owledge 0.6.0__py3-none-any.whl

owledge-0.6.0.data/data/AGENTS.template.md

@@ -0,0 +1,88 @@
+---
+title: "Agent Operating Notes Template"
+date: "YYYY-MM-DD"
+version: "0.1.0"
+status: "template"
+owner: "OWNER"
+tenant_id: "TENANT_ID"
+customer_id: "CUSTOMER_ID"
+project: "PROJECT_NAME"
+project_id: "PROJECT_ID"
+type: "template"
+session_id: "YYYY-MM-DD-short-task-title"
+agent_id: "orchestrator"
+parent_session: ""
+related_files:
+  - "PROJECT_CONTEXT.md"
+tags:
+  - agents
+  - runtime-instructions
+---
+
+# Agent Operating Notes
+
+This file is intentionally mirrored with the sibling runtime instruction file. Keep `AGENTS.md` and `CLAUDE.md` in sync.
+
+## Start Here
+
+1. Read `USER_CONTEXT.md` when present for private user preferences, goals, and agent collaboration defaults.
+2. Read `PROJECT_CONTEXT.md` before starting substantial project work.
+3. Check `global-memory/preferences/`, `global-memory/goals/`, and `global-memory/research/` when planning or personal context matters.
+4. Check `agent-memory/README.md` for the multi-agent documentation workflow.
+5. Check active local plans in `agent-plans/` if the task references planning, architecture, or handoff context.
+6. Check `agent-memory/ideas/` and `global-memory/ideas/` before drafting new plans or proposing new projects.
+7. Use `agent-memory/pi-agent/` reports when the task asks about trends, parallels, recurring errors, or central project candidates.
+8. Use `agent-memory/pi-agent/red-team/` scorecards when PI findings or agent outputs need quality challenge.
+9. Use `README.md` and `docs/` for user-facing setup details.
+
+## Memory Layers
+
+| Layer | Path | Tracked | Purpose |
+| --- | --- | --- | --- |
+| Global user context | `USER_CONTEXT.md` | No by default | Private user profile, preferences, goals, and agent collaboration defaults |
+| Global user memory | `global-memory/` | No by default | Private preferences, goals, daily notes, tasks, ideas, research, patterns, and coach reports |
+| Canonical project router | `PROJECT_CONTEXT.md` | Yes | Current vision, state, architecture, plans, decisions, and next actions |
+| Stable agent memory | `agent-memory/` | Yes | Session summaries, ADRs, templates, and reusable workflow documentation |
+| Idea memory | `agent-memory/ideas/` | Yes | Captured ideas and future project opportunities |
+| PI intelligence | `agent-memory/pi-agent/` | Yes | Candidate reports, parallels, trends, recurring errors, and central project concepts |
+| PI red team | `agent-memory/pi-agent/red-team/` | Yes | Scorecards and challenge reports for PI findings |
+| Local working plans | `agent-plans/` | No | Volatile task plans and scratch planning notes |
+| User-facing docs | `README.md`, `docs/` | Yes | Setup, operations, release, and deployment guides |
+
+## Multi-Agent Session Rules
+
+- One writer per file.
+- Subagents never edit canonical docs.
+- Orchestrators own deltas.
+- Reviewers own reviews.
+- Stable findings are promoted only after review.
+
+## Markdown Memory Rules
+
+- Treat Markdown frontmatter and typed edges as the durable source of truth.
+- Use stable `memory_id` values for links, graph edges, and RAG ids.
+- Read `USER_CONTEXT.md`, relevant global preferences/goals/research, `PROJECT_CONTEXT.md`, then `agent-memory/indexes/memory-index.jsonl`, then relevant compiled/canonical docs.
+- Workers write only sessions, evidence, handoffs, and task notes.
+- Ideas are captured in `agent-memory/ideas/` as drafts and checked before new plans.
+- Global ideas, preferences, goals, daily notes, tasks, research and coach reports live under `global-memory/` and remain private by default.
+- Daily notes, personal tasks, and onboarding profiles are deep-dive sources only and are not default RAG input.
