@taskless/cli-nightly@0.13.0-20261014230341x6284621
Nightly build of [`@taskless/cli`](https://www.npmjs.com/package/@taskless/cli) — the same source and the same build, published from an unreleased commit on `main`.
AI Security Review
scanned 21d ago · by lpm-firewall-aiThe executable risk is limited to a user-invoked setup command that places Taskless-owned skill files in detected agent directories. I found no install-hook attack, credential redirect, or executable obfuscated payload in the inspected source.
Decision evidence
public snapshot- The CLI detects agent installations such as Claude Code and OpenCode and selects those directories as possible skill targets.
- The user-invoked `taskless init` command is for installing or updating Taskless skills and calls the installer.
- The installer writes package-owned Taskless skill files into selected agent directories and records the installation state.
- The manifest scripts are build, generation, and test tasks; there is no npm install lifecycle hook.
- The flagged `eval(payload)` and `api.example.com` strings occur in embedded rule demonstration fixtures.
- The agent setup is limited to Taskless skills and reference files in detected agent directories; the source inspected does not show an unrelated agent-control write or credential disclosure.
- The flagged evaluation and API URL samples are fixture content, not executable payload loading or a credential redirect.
Source & flagged code
4 flaggedManifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.
dist/index.jsView on unpkgFindings
Affected versions and remediation
This report applies to @taskless/cli-nightly@0.13.0-20261014230341x6284621.
See all versions of @taskless/cli-nightly for other recorded verdicts.
- Review the evidence and your use of @taskless/cli-nightly@0.13.0-20261014230341x6284621 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.