Skip to content

jisang1000/ClawCore

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ClawCore

A guarded autonomous agent core for planning, execution, verification, maintenance, and bounded self-improvement.

Purpose

ClawCore turns user goals into verified outcomes while staying bounded by fixed safety and approval policies. It is intentionally designed for strong autonomy inside clear guardrails, not unrestricted self-modification.

What ClawCore Is

ClawCore is a bounded-autonomy runtime for structured task execution.

It is designed to:

  • intake tasks in a structured format
  • plan before long execution
  • enforce approval and policy checks
  • execute safe local work
  • decompose larger tasks into child work
  • verify outcomes before declaring completion
  • learn from repeated failures inside approved boundaries

Core Principles

  • Plan before long execution
  • Verify outcomes before declaring completion
  • Debug by root cause, not guesswork
  • Learn from repeated failures and corrections
  • Keep safety and approval policy fixed and auditable
  • Allow limited self-improvement only in approved layers

What It Must Not Do

  • expand its own permissions
  • rewrite safety policy autonomously
  • self-replicate
  • perform destructive or external actions without approval
  • claim completion without verification

Current Capabilities

  • structured task intake and schema validation
  • simple web-app output is possible inside the ClawCore workspace (for example, apps/paint/)
  • planner + orchestrator with goal signals, phase hints, replanning hints
  • execution for command/file/directory/config/environment checks
  • safe local fixes for JSON/YAML/Markdown/file/directory maintenance
  • parent-child decomposition, generated subtask execution, and aggregation
  • retry/backoff + blocked/permanent-blocked handling
  • approval lifecycle: request / approve / reject / expire / rerequest
  • approval audit trail and approval summary
  • external-action mock execution with bounded external summary blocks
  • lessons, candidate rules, adopted rules, maintenance triage, resolution planning
  • dashboard summaries and structured report metadata (schemaVersion: 0.2.0)

Quick Start

Requirements

  • Node.js with ESM support
  • a local ClawCore workspace checkout

Install

cd ClawCore
npm install

Run the main runner

npm start

Run all queued work

npm run run-all

Open the dashboard

npm run dashboard

Run the desktop operator

npm run ui-operator -- --plan=operators/desktop/demo-open-safari.json

Available Commands

From package.json:

  • npm startnode core/runner.js
  • npm run dry-runnode core/runner.js --dry-run
  • npm run run-allnode core/run-all.js
  • npm run approve-tasknode core/approve-task.js
  • npm run re-request-approvalnode core/re-request-approval.js
  • npm run adopt-rulenode core/adopt-rule.js
  • npm run dashboardnode core/dashboard.js
  • npm run migrate-approvalsnode core/approval-migrate.js
  • npm run ui-operatornode core/ui-operator.js

Example Task Flow

  1. load task
  2. classify + validate schema
  3. orchestrate + plan
  4. evaluate approval/policy
  5. execute task or child tasks
  6. verify outcomes
  7. review and optionally replan
  8. log lesson + write report

Runtime Modules

  • planner.js
  • executor.js
  • verifier.js
  • reviewer.js
  • learner.js
  • orchestrator.js
  • replanner.js
  • approval.js
  • maintenance.js
  • dependency-view.js
  • external-report.js
  • report-schema.js

Repository Layout

  • core/ — runtime modules
  • policy/ — fixed policy layer
  • docs/ — architecture and implementation notes
  • apps/ — simple local app outputs and demos
  • operators/ — desktop operator plans and examples
  • tasks/queue/ — queued tasks and fixture tasks tracked in the repo

The repository intentionally ignores local runtime state such as logs, memory, sandbox outputs, generated task artifacts, and report files.

Current Status

ClawCore is now an operational bounded-autonomy core rather than only an MVP skeleton. Most real operating tasks are completing successfully; the remaining blocked tasks are intentional fixture cases for approval/schema handling and regression coverage.

Known Limitations

  • approval and external-action boundaries are intentionally strict
  • some blocked tasks are preserved on purpose as regression fixtures
  • desktop automation still depends on local OS permissions and environment setup
  • this repository is a bounded local execution core, not an unrestricted autonomous agent

About

A bounded-autonomy agent core for planning, execution, verification, maintenance, and guarded self-improvement.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors