Network AI
v4.15.3Local Python orchestration skill: multi-agent workflows via shared blackboard file, permission gating, token budget scripts, and persistent project context....
⭐ 6· 2k·8 current·8 all-time
byJovan Marinovic@jovancoding
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (local orchestration, blackboard, permission gating, token budgets) align with the included Python scripts and declared requirements (only python3). There are no unexpected credentials, external binaries, or cloud-specific config requested.
Instruction Scope
SKILL.md and the scripts confine operations to local files (data/ and swarm-blackboard.md) and provide explicit orchestration/delegation protocols. The docs repeatedly note that the host platform's sessions_send (outside this skill) may perform external model calls; that is a deliberate separation but gives the agent/operator responsibility to configure the host runtime if they want to prevent external network calls.
Install Mechanism
No install spec, no downloads, and the code claims to use only Python stdlib. requirements.txt is documentation-only and commented. This is low-risk from an install perspective.
Credentials
The skill requests no environment variables or external credentials. It does persist audit logs and grant tokens locally; the SKILL.md correctly warns not to include PII/secrets in justification fields. The only caution: local files (data/) may contain free-text justifications and grants, so protect their file permissions.
Persistence & Privilege
The skill writes persistent state under its own data/ directory (active_grants.json, audit_log.jsonl, project-context.json, etc.) but does not request elevated system privileges or modify other skills or global agent settings. always:false and normal invocation semantics are appropriate.
Assessment
This skill appears to do what it claims: local-only orchestration using Python and local files. Before installing, review and accept these trade-offs: (1) It creates and updates files under the skill directory (data/ and swarm-blackboard.md) — consider running it in an isolated workspace and set restrictive file permissions to protect audit logs and grant tokens. (2) The audit log and permission justification fields can contain free-text supplied by agents; do not put PII, secrets, or credentials into justifications. (3) Although the bundled scripts claim to make no network calls, the orchestration protocol expects the host platform (sessions_send) to perform any external model/API calls — if you need to prevent outbound network use, disable or control sessions_send at the platform level prior to using this skill. (4) If you require extra assurance, inspect the full scripts in this bundle (they are included) and run them in a sandboxed environment before granting broader access.Like a lobster shell, security has layers — review code before you run it.
auditvk9729450grsa8stweh9ff11gms81xr4jautogenvk9729450grsa8stweh9ff11gms81xr4jblackboardvk9729450grsa8stweh9ff11gms81xr4jcrewaivk9729450grsa8stweh9ff11gms81xr4jlangchainvk9729450grsa8stweh9ff11gms81xr4jlatestvk9704v1z4xbkc6vd2tnkp538qs84568bmcpvk9729450grsa8stweh9ff11gms81xr4jmulti-agentvk9729450grsa8stweh9ff11gms81xr4jorchestrationvk9729450grsa8stweh9ff11gms81xr4jpermissionsvk9729450grsa8stweh9ff11gms81xr4jsecurityvk9729450grsa8stweh9ff11gms81xr4jswarmvk9729450grsa8stweh9ff11gms81xr4j
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🐝 Clawdis
Binspython3