The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests service names and log-derived metrics from the PostHog API, creating a vulnerability surface where untrusted data could influence agent behavior.
Ingestion points: The agent calls posthog:logs-services and posthog:logs-count-ranges (in SKILL.md) to retrieve candidate services and their log distributions.
Boundary markers: There are no explicit delimiters or instructions to the agent to ignore potentially malicious content embedded within service names or log metadata.
Capability inventory: The skill has write access to the project's alert configuration and can create notification destinations (Slack channels and Webhook URLs) via posthog:logs-alerts-create and posthog:logs-alerts-destinations-create.
Sanitization: No sanitization steps are defined for the ingested data before it is interpolated into alert names or configurations.
[COMMAND_EXECUTION]: The workflow relies on the execution of local scripts to process data and generate threshold suggestions.
In SKILL.md, the agent is instructed to pipe JSON output from log tools into the included scripts/baseline_stats.py file using python3.
In references/volume-floor-alerts.md, a python3 -c one-liner is provided as a template for the agent to calculate specific volume-floor thresholds.
These executions are constrained to the skill's provided scripts and standard Python libraries, operating on locally retrieved JSON data.

authoring-log-alerts