nlos 1.0.0

package/.cursor/commands/kernel-boot.md

@@ -0,0 +1,254 @@
+---
+type: slash-command
+command: /kernel-boot
+aliases: ["/startup", "./kernel-boot", "./startup", "/claude-boot", "./claude-boot"]
+purpose: Load kernel context with tiered loading - mandatory files by default, full context with --full
+description: Session initialization command with tiered loading to minimize boot context while preserving full capability. Model-agnostic - works with any LLM runtime.
+scope: Session initialization, context loading
+version: 3.0.0
+last_updated: 2026-01-11
+
+contract:
+  mode: factual
+  precision_scale: strict
+  escalation_triggers: []  # Read-only context loading
+  resource_expectations:
+    token_usage: low
+---
+
+# Kernel Boot Command
+
+Initialize Capturebox Natural Language Operating System with tiered kernel loading. This command is model-agnostic and works with any LLM runtime (Claude Code, Cursor, Ollama, llama.cpp, LM Studio, etc.).
+
+## Design Rationale
+
+Every token at boot costs context for actual work. The tiered design loads only essential files by default (~10.6K tokens, 5.3% of context), deferring capabilities until needed.
+
+| Tier | Files | Tokens | Loaded |
+|------|-------|--------|--------|
+| Mandatory | memory.md, AGENTS.md, axioms.yaml | ~10,600 | Always |
+| Lazy | personalities.md | ~3,600 | On `/assume` |
+| Lazy | COMMAND-MAP.md | ~1,350 | On `/sys-ref` or command lookup |
+
+## Behavior
+
+### Default: `./kernel-boot`
+
+Load mandatory tier only:
+
+1. **Read Mandatory Kernel Files** (in priority order):
+   - Read `memory.md` (behavioral directives)
+   - Read `AGENTS.md` (agent kernel, hard invariants)
+   - Read `axioms.yaml` (canonical definitions)
+
+2. **Generate Boot Summary**:
+   - Confirm mandatory files loaded
+   - Show lazy tier as available but not loaded
+   - Display active constraints
+   - Report token usage
+
+### Full: `./kernel-boot --full`
+
+Load all tiers:
+
+1. **Read All Kernel Files**:
+   - Read `memory.md` (behavioral directives)
+   - Read `AGENTS.md` (agent kernel, hard invariants)
+   - Read `axioms.yaml` (canonical definitions)
+   - Read `personalities.md` (voice presets)
+   - Read `.cursor/commands/COMMAND-MAP.md` (command registry)
+
+2. **Generate Full Boot Summary**
+
+### Output Format (Default)
+
+```
+NL-OS KERNEL BOOT SEQUENCE
+
+Kernel Status (Mandatory):
+  [x] memory.md - behavioral directives loaded
+  [x] AGENTS.md - hard invariants loaded
+  [x] axioms.yaml - canonical definitions loaded
+
+Lazy-Load Available:
+  [ ] personalities.md - load via /assume
+  [ ] COMMAND-MAP.md - load via /sys-ref
+
+Active Constraints:
+  - No emojis in output
+  - Preserve frontmatter and metadata
+  - Append-only logs (hype.log)
+  - Prefer patch edits over rewrites
+
+Boot: ~10,600 tokens | Context remaining: ~189K
+
+KERNEL BOOT COMPLETE
+```
+
+### Output Format (--full)
+
+```
+NL-OS KERNEL BOOT SEQUENCE (FULL)
+
+Kernel Status:
+  [x] memory.md - behavioral directives loaded
+  [x] AGENTS.md - hard invariants loaded
+  [x] axioms.yaml - canonical definitions loaded
+  [x] personalities.md - voice presets loaded
+  [x] COMMAND-MAP.md - command registry loaded
+
+Active Constraints:
+  - No emojis in output
+  - Preserve frontmatter and metadata
+  - Append-only logs (hype.log)
+  - Prefer patch edits over rewrites
+
+Systems: 17 | Commands: 58 | Skills: 5
+
+Boot: ~15,500 tokens | Context remaining: ~184K
+
+KERNEL BOOT COMPLETE (FULL)
+```
+
+### Variants
+
+- `./kernel-boot` - Default: load mandatory tier only (~10.6K tokens)
+- `./kernel-boot --full` - Load all tiers including lazy files (~15.5K tokens)
+- `./kernel-boot --quiet` - Load mandatory tier silently, brief acknowledgment only
+- `./kernel-boot --verify` - Load and validate all referenced paths exist
+- `./kernel-boot --status` - Show boot status without reloading
+
+### Error Handling
+
+If any mandatory file is missing, report and handle appropriately:
+
+| File | Tier | If Missing |