+- Research cards must include source URL, source date, retrieval date, version context, valid-until date, and confidence.
+- Coach recommendations must cite source memory IDs and include a next action.
+- PI Agent findings are candidate intelligence in `agent-memory/pi-agent/` and require review before promotion.
+- PI Red Team evaluations score outputs from 1-100 and classify them as `block`, `revise`, `accept`, or `promote-candidate`.
+- Orchestrators and memory curators own promotion into canonical, compiled, patterns, and lessons.
+- Attach evidence and gate reports before requesting promotion.
+- Do not mark work `done` unless the required QA gates pass.
+- Build a task-specific context pack instead of pasting long transcripts.
+- Do not export raw sessions to shared RAG by default.
+- Preserve contradictions through `contradicts` edges instead of overwriting history.
+- Do not read, claim, export, or promote artifacts from another `tenant_id` unless `shared_scope` explicitly allows it.
+
+## Optional Runtime Adapters
+
+SQLite, HTTP, MCP, or runtime plugins may coordinate work, but they must not become the canonical memory store. Generated indexes and RAG exports can always be rebuilt from Markdown.
+
+The optional `plugins/agent-memory-cowork/` adapter can capture Claude/Cowork runtime events into private sessions and draft summaries. Treat those drafts as unreviewed until a memory curator promotes them.
+
+The optional `plugins/pi-agent-workspace/` adapter can run workspace checks and PI intelligence reports. Treat generated PI reports as review inputs, not accepted memory.
+The optional Red Team PI Agent challenges PI reports and agent outputs before promotion decisions.

owledge-0.6.0.data/data/CLAUDE.template.md

@@ -0,0 +1,88 @@
+---
+title: "Agent Operating Notes Template"
+date: "YYYY-MM-DD"
+version: "0.1.0"
+status: "template"
+owner: "OWNER"
+tenant_id: "TENANT_ID"
+customer_id: "CUSTOMER_ID"
+project: "PROJECT_NAME"
+project_id: "PROJECT_ID"
+type: "template"
+session_id: "YYYY-MM-DD-short-task-title"
+agent_id: "orchestrator"
+parent_session: ""
+related_files:
+  - "PROJECT_CONTEXT.md"
+tags:
+  - agents
+  - runtime-instructions
+---
+
+# Agent Operating Notes
+
+This file is intentionally mirrored with the sibling runtime instruction file. Keep `AGENTS.md` and `CLAUDE.md` in sync.
+
+## Start Here
+
+1. Read `USER_CONTEXT.md` when present for private user preferences, goals, and agent collaboration defaults.
+2. Read `PROJECT_CONTEXT.md` before starting substantial project work.
+3. Check `global-memory/preferences/`, `global-memory/goals/`, and `global-memory/research/` when planning or personal context matters.
+4. Check `agent-memory/README.md` for the multi-agent documentation workflow.
+5. Check active local plans in `agent-plans/` if the task references planning, architecture, or handoff context.
+6. Check `agent-memory/ideas/` and `global-memory/ideas/` before drafting new plans or proposing new projects.
+7. Use `agent-memory/pi-agent/` reports when the task asks about trends, parallels, recurring errors, or central project candidates.
+8. Use `agent-memory/pi-agent/red-team/` scorecards when PI findings or agent outputs need quality challenge.
+9. Use `README.md` and `docs/` for user-facing setup details.
+
+## Memory Layers
+
+| Layer | Path | Tracked | Purpose |
+| --- | --- | --- | --- |
+| Global user context | `USER_CONTEXT.md` | No by default | Private user profile, preferences, goals, and agent collaboration defaults |
+| Global user memory | `global-memory/` | No by default | Private preferences, goals, daily notes, tasks, ideas, research, patterns, and coach reports |
+| Canonical project router | `PROJECT_CONTEXT.md` | Yes | Current vision, state, architecture, plans, decisions, and next actions |
+| Stable agent memory | `agent-memory/` | Yes | Session summaries, ADRs, templates, and reusable workflow documentation |
+| Idea memory | `agent-memory/ideas/` | Yes | Captured ideas and future project opportunities |
+| PI intelligence | `agent-memory/pi-agent/` | Yes | Candidate reports, parallels, trends, recurring errors, and central project concepts |
+| PI red team | `agent-memory/pi-agent/red-team/` | Yes | Scorecards and challenge reports for PI findings |
+| Local working plans | `agent-plans/` | No | Volatile task plans and scratch planning notes |
+| User-facing docs | `README.md`, `docs/` | Yes | Setup, operations, release, and deployment guides |
+
+## Multi-Agent Session Rules
+
+- One writer per file.
