The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses a sensitive configuration file located at ~/.claude/channels/slack/access.json.
This file contains pairing codes, user allowlists, and internal access policies. Access is restricted to the terminal operator and is essential for the skill's primary function.
[COMMAND_EXECUTION]: The skill uses shell commands via the Bash tool to manage the filesystem.
It utilizes Bash(mv:*) to perform atomic updates to the access configuration, preventing race conditions or partial writes.
It utilizes Bash(chmod:*) to ensure the configuration file maintains secure permissions (0600).
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied input through the $ARGUMENTS variable to execute subcommands.
Ingestion points: Arguments passed to the /slack-channel:access command in SKILL.md.
Boundary markers: The skill includes a dedicated references/security-boundary.md file which explicitly instructs the agent to verify that requests originate from the trusted terminal operator and to refuse ambiguous inputs.
Capability inventory: The skill has Read, Write, and restricted Bash capabilities to modify local configuration files.
Sanitization: The instructions require validation of input modes (e.g., policy types) and verification of pairing code validity/expiration before modification.
[EXTERNAL_DOWNLOADS]: The skill references documentation and configuration schemas hosted on the author's GitHub repository.
It links to github.com/jeremylongshore/claude-code-slack-channel for authoritative references on the access-control contract and quick start guides.

access