | {one-line description} |
+```
+
+- In summary tables, badge order is: STABLE first, then BETA, then ALPHA (most trustworthy at top).

package/templates/.claude/skills/knowledge-extractor/references/quality-gates.md

@@ -0,0 +1,98 @@
+# Quality Gates Reference
+
+Gates applied to every candidate finding during the harvest process.
+All 8 gates must be evaluated before a finding can be accepted into the toolkit.
+A gate failure is a hard block unless overridden by the user (see Override Protocol).
+
+## Gate Definitions
+
+| Gate | Name | Check | Type | Pass Criteria | Failure Message |
+|------|------|-------|------|---------------|-----------------|
+| G1 | When-To-Use | Finding has a "When to use" section or equivalent guidance | deterministic | Section exists and is non-empty | Missing "When to use" guidance |
+| G2 | When-Not-To-Use | Finding has a "When NOT to use" section or equivalent anti-guidance | deterministic | Section exists and is non-empty | Missing "When NOT to use" guidance |
+| G3 | Decontextualized | No project-specific names, paths, slugs, dates, or hardcoded values remain | semantic | Haiku confirms no project-specific references | Project-specific references remain: {details} |
+| G4 | Concrete Example | At least one concrete usage example is provided | semantic | Haiku confirms example is concrete and actionable | No concrete example provided |
+| G5 | Reusability Threshold | Reusability confidence score meets minimum | deterministic | confidence >= 0.5 | Confidence {score} < 0.5 threshold |
+| G6 | Not Duplicate | Finding is not a duplicate of an existing toolkit entry | deterministic | No existing entry with content similarity > 0.85 | Duplicate of existing entry: {existing_name} |
+| G7 | Maturity Label | Finding has a maturity level assigned | deterministic | maturity field is one of: alpha, beta, stable | Missing maturity label |
+| G8 | Brutal Honesty | Finding survives self-critical review — is it genuinely useful or just noise? | semantic | Haiku confirms finding has genuine reuse value beyond the source project | Fails brutal-honesty review: {reason} |
+
+## Gate Types
+
+**Deterministic gates** (G1, G2, G5, G6, G7) are checked by the orchestrator without LLM calls.
+They operate on the structured finding JSON: presence of fields, numeric thresholds, set membership.
+
+**Semantic gates** (G3, G4, G8) require a haiku LLM call with a focused single-question prompt.
+The prompt must ask for a binary YES/NO answer plus a short explanation (1-2 sentences max).
+Use haiku — these are high-frequency, pattern-matching evaluations that do not require deep reasoning.
+
+## Pass Assignment (which pass each gate belongs to)
+
+Gates are evaluated in two passes to minimize unnecessary LLM cost.
+
+### Pass 1 — Deterministic (zero LLM cost)
+
+Run all deterministic gates first. If any deterministic gate fails, reject immediately.
+Do NOT proceed to Pass 2 for a rejected finding.
+
+Gates in Pass 1 (in evaluation order):
+1. G7 — Maturity Label (cheapest check: field presence)
+2. G5 — Reusability Threshold (numeric comparison)
+3. G1 — When-To-Use (section presence)
+4. G2 — When-Not-To-Use (section presence)
+5. G6 — Not Duplicate (similarity comparison against existing entries)
+
+### Pass 2 — Semantic (haiku LLM calls, parallelizable)
+
+Run only after all Pass 1 gates pass.
+Gates G3, G4, G8 may be evaluated in parallel (3 concurrent haiku calls).
+
+Gates in Pass 2:
+- G3 — Decontextualized
+- G4 — Concrete Example
+- G8 — Brutal Honesty
+
+## Override Protocol
+
+A user may override a gate for a specific finding using the syntax:
+
+```
+пропусти G3 для #X
+```
+
+Where `#X` is the finding number (1-indexed, matching the extraction output list).
+
+### Override Rules
+
+- Multiple gates may be overridden in a single command using comma-separated syntax: `"пропусти G3,G4 для #14"`.
+- G8 (Brutal Honesty) cannot be overridden. If a finding fails G8, it must be revised or dropped.
+- G6 (Not Duplicate) cannot be overridden. If a finding is a duplicate, either update the existing entry or drop the new one.
+- G5 (Reusability Threshold) can be overridden only if the user provides a rationale explaining why low-confidence is acceptable for this finding.
+- All overrides are logged in the harvest output with the gate ID, finding number, and user rationale.
+
+### Override Log Format
+
+When an override is applied, append to the harvest session log:
+
+```
+OVERRIDE: G{N} skipped for finding #{X} — user rationale: "{rationale}"
+```
+
+## Scoring
+
+After all gates are evaluated for a finding:
+
+```
+gate_pass_rate = gates_passed / 8
+```
+
+This score is recorded in the finding's metadata alongside the per-gate pass/fail results.
+
+| gate_pass_rate | Interpretation |
+|----------------|----------------|
+| 1.0 | Clean pass — add to toolkit |
+| 0.875 (7/8) | One gate overridden — add with override note |
+| < 0.875 | Rejected — do not add to toolkit |
+
+The gate_pass_rate is surfaced in the harvest summary table for each finding.
+It is not used as a quality score for ranking — a finding either passes all required gates or it does not.

