The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple external sources, creating a vulnerability surface where malicious instructions embedded in logs or messages could potentially influence agent behavior.
Ingestion points: External data enters the agent context via scripts/axiom-query (logs), scripts/grafana-query (metrics/logs), scripts/slack (messages), and scripts/sentry-api (error events).
Capability inventory: The agent has permissions for file writes (scripts/mem-write), network operations (scripts/slack, scripts/axiom-api), and Git operations (scripts/mem-sync, scripts/mem-share).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands within ingested data were found in the processing scripts.
Sanitization: Not implemented for external content before it is interpolated into the agent's context.
[CREDENTIALS_UNSAFE]: The skill manages sensitive API credentials and tokens for multiple observability platforms.
Evidence: scripts/config and scripts/init manage the storage and retrieval of Bearer tokens and Basic Auth credentials within the ~/.config/axiom-sre/config.toml file.
Security controls: The implementation follows security best practices by using umask 077 for the configuration directory and passing secrets via environment variables (e.g., eval "$(scripts/config ...)") to avoid exposure in command-line arguments.
[EXTERNAL_DOWNLOADS]: The skill facilitates team collaboration by synchronizing shared knowledge bases from external repositories.
Evidence: scripts/mem-sync and scripts/org-add utilize git pull and git clone to synchronize markdown-based memory entries from user-configured Git repositories.
[COMMAND_EXECUTION]: The skill employs a comprehensive suite of shell scripts to wrap API calls and system tools.
Evidence: The provided scripts execute various CLI utilities including curl, jq, git, kubectl, gh, and timeout/gtimeout to perform investigative tasks.

axiom-sre