The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes output from 'bd timeline logs' to identify user journey steps and identifiers. This external data represents an attack surface where malicious content in logs could attempt to influence the agent's logic or the content of the workflows it deploys.
Ingestion points: File references/discovery.md (Step 3) processes log data fetched from user-provided session IDs via bd timeline logs {{SESSION_ID}}.
Boundary markers: Absent. The instructions do not provide explicit delimiters or warnings to the agent to treat the ingested log content as untrusted data.
Capability inventory: The agent has the capability to execute shell commands (via the bd CLI) to create, update, and deploy workflows, alerts, and dashboards using parameters extracted from these logs.
Sanitization: Absent. The skill does not define specific sanitization or validation steps for the content extracted from logs before it is interpolated into subsequent command arguments or JSON payloads.
[DYNAMIC_EXECUTION]: The skill dynamically generates complex JSON payloads for bitdrift workflows, alerts, and dashboards based on user-supplied values and analysis of historical data.
Evidence: Multiple templates in references/funnel.md, references/sankey.md, and references/dashboard.md are populated at runtime and then passed to bd workflow create or bd dashboard create commands.
[COMMAND_EXECUTION]: The skill relies on executing shell commands through the bd CLI tool. User-provided identifiers such as Session IDs, Journey Names, and Step Identifiers are interpolated into these commands.
Evidence: Commands like bd timeline hydrate {{SESSION_ID}} and bd workflow describe {{WORKFLOW_ID}} are used throughout the setup phases, presenting a potential path for command injection if the underlying agent does not properly handle shell metacharacters in user input.

bd-cuj