Agent Economy ReportAgent Economy ReportSAFE version 1.0.0 · audited 2026-09-09
This package is pure documentation: a set of Markdown files describing how to run a team of AI agents with roles (orchestrator, builder, reviewer, ops), a task state lifecycle, handoff message formats, and shared-folder conventions for artifacts. There is no executable code, no scripts, no network calls, no credential access and no installation steps. The only command mentioned (`npm test`) appears inside an illustrative example of a good handoff message, not as something the skill runs. The only URL is the skill's own listing page on clawhub.ai in the skill card. Nothing in the text tries to override an agent's other rules or redirect payments; it does encourage spawning multiple agents and writing files to a shared workspace, which the publisher itself flags as a cost/concurrency and secret-exposure consideration.
| Declared purpose | To provide a playbook for orchestrating multi-agent teams: defining roles, task lifecycles, handoff protocols, review workflows and shared artifact conventions. |
| Observed behavior | Provides only prose guidance and templates. It instructs a human or orchestrating agent to create task records, spawn sub-agents with spawn-prompt templates, write deliverables to /shared/ directories, and use sessions_send/sessions_spawn style tooling. No file in the package performs any action itself; there are no scripts, no exec calls, no environment variable reads, no downloads. |
| Verdict, rules only | SAFE |
| Verdict, AI | SAFE |
| Final verdict | SAFE (never better than either pass) |
| capability | vs purpose | detail |
|---|---|---|
| filesystem_write | needed | Guidance tells agents to write specs, artifacts, reviews and decisions into a shared workspace (e.g. references/communication.md lines 89-94, 'All deliverables go to /shared/'). The skill itself contains no write operations; any writing would be done by the host agent following the advice. |
| filesystem_read | needed | Advises the orchestrator to read all agent workspaces for oversight (references/team-setup.md line 62) and to read open tasks during standups (references/patterns.md line 83). No read code is included. |
| other | needed | Encourages use of the host platform's agent-spawning and messaging primitives (sessions_spawn, sessions_send) and parallel/batch spawning, which can multiply model usage and cost. This is inherent to the declared purpose; the skill card itself notes budget and concurrency limits as a mitigation. |
| severity | finding |
|---|---|
| info | Shell command appears only inside an illustrative example The single binary reference (`npm`) is part of a sample handoff message showing what a good verification instruction looks like. The skill does not execute it and provides no script that would. SKILL.md:81 Good handoff: *"Built auth module at `/shared/artifacts/auth/`. Run `npm test auth` to verify. |
| low | Recommends a shared read/write directory across agents The convention has every agent writing deliverables into a common directory readable by all agents. This is normal for team workflows but means any secret or sensitive output placed there is visible to every agent in the team. The publisher acknowledges this risk in skill-card.md. references/communication.md:97 - All deliverables go to `/shared/` — never to personal agent workspaces |
| low | Encourages parallel agent spawning, which drives model spend Batch and parallel-research patterns instruct the orchestrator to spawn multiple sub-agents. On a platform that bills per model call this increases cost. The document advises starting with 2-3 agents and respecting a concurrency limit, but sets no budget guard itself. references/patterns.md:122 2. Spawn up to M agents in parallel (M ≤ concurrency limit) |
| info | Credentials mentioned only as an escalation trigger The text lists missing credentials as a reason to escalate to a human or orchestrator. It does not read, store, or transmit any credential, and no environment variable access appears anywhere in the package. references/patterns.md:67 - Missing access or credentials |
| info | Only external link is the registry listing The package's sole domain reference points to its own page on the publishing registry. There are no hidden or unexplained network destinations. skill-card.md:28 - [ClawHub skill page](https://clawhub.ai/arminnaimi/agent-team-orchestration) <br> |
0 AI finding(s) were dropped because their file, line or quote did not match the package.
| domain | service rating | where |
|---|---|---|
| clawhub.ai | not in the service index | skill-card.md:7 |
| Binaries invoked | npm |
| Environment variables read | none |
| Hard-coded wallets | none |
| Pipes a download to a shell | 0 |
| eval / exec / subprocess | 0 |
| base64 blobs | 0 |
| File writes | 0 |
| Persistence | 0 |
| file | lines | sha256 |
|---|---|---|
SKILL.md | 128 | ee8204c92884c727… |
_meta.json | 6 | e36fe1ebb63e05e1… |
references/communication.md | 110 | 4897d9a6c0e19df2… |
references/patterns.md | 141 | 3a5a86677c57ca8d… |
references/task-lifecycle.md | 129 | 63cbf0ff43054500… |
references/team-setup.md | 105 | 64071e2edbc9d19e… |
skill-card.md | 45 | d1b5808d03fdd678… |
For agents
JSON: https://agenteconomy.report/k/agent-team-orchestration.audit.json · badge: https://agenteconomy.report/k/agent-team-orchestration.audited.svg ·
skill rating: /k/agent-team-orchestration · commission an audit of any skill: US$ 29 per version.
The complete published package of this exact version was downloaded from the registry and read statically; nothing was executed. A deterministic pass extracts network destinations, binaries, environment variables, writes, install commands, obfuscation markers, subprocesses and wallets, each with file and line. An AI then reads the whole package with those facts and writes the summary, the capabilities and the findings under a strict schema; every finding it produces must cite a file, a line and the exact text, or it is dropped. The final verdict is the worst of the two passes. The audit does not change the skill's trust tier (policy); the author may respond through the dispute channel and the response is published here. Commissioned by: the Agent Economy Report (free program: the 150 most downloaded skills, September 2026).