All in Luna

https://github.com/zenx0x/allinluna
Admin
All in Luna is a durable, resource-aware multi-agent orchestration runtime that breaks complex...
Tokens:14,421
Snippets:244
Trust Score:6.8
License:Apache-2.0
Update:1 month ago
Show doc for...
Tokens:
Raw
### Install and Initialize All in Flash Profile

Source: https://github.com/zenx0x/allinluna/blob/main/plugins/deepseek-harness/README.md

Commands to install the required Python package, initialize the DSH profile, and start the DeepSeek Harness environment.

```powershell
python -m pip install "allinluna==2.0.0rc3"
npx @zenx0x/allinflash@0.2.0 init --profile allinflash
dsh --profile allinflash
```

--------------------------------

### Manage runs via CLI

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Install the package and use CLI commands to start, monitor, and drive task execution.

```bash
python -m pip install "allinluna==2.0.0rc3"

allinluna start --goal "Finish the authentication refactor"
allinluna status RUN_ID
allinluna drive RUN_ID
```

--------------------------------

### Install for development

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Use editable mode for repository development.

```bash
python -m pip install -e .
```

--------------------------------

### Prepare the canary action

Source: https://github.com/zenx0x/allinluna/blob/main/docs/release-canaries/v2.0.0-rc.2/README.md

Initialize the canary database and start the action using the CLI.

```powershell
$db = Join-Path $env:TEMP "allinluna-rc2-desktop-canary.db"
allinluna --db $db start `
  --intent-id desktop-canary-v2-rc2 `
  --goal "Run the v2.0.0-rc.2 real Desktop canary" `
  --model "gpt-5.3-codex-spark" `
  --reasoning high
```

--------------------------------

### Install Development Profile

Source: https://github.com/zenx0x/allinluna/blob/main/plugins/deepseek-harness/README.md

Alternative installation path for repository contributors using the local checkout and development scripts.

```bash
python scripts/install_allinflash_profile.py --force
```

--------------------------------

### Verify Installation

Source: https://github.com/zenx0x/allinluna/blob/main/plugins/deepseek-harness/README.md

Command to check the health and configuration of the installed All in Flash profile without launching the server.

```powershell
npx @zenx0x/allinflash@0.2.0 doctor --profile allinflash
```

--------------------------------

### Run distribution and installation validation scripts

Source: https://github.com/zenx0x/allinluna/blob/main/docs/releases/v2-rc2-distribution-integrity.md

Execute these commands to verify distribution integrity and installation consistency before release.

```text
python scripts/validate_distributions.py
python scripts/validate_installations.py
python -m pytest tests/test_distributions.py tests/integration/test_distribution_integrity.py -q
```

--------------------------------

### Prepare the runtime environment

Source: https://github.com/zenx0x/allinluna/blob/main/docs/phase3-desktop-canary/README.md

Initializes the runtime database and starts the canary process using the specified model and reasoning level.

```powershell
$env:PYTHONPATH = "plugins/allinluna/runtime"
$db = "canary-runtime.db"
python -m allinluna_runtime.cli --db $db start --goal "Run the phase 3 Desktop canary" --model "gpt-5.3-codex-spark" --reasoning high
```

--------------------------------

### CLI Command Reference

Source: https://github.com/zenx0x/allinluna/blob/main/docs/migration-guide.md

Standard commands for starting, driving, and managing task lanes in the All in Luna CLI.

```text
allinluna start --goal "..."
allinluna drive RUN_ID
allinluna lane start RUN_ID TASK_ID
allinluna lane tick RUN_ID TASK_ID
allinluna lane drive RUN_ID TASK_ID
allinluna lane handoff RUN_ID TASK_ID
```

--------------------------------

### Submit a goal to All in Luna

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/quickstart.md

Examples of natural language goal submissions for the All in Luna plugin.

```text
Use All in Luna to finish the authentication refactor.
Keep independent parts moving in parallel where possible.
```

```text
用 All in Luna 完整完成这个项目。
能独立推进的部分尽量并行。
```