+- Subagents never edit canonical docs.
+- Orchestrators own deltas.
+- Reviewers own reviews.
+- Stable findings are promoted only after review.
+
+## Markdown Memory Rules
+
+- Treat Markdown frontmatter and typed edges as the durable source of truth.
+- Use stable `memory_id` values for links, graph edges, and RAG ids.
+- Read `USER_CONTEXT.md`, relevant global preferences/goals/research, `PROJECT_CONTEXT.md`, then `agent-memory/indexes/memory-index.jsonl`, then relevant compiled/canonical docs.
+- Workers write only sessions, evidence, handoffs, and task notes.
+- Ideas are captured in `agent-memory/ideas/` as drafts and checked before new plans.
+- Global ideas, preferences, goals, daily notes, tasks, research and coach reports live under `global-memory/` and remain private by default.
+- Daily notes, personal tasks, and onboarding profiles are deep-dive sources only and are not default RAG input.
+- Research cards must include source URL, source date, retrieval date, version context, valid-until date, and confidence.
+- Coach recommendations must cite source memory IDs and include a next action.
+- PI Agent findings are candidate intelligence in `agent-memory/pi-agent/` and require review before promotion.
+- PI Red Team evaluations score outputs from 1-100 and classify them as `block`, `revise`, `accept`, or `promote-candidate`.
+- Orchestrators and memory curators own promotion into canonical, compiled, patterns, and lessons.
+- Attach evidence and gate reports before requesting promotion.
+- Do not mark work `done` unless the required QA gates pass.
+- Build a task-specific context pack instead of pasting long transcripts.
+- Do not export raw sessions to shared RAG by default.
+- Preserve contradictions through `contradicts` edges instead of overwriting history.
+- Do not read, claim, export, or promote artifacts from another `tenant_id` unless `shared_scope` explicitly allows it.
+
+## Optional Runtime Adapters
+
+SQLite, HTTP, MCP, or runtime plugins may coordinate work, but they must not become the canonical memory store. Generated indexes and RAG exports can always be rebuilt from Markdown.
+
+The optional `plugins/agent-memory-cowork/` adapter can capture Claude/Cowork runtime events into private sessions and draft summaries. Treat those drafts as unreviewed until a memory curator promotes them.
+
+The optional `plugins/pi-agent-workspace/` adapter can run workspace checks and PI intelligence reports. Treat generated PI reports as review inputs, not accepted memory.
+The optional Red Team PI Agent challenges PI reports and agent outputs before promotion decisions.

owledge-0.6.0.data/data/DESIGN.md

@@ -0,0 +1,114 @@
+---
+selected_report_design: "atlas-command"
+design_catalog_version: "0.1.0"
+status: "draft"
+updated_at: "2026-05-12"
+source_of_truth: "DESIGN.md"
+applies_to:
+  - decision
+  - handoff
+  - rag-readiness
+  - agent-activity
+  - project-dashboard
+  - website-ui
+---
+
+# Report Design System
+
+`DESIGN.md` is the central design decision file for all generated Agent Memory HTML reports. Markdown remains the source of truth. HTML files are generated views and must read this file before choosing colors, spacing, density, and report personality.
+
+## Selected Design
+
+| Field | Value |
+| --- | --- |
+| Selected design id | `atlas-command` |
+| Selected design name | Atlas Command |
+| Intended use | Dense enterprise dashboards, multi-agent status reports, RAG readiness views |
+| Default density | Compact |
+| Default mode | Light |
+
+## Persistence Rule
+
+When a user chooses a report style in `REPORT_DESIGN_SELECTOR.html`, persist the chosen style by updating `selected_report_design` in this file and keeping the selected row below unchanged. Generated reports should include the selected design id in their metadata.
