Nm Abstract Subagent Testing
v1.8.2Test skills via RED/GREEN/REFACTOR TDD with fresh subagents
⭐ 0· 11·0 current·0 all-time
by@athola
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description (subagent TDD testing) matches the content: the skill is an instruction-only guide for running baseline/with-skill/rationalization tests using fresh subagents. It does not declare or request any unrelated binaries, env vars, or credentials.
Instruction Scope
SKILL.md only instructs how to create fresh conversations, run identical prompts, compare outputs, and document rationalizations. It does not instruct reading local files, accessing external endpoints, or exfiltrating data. The only external dependency mentioned is an optional 'Claude Code plugin' for a richer experience — this is descriptive, not required by the skill.
Install Mechanism
No install spec and no code files to install; this is an instruction-only skill. There is no download, extraction, or third-party package installation described.
Credentials
The skill declares no required environment variables, credentials, or config paths. The instructions do not ask for any secrets or unrelated credentials.
Persistence & Privilege
Flags show always:false and user-invocable:true (defaults). The skill does not request persistent presence or elevated system privileges, nor does it modify other skills or system-wide settings.
Assessment
This skill is a methodology document for testing skills with fresh subagent instances and appears internally consistent. Before using it: (1) avoid putting real secrets, credentials, or private data into test prompts or logs, since tests capture full model responses; (2) when spinning fresh subagents, ensure those subagents are given only the minimal permissions needed (so tests can't inadvertently access other services or secrets); (3) if you install or use the optional 'Claude Code' plugin mentioned, review that plugin separately for its install/credential requirements; (4) confirm your platform supports creating fresh subagents as the document intends. Other than these operational precautions, the skill itself contains no instructions that are disproportionate or suspicious.Like a lobster shell, security has layers — review code before you run it.
latestvk9760yhjeqkr1384wg20b6w2hd84b2xy
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧪 Clawdis