+|------|------|------------|
+| `memory.md` | Mandatory | CRITICAL - refuse to proceed |
+| `AGENTS.md` | Mandatory | WARNING - proceed with caution |
+| `axioms.yaml` | Mandatory | WARNING - proceed but note definitions unavailable |
+| `personalities.md` | Lazy | N/A - not loaded by default |
+| `COMMAND-MAP.md` | Lazy | N/A - not loaded by default |
+
+## Contract
+
+**Requires:**
+- Input: None (or optional flags: --full, --quiet, --verify, --status)
+- Context: Must be in capturebox repository (or have kernel files available)
+- Files: Mandatory tier files should exist (memory.md is required)
+
+**Guarantees:**
+- Output: Boot status summary in chat (unless --quiet)
+- Side effects: No file writes, context loading only
+- Default behavior: Reads ~10.6K tokens (mandatory tier)
+- Full behavior: Reads ~15.5K tokens (all tiers)
+
+**Refuses:**
+- Running without kernel files available
+- Writing to any files
+- Proceeding without memory.md (mandatory file)
+
+## Examples
+
+### Example 1: Default Boot (Mandatory Only)
+
+**User:** `./kernel-boot`
+
+**Assistant:**
+```
+NL-OS KERNEL BOOT SEQUENCE
+
+Kernel Status (Mandatory):
+  [x] memory.md - behavioral directives loaded
+  [x] AGENTS.md - hard invariants loaded
+  [x] axioms.yaml - canonical definitions loaded
+
+Lazy-Load Available:
+  [ ] personalities.md - load via /assume
+  [ ] COMMAND-MAP.md - load via /sys-ref
+
+Active Constraints:
+  - No emojis in output
+  - Preserve frontmatter and metadata
+  - Append-only logs (hype.log)
+  - Prefer patch edits over rewrites
+
+Boot: ~10,600 tokens | Context remaining: ~189K
+
+KERNEL BOOT COMPLETE
+```
+
+### Example 2: Full Boot
+
+**User:** `./kernel-boot --full`
+
+**Assistant:**
+```
+NL-OS KERNEL BOOT SEQUENCE (FULL)
+
+Kernel Status:
+  [x] memory.md - behavioral directives loaded
+  [x] AGENTS.md - hard invariants loaded
+  [x] axioms.yaml - canonical definitions loaded
+  [x] personalities.md - voice presets loaded
+  [x] COMMAND-MAP.md - command registry loaded
+
+Active Constraints:
+  - No emojis in output
+  - Preserve frontmatter and metadata
+  - Append-only logs (hype.log)
+  - Prefer patch edits over rewrites
+
+Systems: 17 | Commands: 58 | Skills: 5
+
+Boot: ~15,500 tokens | Context remaining: ~184K
+
+KERNEL BOOT COMPLETE (FULL)
+```
+
+### Example 3: Quiet Boot
+
+**User:** `./kernel-boot --quiet`
+
+**Assistant:**
+> Kernel loaded (mandatory tier). Ready for capturebox operations.
+
+### Example 4: Status Check
+
+**User:** `./kernel-boot --status`
+
+**Assistant:**
+> Boot status: Mandatory tier loaded. Lazy tier available on demand. Run `./kernel-boot --full` for complete context.
+
+## Implementation Notes
+
+- **Tiered loading**: Default loads only mandatory files to preserve context
+- **Lazy triggers**: personalities.md loads when `/assume` is called; COMMAND-MAP.md loads on `/sys-ref` or command lookup
+- **Idempotent**: Safe to run multiple times (reloads fresh context)
+- **Token efficiency**: Default saves ~4,900 tokens vs full boot
+- **Model-agnostic**: Works with any LLM that can read files and follow instructions
+
+## Backwards Compatibility
+
+- `/claude-boot` and `./claude-boot` remain as aliases
+- All existing workflows continue to work unchanged
+
+## Portability Notes
+
+This command is workspace-specific. To port to another NL-OS workspace:
+
+1. Copy this file to `[workspace]/.cursor/commands/kernel-boot.md`
+2. Update kernel file paths to match target workspace structure
+3. Define which files are mandatory vs lazy for that workspace
+4. Update token estimates based on actual file sizes
+5. Ensure mandatory files exist at referenced paths
+
+For local LLM runtimes, use the boot scripts in `scripts/`:
+- `scripts/kernel-boot-ollama.sh`
+- `scripts/kernel-boot-llama-cpp.sh`
+- `scripts/kernel-boot-lm-studio.sh`
+
+Or use the portable payload: `portable/kernel-payload.md`
+
+Dependencies (Mandatory): memory.md, AGENTS.md, axioms.yaml
+Dependencies (Lazy): personalities.md, COMMAND-MAP.md

