Subagent Dev
v1.0.1Use when executing implementation plans with independent tasks - coordinates task execution by dispatching subagents per task with verification checkpoints,...
⭐ 0· 169·1 current·1 all-time
by@axelhu
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name and description match the instructions: the SKILL.md describes dispatching per-task subagents via sessions_spawn, two-stage review (spec compliance + code quality), model selection, and integration with other workflow skills. None of the declared requirements (none) conflict with that purpose.
Instruction Scope
Instructions stay within the stated scope: extract tasks from a plan, spawn subagents with sessions_spawn(mode="run"), answer subagent questions, run tests, commit, and perform reviews. The doc explicitly prescribes how to pass context and handle subagent states. It does not instruct the agent to collect unrelated system secrets or contact external endpoints beyond the platform session model.
Install Mechanism
No install spec or code files — instruction-only skills are lowest-risk from an install perspective. There are no downloads, packages, or binaries declared.
Credentials
The skill declares no env vars or credentials, which is coherent. However, the workflow assumes the agent can access a project directory, run tests, and perform git commits; those actions typically require filesystem access and possibly VCS credentials configured in the environment even though they are not declared. This is a practical expectation, not necessarily a mismatch, but you should be aware the agent will need repository access to carry out the described steps.
Persistence & Privilege
Skill is instruction-only, does not request persistent presence (always:false) and does not modify other skills' configurations. Model invocation is not disabled (default), which is normal — the skill can be invoked autonomously if the agent decides to.
Assessment
This skill appears internally consistent with its purpose, but before enabling it consider: 1) Run it first in a sandbox/test repository — it expects to run tests and make commits, so avoid letting it modify production branches. 2) Ensure the agent's repository and git credentials are restricted and reviewed (the skill doesn't ask for creds but assumes git access). 3) Verify and trust the other workflow skills it references (superpowers-writing-plans, superpowers-tdd, superpowers-finishing-branch). 4) If you want tighter control, require manual approval before commits or disable autonomous invocation for agents using this skill. If you need more assurance, provide the actual runtime environment details (what sessions_spawn can access, which files/cwd the agent will see) and test the skill in a controlled environment.Like a lobster shell, security has layers — review code before you run it.
latestvk97edc0dx3bf0n0mkhgqf9qjax83n3nt
License
MIT-0
Free to use, modify, and redistribute. No attribution required.