+
+## Branding Model
+
+Current v0.2 behavior is intentionally simple: one global `selected_report_design` controls all generated reports in a project.
+
+Future multi-brand behavior should add named profiles without breaking the current field:
+
+```yaml
+selected_report_design: "atlas-command"
+brand_profiles:
+  default: "atlas-command"
+  internal: "signal-grid"
+  customer_acme: "nordic-clarity"
+  executive: "executive-ledger"
+```
+
+Report generation can then accept a future `brand_profile` option while still falling back to `selected_report_design`.
+
+## Design Catalog
+
+| ID | Name | Best For | Visual Character | Primary | Background | Risk |
+| --- | --- | --- | --- | --- | --- | --- |
+| `atlas-command` | Atlas Command | Enterprise operations, RAG readiness, multi-agent dashboards | Command-center clarity with restrained blue, graphite, and crisp tables | `#245c88` | `#f6f7f9` | Can feel too operational for client-facing brand reports |
+| `glass-ledger` | Glass Ledger | Governance, audit trails, compliance-adjacent decisions | Translucent panels, precise borders, cool neutrals | `#317873` | `#eef5f3` | Needs careful contrast for dense tables |
+| `signal-grid` | Signal Grid | Metrics, retrieval evals, performance reports | High-signal grids, amber accents, terminal-inspired density | `#b36b00` | `#111418` | Dark mode may be too intense for executive exports |
+| `courtroom-brief` | Courtroom Brief | ADRs, decisions, risk tradeoffs, evidence chains | Formal legal brief styling with serif headings and warm paper | `#7a3b1f` | `#fbf7ef` | Less suitable for technical dashboards |
+| `mission-control` | Mission Control | Agent orchestration, task flow, delegation reports | Dark control-room panels, cyan status accents, strong grouping | `#35b7d8` | `#071117` | Should not be used for printable reports |
+| `blueprint-studio` | Blueprint Studio | Architecture, UI systems, implementation plans | Blueprint grid, technical annotations, crisp cyan lines | `#1178a8` | `#edf7fb` | Can overemphasize technical structure |
+| `executive-ledger` | Executive Ledger | Client status, management updates, board-style summaries | Quiet premium editorial system with navy, ivory, and gold accents | `#9b7a2f` | `#faf8f1` | Lower visual energy for engineering teams |
+| `graph-aurora` | Graph Aurora | GraphRAG, cross-project parallels, semantic maps | Layered graph feel, saturated accent pairs, soft dark canvas | `#78d6a3` | `#10151f` | Needs restraint to avoid decorative noise |
+| `monolith-minimal` | Monolith Minimal | Dense technical review, source-heavy reports | Brutal clarity, monochrome, hard lines, compact cards | `#111111` | `#f3f3f0` | Too severe for customer storytelling |
+| `workshop-canvas` | Workshop Canvas | PM to dev handoffs, planning sessions, collaborative reviews | Whiteboard structure, sticky-note accents, approachable hierarchy | `#d66a28` | `#fffaf2` | Can feel informal for compliance reports |
+| `neon-console` | Neon Console | Runtime logs, agent activity, dev demos | Black console canvas, green/cyan highlights, monospace data blocks | `#39ff88` | `#050806` | Not ideal for long-form reading |
+| `nordic-clarity` | Nordic Clarity | Customer-safe reports, documentation, handoffs | Calm light surfaces, cool grays, precise blue-green accents | `#3f8f95` | `#f7faf9` | May feel too quiet for brand selection |
+| `evidence-vault` | Evidence Vault | Source provenance, QA gates, audit reports | Secure archive look, dark slate, red/orange status semantics | `#d24b35` | `#111820` | Strong security tone can make normal reports feel high-risk |
