The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the creation of a temporary file in /tmp containing an App Store Connect private key (.p8). While it uses chmod 600 and deletes the file after use, storing private keys in world-writable temporary directories is a suboptimal practice on shared systems.
[CREDENTIALS_UNSAFE]: The skill stores an OP_SERVICE_ACCOUNT_TOKEN in a persistent tmux environment using tmux set-environment. This persistence mechanism makes the sensitive token available to any process running within or attached to that specific tmux session.
[COMMAND_EXECUTION]: The skill executes a local repository script ./scripts/release-binaries.sh to perform high-privilege actions, including publishing to the npm registry and creating GitHub releases.
[DYNAMIC_EXECUTION]: Uses op run combined with bash -c to dynamically assemble and execute shell commands that manage authentication keys and interact with Apple's notarization services.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Reads metadata and versioning data from project files (package.json, CHANGELOG.md) and external tool outputs (npm view, gh release view).
Boundary markers: Absent; the skill does not use delimiters to isolate untrusted data from the instructions.
Capability inventory: Includes file system writes, network operations (publishing), and subprocess execution.
Sanitization: Absent; the skill interpolates content from local files and command outputs directly into shell environments and script arguments.

release-peekaboo