package/templates/.claude/skills/knowledge-extractor/templates/artifact-card.md

@@ -0,0 +1,63 @@
+# Artifact Card Template
+
+## Usage
+This template is rendered once per approved finding during Module 04 (Integrate).
+Replace all {PLACEHOLDER} values with finding data.
+
+## Template
+
+---
+
+### {NAME}
+
+**Category:** {CATEGORY}
+**Maturity:** {MATURITY_BADGE} {MATURITY_LEVEL}
+**Reusability Confidence:** {CONFIDENCE}
+**Extracted from:** {SOURCE_PROJECT}
+**Last updated:** {DATE}
+
+#### Description
+
+{DESCRIPTION}
+
+#### When to use
+
+{WHEN_TO_USE}
+
+#### When NOT to use
+
+{WHEN_NOT_TO_USE}
+
+#### Example
+
+{EXAMPLE}
+
+#### Changelog
+
+- v1: Initial extraction from {SOURCE_PROJECT} ({DATE})
+
+---
+
+## Placeholder Reference
+
+| Placeholder | Source | Example |
+|-------------|--------|---------|
+| {NAME} | finding.title | "Agent Swarm Topology" |
+| {CATEGORY} | finding.category | "patterns" |
+| {MATURITY_BADGE} | maturity-model.md lookup | "🔴" |
+| {MATURITY_LEVEL} | finding.maturity | "Alpha" |
+| {CONFIDENCE} | finding.confidence | "0.85" |
+| {SOURCE_PROJECT} | session.slug | "bank-kc-automation" |
+| {DATE} | current date | "2026-03-03" |
+| {DESCRIPTION} | finding.content (first paragraph) | "..." |
+| {WHEN_TO_USE} | finding.when_to_use | "..." |
+| {WHEN_NOT_TO_USE} | finding.when_not_to_use | "..." |
+| {EXAMPLE} | finding.example | "..." |
+
+## Placement Rules
+
+For **skills** category: This template becomes the initial SKILL.md skeleton
+For **commands** category: This template becomes the command .md file body
+For **rules** category: Convert to rule table format (see .claude/rules/ convention)
+For **templates/patterns/snippets** category: This card is appended to TOOLKIT_HARVEST.md under the relevant section
+For **hooks** category: This card is appended to TOOLKIT_HARVEST.md hooks section (or created inline)

