The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of a custom binary slack-pp-cli from non-whitelisted sources. Instructions include npx -y @mvanhorn/printing-press-library install slack and go install github.com/mvanhorn/printing-press-library/library/productivity/slack/cmd/slack-pp-cli@latest.\n- [CREDENTIALS_UNSAFE]: The skill manages and accesses sensitive configuration files, specifically credentials.toml and config.toml, which store Slack bearer tokens (SLACK_BOT_TOKEN and SLACK_USER_TOKEN).\n- [DATA_EXFILTRATION]: The skill provides a --deliver webhook:<url> flag that allows the agent to POST command output, which may contain sensitive Slack message history or user data, to any external URL.\n- [INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface as it mirrors external Slack content (messages, user profiles, thread history) into a local SQLite database and processes this untrusted data to answer queries.\n
Ingestion points: Slack workspace data synced to data.db.\n
Boundary markers: None mentioned for data interpolation.\n
Capability inventory: File write, shell command execution, and network POSTing (via webhooks).\n
Sanitization: PII stripping is advised during the teach step but not guaranteed for general history mirroring.\n- [DYNAMIC_EXECUTION]: The 'Automatic Learning' loop utilizes 'playbooks' stored in JSON files to record and replay sequences of shell commands with variable substitution ({slot}), creating a mechanism for dynamic command assembly and execution.\n- [COMMAND_EXECUTION]: The skill relies on the execution of the slack-pp-cli binary via the shell and explicitly requires the Read Bash tool in its frontmatter.

pp-slack