v1.0.3-final - Final Hermes Runtime Refresh

v1.0.3-final - Final Hermes Runtime Refresh

This is a small final refresh for the archived public snapshot.

Changes

• Add a secret-free Hermes Agent config template.
• Add a Hermes environment example with empty credential placeholders only.
• Add install-hermes.sh to prepare ~/.hermes from the existing Agent SOUL files.
• Add deploy.sh as a conservative local deployment refresh wrapper.
• Document the Hermes Agent parallel runtime path in the Chinese README and quick-start guide.
• Fix the quick-start clone URL to the current public repository.

Scope

No full runtime test pass was rerun. Validation was limited to shell syntax checks, whitespace checks, and local Markdown link checks.

v1.0.2-final - Final Public Runtime Refresh

v1.0.2-final - Final Public Runtime Refresh

This is a small maintenance refresh for the archived final public snapshot.

Changes

• Align runtime guidance with OpenClaw v2026.5.19+ expectations by documenting Node.js 22.19+.
• Add the same Node.js 22.19+ engine requirement to the frontend package metadata and lockfile.
• Harden install.sh so it detects older Node.js versions before attempting the frontend build.
• Clean up one find invocation in the installer for more predictable shell behavior.
• Default the local backend/service binding to 127.0.0.1 while still allowing overrides through AGENTORCHESTRATOR_API_HOST.
• Keep the Chinese README entry, preview images, diagrams, and local README links intact.

Scope

No full runtime test pass was rerun for this archived snapshot refresh. The update was limited to code-level compatibility and documentation alignment.

v1.0.1-final - Final Public Snapshot Refresh

v1.0.1-final - Final Public Snapshot Refresh

This is a documentation and hardening refresh for the final public snapshot.
The project remains an archived final version rather than an actively maintained product line.

What changed

• Restored the main README.md as the Chinese authoritative entry point.
• Restored README navigation links and flow diagrams that were missing from the first final snapshot README rewrite.
• Kept the sanitized 19-image preview gallery in the main README.
• Refined docs/project-reflections.md to explain why the project stops here: good results, but too much orchestration overhead, token cost, waiting, and OpenClaw stall/recovery friction for the current model/tool era.
• Replaced machine-specific registry workspace paths with <openclaw-home>/workspace-* placeholders.
• Removed stale public references to real server packaging paths and default admin/admin wording.
• Added pytest-asyncio and asyncio_mode = auto for async test compatibility.
• Hardened public defaults: backend secret_key now defaults to a per-process random development key unless explicitly set by environment.
• Made OpenClaw home/workspace resolution environment-driven in scripts.
• Made script sync fall back to file copy when symlinks are unavailable, which helps Windows users without symlink privileges.
• Removed shell string construction from specialist/search smoke paths where task IDs, summaries, or queries could contain unsafe characters.

Validation performed

• Static scan for common private machine paths and hardcoded default credential strings.
• Python syntax compilation for the changed Python files.
• git diff --check.

Full test suites were intentionally not rerun for this refresh.

Release intent

This release supersedes v1.0.0-final as the preferred public snapshot. It keeps the same archival intent while making the repository presentation and public defaults cleaner.

v1.0.0-final - Final Public Snapshot

Multi-Agent Orchestrator v1.0.0-final

This is the final public snapshot of Multi-Agent Orchestrator.

The repository is being published as an archived, sanitized reference release rather than an actively maintained product. It captures the latest tuned state of the system: backend and frontend source, agent role definitions, registry specs, public documentation, examples, tests, and a refreshed README with 19 sanitized preview images.

Why this is final

The system proved that a task-centered multi-agent workbench can produce useful results: it can accept tasks, route work across roles, expose progress, preserve working context, and continue in a target-mode style without constant manual steering.

The project is being archived because the model and tool ecosystem has moved faster than this orchestration architecture. The external coordination layer became too expensive in practice: high token consumption, slow handoffs, heavy state management, and frequent waiting when the underlying OpenClaw runner stalled and had to recover. The final outputs were often good, but the cost and latency made continued investment hard to justify.

Included

• Sanitized final public source snapshot.
• Updated README with final positioning and preview gallery.
• Updated project reflection explaining why development stops here.
• Backend/frontend source, agent definitions, registry specs, scripts, docs, examples, and tests.
• Release assets in both .zip and .tar.gz formats.

Removed or excluded

• Runtime data, logs, ledgers, task workspaces, cold archives, snapshots, process files, caches, virtual environments, and dependency folders.
• Databases, auth/session material, cookies, credentials, API keys, local tokens, and machine-specific paths.
• Internal review notes, audit outputs, old screenshots, private task history, and temporary release-preparation files.

Validation

• Public safety scan passed for the repository tree.
• Release archive safety scan passed for both .zip and .tar.gz.
• git diff --check passed.
• Python syntax compilation passed for agentorchestrator, scripts, and tests.
• Frontend npm ci && npm run build passed.

Known caveats:

• Full pytest did not pass in the Windows release environment. Failures include missing async pytest plugin support, Windows symlink privilege limitations, and legacy tests that depend on local encoding/path behavior.
• npm audit reports 7 dependency advisories in the current frontend lockfile: 3 moderate and 4 high.

This release is intended as a final public archive and reference implementation, not as a promise of future maintenance.