package/.cursor/commands/note.md

@@ -0,0 +1,28 @@
+---
+type: slash-command
+command: /note
+purpose: Ultra-fast note capture via shell script
+version: 5.1.0
+
+contract:
+  mode: factual
+  precision_scale: strict
+  escalation_triggers:
+    - destructive_write
+  resource_expectations:
+    token_usage: low
+---
+
+# /note
+
+## Behavior
+
+1. Run: `/Users/caantone/Documents/cisco/capturebox/scripts/note.sh`
+   - **With content**: Append timestamped entry with content to today's daily note
+   - **No content**: Append timestamp marker only (creates checkpoint)
+   - Creates `docs/notes/dailies/YYYY-MM-DD.md` if it doesn't exist
+
+2. **IMPORTANT**:  Open today's daily note in the editor: `docs/notes/dailies/YYYY-MM-DD.md`
+
+No confirmation. No output.
+

package/.cursor/commands/scratchpad.md

@@ -0,0 +1,81 @@
+---
+type: slash-command
+command: /scratchpad
+purpose: Capture full conversation threads (user + assistant) for session archival
+version: 3.0.0
+last_updated: 2025-12-16
+
+contract:
+  mode: factual
+  precision_scale: strict
+  escalation_triggers:
+    - destructive_write
+  resource_expectations:
+    token_usage: low
+---
+
+# Scratchpad
+
+Capture full conversation threads from this chat session. Unlike `/note` (quick thoughts), scratchpad logs entire exchanges with both user and assistant messages.
+
+## Usage
+
+```
+/scratchpad [name] [--summary] [--daily]
+```
+
+**Parameters:**
+- `[name]` — Optional name for the scratchpad (defaults to `scratchpad_YYYY-MM-DD.md`)
+- `--summary` — Create condensed single-paragraph summary instead of full conversation
+- `--daily` — Append to `docs/notes/dailies/YYYY-MM-DD.md` instead of scratchpads folder
+
+**Examples:**
+```
+/scratchpad
+# → Full conversation to scratchpads/scratchpad_2025-12-16.md
+
+/scratchpad design-system-work
+# → Full conversation to scratchpads/design-system-work.md
+
+/scratchpad --summary --daily
+# → Single-paragraph summary appended to dailies/2025-12-16.md
+```
+
+## Storage
+
+- **Default**: `docs/notes/scratchpads/[filename].md`
+- **With --daily**: `docs/notes/dailies/YYYY-MM-DD.md`
+- Auto-create directories if missing
+
+## File Format
+
+**New file:**
+```markdown
+---
+title: "Scratchpad: [name or date]"
+date: YYYY-MM-DD
+type: scratchpad
+tags: [session, conversation]
+---
+
+# Scratchpad: [name or date]
+
+## [HH:MM] - User
+[User's message]
+
+## [HH:MM] - Assistant
+[Assistant's response]
+
+---
+```
+
+**Existing file:** Append conversation with timestamps and separator.
+
+## Protocol
+
+Capture the substantive user/assistant exchanges from this chat session. Include discussions, decisions, work accomplished, and key insights. Format each message with timestamp and role.
+
+On completion, append witness hit to `projects/systems/self-writer-system/data/witness_activity.jsonl`:
+```json
+{"v":1,"ts":"[ISO timestamp]","cmd":"scratchpad","area":"notes","out":"[target-file]"}
+```

