The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands including git status, find, git rev-parse, wt config, and rg (ripgrep) to understand the project structure and gather context. These are standard operations for a development-focused agent skill.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of ingesting data from the local repository and external research sources.
Ingestion points: Content is read from various project directories (docs, app, resources, tests) and gathered via external research capabilities when seeking best practices.
Boundary markers: There are no explicit delimiters or specific 'ignore embedded instructions' warnings applied to the data being searched or retrieved.
Capability inventory: The skill possesses the ability to execute shell commands and write research artifacts in TOML format to the local file system.
Sanitization: The instructions do not specify any validation or sanitization for the content retrieved from external sources before it is processed or stored.
[DATA_EXFILTRATION]: The skill proactively mitigates data exposure risks by instructing the agent to 'Never read secret files such as .env' during its evidence-gathering phase.

wt-idea