+| `product-lab` | Product Lab | Website/UI decisions, design tokens, experiments | Modular design-board layout with vivid swatches and sliders | `#e24d7b` | `#fff6f8` | Needs token discipline to stay polished |
+| `zeus-celestial` | Zeus Celestial | Vision decks, strategic narratives, Zeus-style AI agency reports | High-contrast cosmic editorial with electric blue and silver | `#65a9ff` | `#080b13` | Use sparingly for polished storytelling, not raw QA |
+
+## Token Contract
+
+Each design must resolve to these tokens:
+
+| Token | Meaning |
+| --- | --- |
+| `report.background` | Page background |
+| `report.panel` | Card and section surface |
+| `report.foreground` | Primary text |
+| `report.ink` | Legacy alias for primary text |
+| `report.muted` | Secondary text |
+| `report.line` | Borders and dividers |
+| `report.accent` | Primary action and emphasis color |
+| `report.accent_2` | Secondary accent for charts, graph edges, or visual contrast |
+| `report.radius` | Card and control radius |
+| `report.density` | Density in pixels or compact/balanced/spacious alias |
+| `report.border_width` | Standard report border width |
+| `report.border_style` | `solid`, `dashed`, `dotted`, or `double` |
+| `report.font_mode` | System, editorial, mono, or hybrid |
+| `button.primary_bg` | Primary button background |
+| `button.primary_fg` | Primary button text |
+| `button.secondary_bg` | Secondary button background |
+| `button.secondary_fg` | Secondary button text |
+| `button.radius` | Button radius |
+| `button.weight` | Button font weight |
+| `button.case` | Button text transform |
+| `ornament.mode` | Optional visual motif: none, grid, beam, ledger, radial |
+| `shadow.color` | Shadow color source |
+| `shadow.strength` | Shadow depth/intensity |
+
+## Merge Note For AI-WorkOS Obsidian Vault
+
+The uploaded AI-WorkOS Obsidian Vault should not replace Agent Memory Kit. It is best treated as an upstream concept library and migration source.
+
+| Vault Concept | Merge Target | Decision |
+| --- | --- | --- |
+| `00_SYSTEM/MERGE_MAP.md` scoring | `find-parallels` and future merge assessment docs | Merge as a cross-project evaluation rubric |
+| `00_SYSTEM/FRONTMATTER_SCHEMA.md` flat Obsidian fields | `agent-memory/schemas/frontmatter.schema.json` | Map concepts, keep stricter Agent Memory fields |
+| `00_SYSTEM/GRAPH_SCHEMA.md` node/edge vocabulary | `agent-memory/schemas/graphrag-*.schema.json` | Merge vocabulary as aliases, keep typed frontmatter edges |
+| `00_SYSTEM/INGEST_PIPELINE.md` raw to reviewed flow | runtime capture and session compaction docs | Merge as lifecycle documentation |
+| `00_SYSTEM/RAG_INDEXING_POLICY.md` tier model | RAG export policy | Merge as index priority tiers |
+| Obsidian folder taxonomy | Optional enterprise hub adapter | Keep separate, do not force into project-local core |
+
+Recommendation: keep the vault as a second-system adapter until a migration tool can map its `type/id/canonical/kb_ready/confidentiality` fields into Agent Memory `memory_id/doc_type/status/review_status/sanitization_status/data_class/edges`.

owledge-0.6.0.data/data/PROJECT_CONTEXT.template.md

@@ -0,0 +1,169 @@
+---
+title: "PROJECT_NAME Project Context"
+date: "YYYY-MM-DD"
+version: "0.1.0"
+status: "active"
+owner: "OWNER"
+memory_id: "mem:TENANT_ID:CUSTOMER_ID:PROJECT_ID:project_context:project-router"
+tenant_id: "TENANT_ID"
+customer_id: "CUSTOMER_ID"
+project: "PROJECT_NAME"
+project_id: "PROJECT_ID"
+doc_type: "project_context"
+visibility: "private"
+data_class: "internal"
+semantic_title: "PROJECT_NAME project context router"
+summary: "Project router for goals, architecture, memory locations, QA gates, and agent operating model."