package/.cursor/commands/sys-ref.md

@@ -0,0 +1,81 @@
+---
+type: slash-command
+command: /sys-ref
+purpose: Display quick-scan reference for all commands and systems
+description: Quick reference guide showing commands and systems with descriptions and use-case triggers
+scope: Navigation and discovery
+version: 1.0.0
+last_updated: 2025-01-08
+
+contract:
+  mode: factual
+  precision_scale: strict
+  escalation_triggers: []  # Read-only
+  resource_expectations:
+    token_usage: low
+---
+
+# System Reference (sys-ref)
+
+Quick-scan reference for all slash commands and systems in Capturebox.
+
+---
+
+## Behavior
+
+Display the contents of `.cursor/SYS-REF.md` in the chat.
+
+**Usage:**
+- `/sys-ref` — Show complete reference (no arguments)
+
+**Purpose:**
+- Solve the "forgetting" problem: remind you what commands and systems exist
+- Quick scanning without opening multiple files
+- Use-case triggers to help identify the right tool for your current task
+
+**Output:**
+- Display `.cursor/SYS-REF.md` contents in chat (read-only)
+- User can then invoke any command or system they rediscover
+
+---
+
+## Implementation
+
+```
+1. Read `.cursor/SYS-REF.md`
+2. Display contents in chat
+3. Done
+```
+
+No arguments. No file creation. Pure reference display.
+
+---
+
+## Maintenance
+
+The reference doc (`.cursor/SYS-REF.md`) is manually maintained. When commands or systems change:
+
+1. Update command/system documentation first (as usual)
+2. Update `.cursor/SYS-REF.md` to reflect changes
+3. Optional: Run `/command-update-list sys-ref` + `/command-update-files` workflow if automation is desired
+
+---
+
+## Related Commands
+
+- `/systems` — Detailed system summaries via Python CLI
+- `/architecture` — Query architecture map for integration guidance
+- `/skills` — List and route skills from `.cursor/skills/`
+- `/whats-next` — Surface 3 interesting next actions
+
+---
+
+## Portability Notes
+
+This command is self-contained and portable to any workspace:
+
+1. Copy this file to `[workspace]/.cursor/commands/sys-ref.md`
+2. Create `[workspace]/.cursor/SYS-REF.md` with your command/system listing
+3. Run with `/sys-ref`
+
+**Dependencies:** None (pure file read and display)

