The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data ingested from external Axiom datasets through the axiom query and axiom stream tools. Maliciously crafted data within these datasets could potentially attempt to influence the agent's behavior.
Ingestion points: Data returned from axiom query "<APL>" and axiom stream <dataset> (referenced in SKILL.md and references/cli.md).
Boundary markers: None detected in the instructions for handling query output.
Capability inventory: The skill uses Bash(axiom query *), Bash(axiom stream *), and other file-system tools (Read, Grep, Glob).
Sanitization: No explicit sanitization or validation of the data retrieved from Axiom is mentioned.
[COMMAND_EXECUTION]: The skill relies on the execution of the axiom CLI tool for its primary functionality. This includes querying datasets, listing configurations, and managing authentication deployments.
Evidence: The allowed-tools section in SKILL.md includes multiple Bash invocations of the axiom command.
[CREDENTIALS_UNSAFE]: The skill provides instructions for managing and exporting sensitive authentication tokens using the CLI.
Evidence: references/cli.md advises using eval $(axiom config export --force) to export credentials to the environment or .env files. While documented as a method for secret management, this provides the agent with access to sensitive authentication materials.

axiom-apl