+concept_tags:
+  - "project-context"
+  - "agent-memory"
+stack_tags: []
+problem_patterns: []
+architecture_patterns:
+  - "markdown-first-memory"
+failure_modes: []
+reusable_lessons: []
+confidence: 1.0
+review_status: "reviewed"
+sanitization_status: "not_required"
+created_at: "YYYY-MM-DDT00:00:00Z"
+updated_at: "YYYY-MM-DDT00:00:00Z"
+source_hash: ""
+edges: []
+type: "project-context"
+session_id: "YYYY-MM-DD-short-task-title"
+agent_id: "orchestrator"
+parent_session: ""
+related_files:
+  - "AGENTS.md"
+  - "CLAUDE.md"
+  - "README.md"
+tags:
+  - project-context
+---
+
+# PROJECT_NAME Project Context
+
+## Project Vision
+
+Describe the product, project, or system vision.
+
+## Current State
+
+Describe what works today and what is verified.
+
+## Current Tech Stack
+
+| Component | Purpose | Why It Exists | Status |
+| --- | --- | --- | --- |
+
+## Painpoints To Solve
+
+- Painpoint 1
+
+## Solved Painpoints
+
+- Solved item 1
+
+## Target Architecture
+
+```text
+User -> System -> Result
+```
+
+## Stack Mapping
+
+| Stack | Use When | Key Files |
+| --- | --- | --- |
+
+## Setup Routes
+
+- Route 1
+
+## Agent Operating Model
+
+Describe how agents and subagents should document work.
+
+## Tenant / Access Model
+
+| Field | Value |
+| --- | --- |
+| Tenant ID | TENANT_ID |
+| Customer ID | CUSTOMER_ID |
+| Project ID | PROJECT_ID |
+| Default Data Class | internal |
+| Shared Scope | none |
+
+## Markdown Memory Core
+
+| Setting | Value |
+| --- | --- |
+| Source Of Truth | Markdown frontmatter and typed edges |
+| Project Router | `PROJECT_CONTEXT.md` |
+| Canonical Memory | `agent-memory/canonical/` |
+| Compiled Memory | `agent-memory/compiled/` |
+| Pattern Memory | `agent-memory/patterns/` |
+| Shared Lessons | `agent-memory/lessons/` |
+| Idea Memory | `agent-memory/ideas/` |
+| PI Intelligence | `agent-memory/pi-agent/` |
+| PI Red Team | `agent-memory/pi-agent/red-team/` |
+| Machine Index | `agent-memory/indexes/memory-index.jsonl` |
+| RAG Exports | `agent-memory/exports/` |
+
+## QA Gates
+
+| Gate | Required | Evidence |
+| --- | --- | --- |
+| Tests | Yes | Gate report |
+| Build / Type Safety | Yes | Gate report |
+| Security | Yes | Gate report |
+| Performance | Project-specific | Gate report |
+| Accessibility | UI projects | Gate report |
+
+## Active Plans
+
+| Plan | Status | Location |
+| --- | --- | --- |
+
+## Idea Backlog
+
+Agents should check `agent-memory/ideas/` before drafting new plans. Related ideas should be linked through `similar_to`, `relates_to`, `derived_from`, or `shared_lesson_for` edges.
+
+| Idea | Status | Relevance | Location |
+| --- | --- | --- | --- |
+
+## PI Agent Intelligence
+
+PI Agent reports are candidate intelligence for parallels, trends, repeated agent errors, and central project opportunities. They help planning and memory curation, but require review before promotion.
+
+| Artifact | Status | Signal | Location |
+| --- | --- | --- | --- |
+
+## Evaluation Frameworks
+
+Use 1-100 scorecards for important agent outputs, PI reports, plans, QA findings, and promotion candidates.
+
+| Framework | Target | Accept Score | Location |
+| --- | --- | ---: | --- |
+| PI Red Team Evaluation | PI intelligence reports | 85 | `agent-memory/pi-agent/red-team/` |
+
+## Session Log
+
+| Date | Session | Outcome |
+| --- | --- | --- |
+
+## Decision Log
+
+- Decision 1
+
+## Validation / QA Status
+
+- QA state
+
+## Next Best Actions
+
+1. Next action
+
+## Open Questions
+
+- Question 1