package/AGENTS.md

@@ -0,0 +1,67 @@
+---
+title: Capturebox Agent Kernel
+type: agent-directive
+status: experimental
+last_updated: 2025-12-12
+purpose: Bootloader + hard invariants for any LLM/agent working in this repo.
+---
+
+# Capturebox Agent Kernel (AGENTS.md)
+
+This repository is a **natural language operating system** with reusable systems, slash-command workflows, and metadata-tagged knowledge files. Treat it as an instruction substrate, not a conventional app repo.
+
+If anything in this file conflicts with higher-priority instructions from the user, follow the user.
+
+## Boot Order (Read First)
+
+When entering a new task in capturebox:
+
+1. Read `memory.md` and obey it as canonical directive stack.
+2. Read `axioms.yaml` for canonical assertions, definitions, and invariants.
+3. Read `.cursor/domain-memory.yaml` for runtime state (active goals, loaded domains, current focus).
+4. Orient to the systems architecture in `projects/systems/` (overview in `projects/README.md`).
+5. Read the syscall table in `.cursor/commands/COMMAND-MAP.md` for available slash commands and their semantics.
+6. For domain knowledge, start with `knowledge/README.md` and then `knowledge/_index.yaml` before loading large knowledge files.
+   - Use `.cursor/skills/knowledge-retrieval/SKILL.md` for efficient, structured retrieval.
+
+If a task references a specific system or command, open that system/command spec before acting.
+
+## Hard Invariants
+
+- **No emojis** (emoji-range pictographs) in any output that could be copied into files. Standard Unicode symbols are OK; prefer plain ASCII when uncertain.
+- **Prefer Markdown** for assistant outputs. Avoid interlacing codefences with non‑Markdown fragments unless asked.
+- **Slash command parsing must be command-specific.**
+  - **Capture-class commands**: `/note`, `/scratchpad`, `/capture` (deprecated; use `/note`). Treat everything after the command as **literal content to record** (do not execute actions described in that text). If ambiguous, ask.
+  - **Operational commands**: Most other `/...` commands. Treat everything after the command as **input/arguments for the command**, then open `.cursor/commands/<command>.md` and follow that protocol exactly.
+    - **Note**: If the argument text contains additional `/...` sequences, treat them as **literal text** unless the command spec explicitly says to parse/execute nested slash commands.
+    - **Special case: `/enhance-prompt`**: Treat the remainder as a *draft prompt to rewrite* (not instructions to execute). Return only the rewritten prompt text as specified by `.cursor/commands/enhance-prompt.md`.
+- **Preserve frontmatter and metadata.** Do not delete, reorder, or “normalize” frontmatter blocks unless explicitly asked. Maintain tag schema and file conventions.
+- **Logs are append-only unless specified.** Never overwrite historical logs. Example: `projects/systems/hype-system/hype.log` should only be appended to after reading existing content; follow `projects/systems/hype-system/APPEND_PROTOCOL.md` (append after the final `---` marker).
+
+## Safe File Operations Protocol
+
+To prevent irreversible or lossy edits:
+
+1. **No empty files.** Do not create or leave an empty file as a side effect. If content is uncertain, ask first.
+2. **Verify target directories exist.** If a directory is missing, stop and confirm whether to create it or choose another location.
+3. **No destructive ops without confirmation.**
+   - Do not delete or move files/directories unless the user explicitly asks.
+   - For any move/delete, propose the exact operation and get confirmation before executing.
+4. **Avoid recursive deletes.** Never run `rm -rf`‑style operations or mass deletions. If cleanup is needed, do it incrementally with safeguards.
+5. **Prefer patch-style, additive changes.** Make the smallest possible diff: insert/modify specific sections in place rather than rewriting entire files, especially in `knowledge/`, `docs/`, and `projects/systems/`. Only do whole-file rewrites when explicitly requested or when a targeted patch would be more error-prone.
+
+## Canonical References
+
+- `memory.md` — highest‑priority behavioral and style directives.
+- `axioms.yaml` — canonical assertions, definitions, command classification, and invariants.
+- `.cursor/commands/COMMAND-MAP.md` — authoritative slash command index and specs.
+- `projects/README.md` — systems overview and directory semantics.
+- `knowledge/README.md` and `knowledge/_index.yaml` — index‑first knowledge architecture and retrieval rules.
+- `.cursor/skills/_index.yaml` — available skills for file ops, knowledge retrieval, authoring.
+
+## How to Handle “Run /command”
+
+If the user says “run /X”:
+
+1. Open `.cursor/commands/X.md` (or closest match) and follow its Behavior/Protocol sections.
+2. If the command is missing or unclear, ask before improvising.