BB agent-harness-builder
Build a complete multi-agent orchestration harness for an OpenClaw (or similar) agentic system — defining a CTO/orchestrator agent, tiered specialist agents, context-routing rules, model assignment, memory limits, Slack channel architecture, and a step-by-step implementation guide. Use this skill whenever someone wants to design, document, structure, or hand off an AI agent fleet; mentions orchestrating multiple agents, an "agent harness," agent roles/profiles, a CTO agent, agent routing, or distributing skills/context/models across agents; or wants a packaged architecture doc set they can give to an orchestrator agent or a team. Trigger this even if they just describe the problem ("I have 20 agents and need one to coordinate them") without using the word "harness."
Build a complete multi-agent orchestration harness for an OpenClaw (or similar) agentic system — defining a CTO/orchestrator agent, tiered specialist agents…
As a process B 73/100 · Nearly there — weak spots: inputs and preconditions, running it twice
What is at stake
The skill contains fragments that, in the wrong hands, cost money or data. Below: what the installer risks and what the author should do.
The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.
Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.
How to improve
- Address the high-severity findings: each costs 18 safety points. If one is a false positive, add the rule id to guard.allow in spec.yaml.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 1
-
high Dangerous commands
cmd-persistencereferences/implementation-guide-template.md:205Persistence mechanism (cron / launchd / scheduled task / autorun registry)launchctl bootstrap gui/$UID ~/Library/LaunchAgents/ai.openclaw.gateway.plist
Files scanned: 7. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 73/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 30Running it twice. 2 mutating operations with no state check
- 60Tools and files. Uses tools (bash) that frontmatter does not declare
- 70When it triggers. States when to use, but not when not to
- 85Steps. 24 steps, 2 vague phrases
- 100Result and completion. Output format and completion criterion are stated
- 100Failures and branches. 2 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 1484 tokens
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +4No input/output examples
- +1No license
- +2Single-language instructions
- +5Description quotes 3 example trigger phrases
- +3Description length 776: enough signal without eating the budget
- +4Structure: 10 headings
- +3Step-by-step instructions: 24 items
- +3Output format is stated explicitly
- +4Reference files are cited in the instructions (5 of 5)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 92.