The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands to gather metadata for the walkthrough. It uses git branch --show-current to identify the active branch, git diff --stat to list file changes, and git log to summarize recent commits. Additionally, it uses mkdir and cat for directory management and reading the local plan.md file. These commands are typical for development tools and are confined to the project's worktree.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by ingesting data from a file that could theoretically contain malicious instructions.\n
Ingestion points: The skill reads the contents of .agent/${BRANCH}/plan.md to extract implementation details for the summary.\n
Boundary markers: There are no explicit delimiters or 'ignore' instructions used when reading from plan.md or when interpolating that data into the walkthrough.md template.\n
Capability inventory: The skill has the ability to execute shell commands (bash) and perform file system read/write operations.\n
Sanitization: The skill does not perform sanitization or structural validation on the content extracted from plan.md before processing it.

aw-create-walkthrough