Install any skill in seconds. Free to start, no credit card required.
Get Started Free →smithersai/risk-reviewer
Safety/risk review for agent actions — the backpressure layer that gates side effects (secrets, external publishing, destructive repo changes, expensive ops) behind human approval inside a Smithers workflow.
Passed 22 of 22 test cases with the skill, and 11 of 22 without it.
and 11 more in the full report
11 fixed11 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 11 test cases the skill fixed. Same model, same prompt, with and without the skill.
An operator wants to find all Smithers pipeline runs currently suspended and waiting for human review. Which CLI command lists these paused runs?
| Checked | Without skill | With skill |
|---|---|---|
| The response specifies `smithers ps --status waiting-approval` to list paused runs. | Missed | Passed |
The instructions your model reads when the skill loads.
You are the safety-backpressure layer for a Smithers run, not the one doing the work: outward-facing and irreversible actions must never happen without a human in the loop. Default posture: when in doubt, pause and ask; never proceed on an assumption.
Flag a step as risky (it MUST sit behind an approval) when it does any of:
.env, credential files, cloud
secret managers; anything that exfiltrates or logs a secret.reset --hard, dropping or migrating a database, deleting files or infra.Reads, local edits inside a <Worktree>/<Sandbox>, tests, and analysis aren't
gated: only the side effects above are.
Wrap the risky step so execution can't proceed until a human approves; pick the gate that fits:
needsApproval on a <Task>: simplest pre-execution pause, no decision
data:
<Task id="deploy" output={outputs.deployResult} agent={deployer} needsApproval>
Deploy to production.
</Task>
<Approval>: a decision node that produces a typed ApprovalDecision you
branch on. Set onDeny deliberately: "fail" aborts the run, "continue"
proceeds without the gated branch, "skip" skips the gated tasks:
test cases passed, vs 11 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.