package/templates/.claude/skills/knowledge-extractor/templates/harvest-report.md

@@ -0,0 +1,102 @@
+# Harvest Report Template
+
+## Usage
+Generated at the end of Module 04 (Integrate) to summarize the harvest session.
+All {PLACEHOLDER} values are computed from the findings registry.
+
+## Template
+
+---
+
+# Harvest Report: {SESSION_ID}
+
+**Date:** {DATE}
+**Target:** {TARGET_PATHS}
+**Scope:** {SCOPE_FILTER_OR_ALL}
+**Steps completed:** {STEPS_COMPLETED} (Extract → Classify → Checkpoint → Gate → Integrate)
+
+## Summary
+
+| Metric | Value |
+|--------|-------|
+| Total findings extracted | {TOTAL_EXTRACTED} |
+| After deduplication | {AFTER_DEDUP} |
+| User removals | {USER_REMOVALS} |
+| User merges | {USER_MERGES} |
+| Approved (passed gates) | {APPROVED_COUNT} |
+| Blocked (failed gates) | {BLOCKED_COUNT} |
+| Gate overrides | {OVERRIDE_COUNT} |
+| Placed artifacts | {PLACED_COUNT} |
+
+## Findings by Category
+
+| Category | Extracted | Approved | Placed | Destination |
+|----------|-----------|----------|--------|-------------|
+| skills | {N} | {N} | {N} | .claude/skills/ |
+| commands | {N} | {N} | {N} | .claude/commands/ |
+| hooks | {N} | {N} | {N} | TOOLKIT_HARVEST.md |
+| rules | {N} | {N} | {N} | .claude/rules/ |
+| templates | {N} | {N} | {N} | TOOLKIT_HARVEST.md |
+| patterns | {N} | {N} | {N} | TOOLKIT_HARVEST.md |
+| snippets | {N} | {N} | {N} | TOOLKIT_HARVEST.md |
+
+## Quality Gate Results
+
+| Gate | Pass | Fail | Override | Pass Rate |
+|------|------|------|----------|-----------|
+| G1 When-To-Use | {N} | {N} | {N} | {%} |
+| G2 When-Not-To-Use | {N} | {N} | {N} | {%} |
+| G3 Decontextualized | {N} | {N} | {N} | {%} |
+| G4 Concrete Example | {N} | {N} | {N} | {%} |
+| G5 Reusability | {N} | {N} | {N} | {%} |
+| G6 Not Duplicate | {N} | {N} | {N} | {%} |
+| G7 Maturity Label | {N} | {N} | {N} | {%} |
+| G8 Brutal Honesty | {N} | {N} | {N} | {%} |
+
+## Coverage Gaps
+
+| Item | Reason |
+|------|--------|
+| {SKIPPED_FILES_TABLE_OR_NONE} | |
+
+**Failed extractors:** {FAILED_EXTRACTORS_OR_NONE}
+
+## Confidence Distribution
+
+| Range | Count | Percentage |
+|-------|-------|------------|
+| 0.9 - 1.0 (Excellent) | {N} | {%} |
+| 0.7 - 0.89 (Good) | {N} | {%} |
+| 0.5 - 0.69 (Acceptable) | {N} | {%} |
+| < 0.5 (Below threshold) | {N} | {%} |
+
+## Memory Integration
+
+- Patterns loaded from memory: {MEMORY_PATTERNS_COUNT}
+- Reward stored: {REWARD_SCORE} ({REWARD_LABEL})
+- Dream trigger updated: {YES_NO}
+
+---
+
+## Placeholder Reference
+
+| Placeholder | Computation |
+|-------------|-------------|
+| {SESSION_ID} | HarvestSession.session_id |
+| {DATE} | Current date (YYYY-MM-DD) |
+| {TARGET_PATHS} | Directory path(s) scanned |
+| {SCOPE_FILTER_OR_ALL} | Scope filter value or "all categories" |
+| {STEPS_COMPLETED} | Count of pipeline steps that ran successfully |
+| {TOTAL_EXTRACTED} | count(all findings at extraction) |
+| {AFTER_DEDUP} | count(findings after Module 02 dedup) |
+| {USER_REMOVALS} | count(findings removed by user at checkpoint) |
+| {USER_MERGES} | count(findings with non-null merged_from) |
+| {APPROVED_COUNT} | count(findings where status = approved) |
+| {BLOCKED_COUNT} | count(findings where status = blocked) |
+| {OVERRIDE_COUNT} | sum(length of all gate_overrides arrays) |
+| {PLACED_COUNT} | count(findings where status = integrated) |
+| {SKIPPED_FILES_TABLE_OR_NONE} | Table of files skipped (>100KB or binary), or "None" |
+| {FAILED_EXTRACTORS_OR_NONE} | List of failed agent lenses, or "None" |
+| {MEMORY_PATTERNS_COUNT} | count(patterns loaded from memory_query) |
+| {REWARD_SCORE} | User checkpoint response mapped to 0.0-1.0 |
+| {REWARD_LABEL} | excellent/good/needs_work/failed (see reward-learning.md) |

package/templates/CLAUDE.md

@@ -30,6 +30,7 @@
 
 /bto [путь или описание]           — Build-Test-Optimize полный цикл
 /bto-build [описание]              — Генерация нового скилла/команды
+/bto-benchmark [путь]              — Детерминистический бенчмаркинг артефакта
 /bto-test [путь]                   — Мульти-агентная оценка артефакта
 /bto-optimize [путь]               — Эволюционная оптимизация промптов
 