--------------------------------

### Map capabilities to routes

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/models-and-performance.md

Example of a portable policy mapping specific capability classes to deployment-specific routes.

```text
planning.semantic → use the deployment's strong planning route
work.mechanical   → use the deployment's fast route
verify.independent → use an independent verification route
```

--------------------------------

### Task Organization Example

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

A visual representation of how All in Luna structures a complex authentication refactor into independent, manageable tasks.

```text
Authentication refactor

├─ Task 1 — Auth backend
│    ├─ session/token logic
│    ├─ API
│    └─ backend tests
│
├─ Task 2 — Frontend auth flow
│    ├─ login
│    ├─ logout
│    └─ protected routes
│
├─ Task 3 — Migration
│    └─ waits for auth contract
│
└─ Task 4 — Integration
     └─ waits for backend + frontend
```

--------------------------------

### Visualize task status

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/quickstart.md

Example output showing the parallel execution state of tasks.

```text
Refactor authentication across backend, frontend, migration, and tests.
```

```text
Auth backend        ● running
Frontend auth flow  ● running
Migration           ○ waiting for auth contract
Integration         ○ waiting for backend + frontend
```

--------------------------------

### Task Hierarchy Visualization

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Example of how a top-level goal is broken down into independent work domains.

```text
Add billing to this app

├─ Billing backend
├─ Checkout UI
├─ Database migration
├─ Integration tests
└─ Documentation
```

--------------------------------

### Recursive Task Decomposition

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Example of how a complex task can be further subdivided into smaller work units.

```text
Backend Task
 ├─ API
 ├─ database changes
 ├─ tests
 └─ migration checks
```

--------------------------------

### Public Relay Action Sequence

Source: https://github.com/zenx0x/allinluna/blob/main/docs/release-canaries/v2.0.0-rc.2/t1-correction-attempt-3-README.md

The sequence of actions performed by the installed CLI to facilitate producer and consumer work handoffs.

```text
lane next-actions
-> signed lane-direct-work/v1
-> bounded local work plus ArtifactStore.put
-> direct-work-result/v1
-> lane ingest-direct-result
-> independently verified work-handoff/v1
-> installed-wheel EvidenceCollector lane-handoff/v1
-> CoordinatorEngine.ingest_handoff
```

--------------------------------

### View CLI help

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Display available CLI commands and options.

```bash
allinluna --help
```

--------------------------------

### Initialize DeepSeek Harness profile

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Set up a dedicated DSH profile for All in Flash integration.

```bash
npx @zenx0x/allinflash@0.2.0 init --profile allinflash
dsh --profile allinflash
```

--------------------------------

### Sol Advisor and All in Luna Orchestration Models

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Compares the architectural orchestration boundaries between Sol Advisor-style systems and All in Luna.

```text
Strong Primary Architect
→ bounded implementation / review workers
```

```text
Global Coordinator
→ multiple persistent Top-level Tasks
→ each Task has its own workflow and workers
```

--------------------------------

### Run performance benchmarks

Source: https://github.com/zenx0x/allinluna/blob/main/docs/phase2-acceptance.md

Execute these scripts to reproduce the scale report measurements for the store, scheduler, and context modules.

```text
python scripts/benchmark_store_scale.py
python scripts/benchmark_scheduler.py
python scripts/benchmark_context.py
python scripts/validate_core_slim.py
```

--------------------------------

### Execute and monitor goals via CLI

Source: https://github.com/zenx0x/allinluna/blob/main/docs/examples/plain-goal.md

Commands for initiating a goal and inspecting the status of a durable run.

```text
allinluna start --goal "Finish the authentication refactor across backend, frontend, migration, and tests"
```

```text
allinluna status RUN_ID
allinluna next-actions RUN_ID
allinluna drive RUN_ID
```

--------------------------------

### Run host conformance checker

Source: https://github.com/zenx0x/allinluna/blob/main/docs/host-conformance.md

Execute the conformance script in either fixture or real mode using a specified trace file.

```powershell
python scripts/host_conformance.py --mode fixture
python scripts/host_conformance.py --mode real --trace C:\path\to\host-trace.json
```

--------------------------------

### CLI Command Reference

Source: https://github.com/zenx0x/allinluna/blob/main/plugins/allinluna/skills/allinluna/SKILL.md

A list of available commands for managing execution runs and task lanes within the AllinLuna system.

```APIDOC
## CLI Commands

### Execution Management
- `allinluna start --goal "..."`: Start a new execution goal.
- `allinluna status RUN_ID`: Get the status of a specific run.
- `allinluna next-actions RUN_ID`: List next actions for a run.
- `allinluna ingest-receipt RUN_ID RECEIPT.json`: Ingest a receipt for a run.
- `allinluna pause RUN_ID`: Pause a run.
- `allinluna resume RUN_ID`: Resume a paused run.
- `allinluna retry RUN_ID --task TASK_ID`: Retry a specific task in a run.
- `allinluna cancel RUN_ID --task TASK_ID`: Cancel a specific task in a run.
- `allinluna reconcile RUN_ID`: Reconcile a run.
- `allinluna drive RUN_ID`: Drive a run.

### Lane Management
- `allinluna lane start RUN_ID TASK_ID`: Start a task lane.
- `allinluna lane status RUN_ID TASK_ID`: Get status of a task lane.
- `allinluna lane tick RUN_ID TASK_ID`: Tick a task lane.
- `allinluna lane drive RUN_ID TASK_ID`: Drive a task lane.
- `allinluna lane next-actions RUN_ID TASK_ID`: List next actions for a task lane.
- `allinluna lane ingest-receipt RUN_ID TASK_ID RECEIPT.json`: Ingest a receipt for a task lane.
- `allinluna lane ingest-direct-result RUN_ID TASK_ID RESULT.json`: Ingest a direct result for a task lane.
- `allinluna lane handoff RUN_ID TASK_ID`: Perform a handoff for a task lane.
```

--------------------------------

### Allinluna CLI Commands

Source: https://github.com/zenx0x/allinluna/blob/main/plugins/allinluna/skills/allinluna/SKILL.md

Standard CLI commands for managing runs and task lanes within the Allinluna environment.

```text
allinluna start --goal "..."
allinluna status RUN_ID
allinluna next-actions RUN_ID
allinluna ingest-receipt RUN_ID RECEIPT.json
allinluna pause RUN_ID
allinluna resume RUN_ID
allinluna retry RUN_ID --task TASK_ID
allinluna cancel RUN_ID --task TASK_ID
allinluna reconcile RUN_ID
allinluna drive RUN_ID
allinluna lane start RUN_ID TASK_ID
allinluna lane status RUN_ID TASK_ID
allinluna lane tick RUN_ID TASK_ID
allinluna lane drive RUN_ID TASK_ID
allinluna lane next-actions RUN_ID TASK_ID
allinluna lane ingest-receipt RUN_ID TASK_ID RECEIPT.json
allinluna lane ingest-direct-result RUN_ID TASK_ID RESULT.json
allinluna lane handoff RUN_ID TASK_ID
```

--------------------------------

### Execute a task with natural language

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Use natural language prompts to initiate task automation without manual configuration.

```text
Use All in Luna to finish the authentication refactor.
Keep independent parts moving in parallel where possible.
```

--------------------------------

### Run Release Validation Commands

Source: https://github.com/zenx0x/allinluna/blob/main/docs/phase3-release-acceptance.md

Execute these commands serially from the repository root to validate the release candidate.

```text
python -m pytest tests -q
python scripts/validate_vnext_tests.py --run --performance --json
python scripts/validate_repository.py
python scripts/validate_core_slim.py
python scripts/validate_distributions.py
python scripts/validate_installations.py
```

--------------------------------

### Conceptual evidence path

Source: https://github.com/zenx0x/allinluna/blob/main/docs/examples/plain-goal.md

Visualizes the hierarchy from goal definition to coordinator completion.

```text
Goal
 -> Top-level Task(s)
 -> WorkUnit(s)
 -> checks / immutable artifacts
 -> work-handoff/v1
 -> lane-handoff/v1
 -> Coordinator completion decision
```

--------------------------------

### Recovering an Interrupted Run

Source: https://github.com/zenx0x/allinluna/blob/main/docs/troubleshooting/common-issues.md