@@ -84,7 +85,8 @@ product-keysarium-2026/
 │       ├── presentation-storyteller/ ← Презентации со storytelling
 │       ├── problem-solver-enhanced/  ← Решение проблем (TRIZ + Game Theory)
 │       ├── reverse-engineering-unicorn/ ← Reverse engineering компаний
-│       └── feature-adr/            ← Adaptive feature development (9 steps, S/M/L/XL)
+│       ├── feature-adr/            ← Adaptive feature development (9 steps, S/M/L/XL)
+│       └── knowledge-extractor/   ← Knowledge harvesting (5 agents, 7 categories, 8 gates)
 ├── lib/                            ← Библиотечные модули
 │   ├── background-workers.md       ← Протокол фоновых воркеров
 │   ├── memory-protocol.md          ← Протокол memory_query / memory_store
@@ -103,6 +105,7 @@ product-keysarium-2026/
 ├── packages/                       ← npm пакеты
 │   ├── dz-keysarium/              ← @dzhechkov/keysarium (research pipeline)
 │   ├── dz-skills-bto/             ← @dzhechkov/skills-bto (BTO evaluation)
+│   ├── dz-feature-adr/            ← @dzhechkov/skills-feature-adr (feature development)
 │   └── dz-keysarium-core/         ← @dzhechkov/keysarium-core (shared framework)
 │       ├── governance/             ← Constitution + shards + checkpoints
 │       ├── memory/                 ← Learning protocol + reward + dream engine
@@ -188,30 +191,35 @@ DISCOVERY → EXPLORE  → RESEARCH → CJM PROTO    → SOLVE    → ARCHITECTU
 6 именованных loops формализуют передачу данных между фазами.
 См. `.claude/rules/feedback-loops.md` для Variable Registry.
 
-## Portable Brain
+## Portable Brain (v1.1)
 
 ```
-/brain-export [путь или "all"]  — Экспорт знаний в JSON контейнер
-/brain-import [файл.json]       — Импорт знаний из другого проекта
+/brain-export [путь или "all"]               — Экспорт знаний в JSON контейнер (v1.1 с manifest)
+/brain-export --delta <parent-brain.json>     — Дельта-экспорт (COW, JSON Patch RFC 6902)
+/brain-import [файл.json]                     — Импорт знаний из другого проекта
 ```
 
-Brain container включает: skill metadata, domain patterns, research summaries,
-harvest patterns, pipeline metrics. Selective import с merge-not-overwrite стратегией.
+Brain container v1.1 включает: SHA-256 manifest с checksum, skill metadata, domain patterns,
+research summaries, harvest patterns, pipeline metrics, reward data.
+Selective import с merge-not-overwrite стратегией. Backward compatible с v1.0.
+
+**v1.1 features:** Integrity manifest, delta exports (COW branching), 2-tier memory index,
+HOT/WARM/COLD/PURGE record lifecycle.
 
 ## keysarium-core — Shared Framework Package
 
 `@dzhechkov/keysarium-core` — домен-агностичный фреймворк для multi-agent пайплайнов. Содержит протоколы, которые используются и в Keysarium, и в BTO, и могут быть использованы любой командой для создания собственного пайплайна.
 
-### Архитектура 3 пакетов
+### Архитектура 4 пакетов
 
 ```
-@dzhechkov/keysarium-core          ← Общий фреймворк (NEW)
-        ^                ^
-        |                |
-   peerDep          peerDep
-        |                |
-@dzhechkov/keysarium    @dzhechkov/skills-bto
-  (исследования)         (оценка)
+            @dzhechkov/keysarium-core              ← Общий фреймворк
+                ^         ^         ^
+                |         |         |
+           peerDep    peerDep   peerDep
+                |         |         |
+@dzhechkov/keysarium  skills-bto  skills-feature-adr
+  (исследования)       (оценка)    (фичи)
 ```
 
 ### Модули core
@@ -219,7 +227,7 @@ harvest patterns, pipeline metrics. Selective import с merge-not-overwrite ст
 | Модуль | Файлы | Назначение |
 |--------|-------|-----------|
 | governance/ | 3 | Constitution, shard protocol, checkpoint/promise protocol |
-| memory/ | 3 | memory_query/store, reward tracking, dream engine |
+| memory/ (v1.1) | 3 | memory_query/store, reward tracking, dream engine, **2-tier index, COW branching, record lifecycle** |
 | orchestration/ | 4 | Queen protocol, 6 topologies, workers, model routing |
 | verification/ | 3 | Witness chain, judge attestation, audit trail |
 | trust-tiers/ | 2 | 4-tier classification, promotion protocol |
@@ -248,12 +256,14 @@ Core — это cleanroom-экстракция из `lib/` и `.claude/rules/`.
 
 После завершения каждого исследования:
 1. Запусти `/harvest researches/<slug>/`
-2. Агент ревьюит проект и находит универсальные паттерны
-3. Классифицирует: skills / commands / rules / templates / patterns / snippets
-4. Деконтекстуализирует и документирует
-5. Обновляет TOOLKIT_HARVEST.md
+2. 5 параллельных агентов-экстракторов сканируют проект через разные линзы
+3. Классифицирует: skills / commands / hooks / rules / templates / patterns / snippets (7 категорий)
+4. Пользователь ревьюит нумерованный список находок с гранулярным контролем (#N команды)
+5. 8 блокирующих quality gates (2-pass: deterministic + semantic)
+6. Автоматическое размещение артефактов + обновление TOOLKIT_HARVEST.md
 
 Для массового harvest: `/harvest all`
+Для фильтрации: `/harvest path/ only patterns,rules`
 
 ## Скиллы: как они работают
 
@@ -283,15 +293,24 @@ feature-adr (Adaptive Feature Development)
   ├── frontend-design (UI реализация, если есть)
   ├── modules/00-08 (9 шагов pipeline)
   └── references/ (ADR template, C4 template, QE checklist)
+
+knowledge-extractor (извлечение знаний)
+  ├── modules/01-extract (5 parallel extractor agents, sonnet)
+  ├── modules/02-classify (7-category classification + dedup)
+  ├── modules/03-gate (8 quality gates, 2-pass: deterministic + haiku)
+  ├── modules/04-integrate (auto-placement + harvest report)
+  ├── references/ (quality-gates, artifact-categories, maturity-model)
+  └── templates/ (artifact-card, harvest-report)
 ```
 
-## BTO — Build-Test-Optimize
+## BTO — Build-Benchmark-Test-Optimize
 
-Система мульти-агентной оценки и оптимизации скиллов/команд.
+Система мульти-агентной оценки и оптимизации скиллов/команд (4-module pipeline).
 
 ```
-/bto [путь]           — Полный цикл: BUILD → TEST → OPTIMIZE
+/bto [путь]           — Полный цикл: BUILD → BENCHMARK → TEST → OPTIMIZE
 /bto-build [описание] — Генерация нового скилла/команды
+/bto-benchmark [путь] — Детерминистический бенчмаркинг артефакта
 /bto-test [путь]      — Оценка: Layer 0 (детерм.) → Layer 1 (haiku) → Layer 2 (3 судьи)
 /bto-optimize [путь]  — Эволюционная оптимизация за 3 раунда
 ```
@@ -301,6 +320,7 @@ feature-adr (Adaptive Feature Development)
 | Layer | Агенты | Модель | Назначение |
 |-------|--------|--------|------------|
 | Layer 0 | 0 | — | Детерминистические пре-чеки (71 проверка) |
+| Benchmark | 3 (B2) | haiku | Golden samples + тесты + consistency + performance |
 | Layer 1 | 1 | haiku | Быстрая оценка по 5 измерениям |
 | Layer 2 | 3 | sonnet | Полная панель: Expert + Critic + Auditor |
 | Meta | 1 | opus | Разрешение разногласий (если delta > 3) |