Use these commands to recompute ready actions from the existing runtime database after an interruption.

```text
allinluna status RUN_ID
allinluna reconcile RUN_ID
allinluna drive RUN_ID
```

--------------------------------

### Compile a goal with the Python API

Source: https://github.com/zenx0x/allinluna/blob/main/docs/examples/plain-goal.md

Uses SinglePublicSkillAPI to transform a goal description and completion criteria into a task graph.

```python
from allinluna_runtime.packs import SinglePublicSkillAPI

compiled = SinglePublicSkillAPI().compile({
    "goal": "Finish the authentication refactor across backend, frontend, migration, and tests",
    "done_when": [
        "the authentication backend changes are implemented",
        "the frontend authentication flow is updated",
        "required migration work is complete",
        "targeted and integration checks pass",
        "changed paths and evidence are available",
    ],
})

print(compiled.task_graph.to_dict())
```

--------------------------------

### Define task and work unit overrides

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/models-and-performance.md

Demonstrates how specific tasks or work units can narrow or specialize the general run policy.

```text
Run default        → balanced
Architecture Task  → stronger reasoning
Mechanical WorkUnit → fast route
Final verification → independent strong route
```

--------------------------------

### All in Luna Orchestration Boundary

Source: https://github.com/zenx0x/allinluna/blob/main/README.en.md

Illustrates the higher-level orchestration boundary used by All in Luna.

```text
Global Coordinator
→ multiple persistent Top-level Tasks
→ each Task has its own workflow and workers
```

--------------------------------

### Define Project-Backed Task Target

Source: https://github.com/zenx0x/allinluna/blob/main/docs/migration-guide.md

Use this JSON structure for repository-backed tasks after resolving the project ID and environment.

```json
{
  "type": "project",
  "projectId": "<resolved-project-id>",
  "environment": {"type": "worktree"}
}
```

--------------------------------

### Model Reasoning Allocation

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

A conceptual mapping of development phases to appropriate reasoning levels for model performance optimization.

```text
Planning        → stronger reasoning
Implementation  → balanced
Mechanical work → fast / efficient
Verification    → strong / independent
```

--------------------------------

### Sol Advisor Architectural Pattern

Source: https://github.com/zenx0x/allinluna/blob/main/README.en.md

Represents the structural approach of Sol Advisor-style systems.

```text
Strong Primary Architect
→ bounded implementation / review workers
```

--------------------------------

### Define Runtime Boundary Flow

Source: https://github.com/zenx0x/allinluna/blob/main/distributions/overlays/research-routes/README.en.md

Visual representation of the data flow from question to explicit handoff.

```text
question or evidence package -> route map -> claims/evidence -> failure/recovery records -> reversible probe -> HumanDecision seam -> explicit handoff
```

--------------------------------

### Define capability classes

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/models-and-performance.md

List of supported capability classes for mapping semantic requirements to concrete model routes.

```text
control.relay
planning.semantic
lane.synthesis
work.mechanical
work.implementation
work.deep-debug
verify.independent
```

--------------------------------

### Run project validation and tests

Source: https://github.com/zenx0x/allinluna/blob/main/CONTRIBUTING.md

Execute the test suite and repository validation scripts to ensure code integrity before submission.

```bash
python -m unittest discover -s tests -v
```

```bash
python scripts/validate_repository.py
```

--------------------------------

### Defining a Projectless Target

Source: https://github.com/zenx0x/allinluna/blob/main/docs/troubleshooting/common-issues.md

Use this explicit target format for work that does not involve a repository.

```json
{"type":"projectless"}
```

--------------------------------

### Ingest and drive the receipt

Source: https://github.com/zenx0x/allinluna/blob/main/docs/release-canaries/v2.0.0-rc.2/README.md

Commands to ingest the desktop receipt and drive the canary status.

```powershell
allinluna --db $db ingest-receipt desktop-canary-v2-rc2 .\desktop-receipt.json
allinluna --db $db drive desktop-canary-v2-rc2
allinluna --db $db status desktop-canary-v2-rc2
```

--------------------------------

### Execute lane-local commands

Source: https://github.com/zenx0x/allinluna/blob/main/docs/user/quickstart.md

Commands for manual control of specific lanes within a run, typically used for debugging or host integrations.

```bash
allinluna lane start RUN_ID TASK_ID
allinluna lane tick RUN_ID TASK_ID
allinluna lane next-actions RUN_ID TASK_ID
allinluna lane ingest-direct-result RUN_ID TASK_ID RESULT.json
allinluna lane handoff RUN_ID TASK_ID
```

--------------------------------

### Parallel Task Status Tracking

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Visualization of concurrent task execution and dependency waiting states.

```text
Billing backend       ● running
Checkout UI           ● running
Documentation         ● running
Database migration    ○ waiting for schema
Integration tests     ○ waiting for backend
```

--------------------------------

### GSD Workflow Steps

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Defines the explicit development workflow sequence used in GSD.

```text
clarify
→ specify
→ decompose
→ implement
→ verify
→ integrate
```

--------------------------------

### Desktop Canary Relay Sequence

Source: https://github.com/zenx0x/allinluna/blob/main/docs/release-canaries/v2.0.0-rc.2/t1-correction-attempt-2-README.md

The sequence of actions performed during the Desktop canary execution.

```text
lane next-actions
→ lane-direct-work/v1
→ bounded local work
→ direct-work-result/v1 with typed export
→ lane ingest-direct-result
→ evidence-backed work-handoff/v1
→ evidence-backed lane-handoff/v1
→ producer export releases consumer
→ root completed
```

--------------------------------

### Define Projectless Task Target

Source: https://github.com/zenx0x/allinluna/blob/main/docs/migration-guide.md

Use this JSON structure for tasks that do not require a project association.

```json
{"type": "projectless", "directoryName": "<runtime-generated-name>"}
```

--------------------------------

### All in Luna Core Responsibilities

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Lists the primary functional areas managed by the All in Luna core system.

```text
top-level tasks
dependencies
scheduling
context
results
recovery
```

--------------------------------

### Ingest receipt and drive execution

Source: https://github.com/zenx0x/allinluna/blob/main/docs/phase3-desktop-canary/README.md

Finalizes the process by ingesting the completed receipt and driving the execution for the specified run ID.

```powershell
python -m allinluna_runtime.cli --db $db ingest-receipt <RUN_ID> .\desktop-receipt.json
python -m allinluna_runtime.cli --db $db drive <RUN_ID>
```

--------------------------------

### Required action fields for relay

Source: https://github.com/zenx0x/allinluna/blob/main/docs/phase3-desktop-canary/README.md

Specifies the exact fields that must be copied from the emitted action when performing the Desktop call.

```text
target, prompt, model, thinking (when present), title
```

--------------------------------

### Task Lifecycle Flow

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Representation of the persistent workflow allowing tasks to be stopped and resumed.

```text
start
→ work
→ stop
→ come back
→ resume
```

--------------------------------

### Evidence layer flow

Source: https://github.com/zenx0x/allinluna/blob/main/examples/resource-policies/README.md

Visual representation of the three separate evidence layers in the resource policy resolution process.

```text
requested capability/policy -> resolved host route -> actual host observation
```

--------------------------------

### Main Conversation Status Indicators

Source: https://github.com/zenx0x/allinluna/blob/main/README.md

Standardized symbols used to represent the state of tasks in the main conversation.

```text
✓ what is done
● what is running
○ what is waiting
! what needs your decision
```

--------------------------------

### Define the canary sequence

Source: https://github.com/zenx0x/allinluna/blob/main/docs/release-canaries/v2.0.0-rc.2/README.md

The sequence of operations for the final corrective qualification run.

```text
SinglePublicSkillAPI.start
→ CoordinatorDriver
→ real Desktop Producer Lane
→ lane-direct WorkUnit and verified export
→ WorkHandoff → LaneHandoff → HandoffProcessor
→ Consumer becomes ready
→ real Desktop Consumer Lane
→ lane-direct WorkUnit → WorkHandoff → LaneHandoff → HandoffProcessor
→ root Run completed
```
Tokens:
Raw