The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to perform repository maintenance and verification tasks.
Evidence: Use of a pipeline involving find, sort, xargs, and sha256sum to generate byte-level fingerprints of the specification tree for integrity checks.
Evidence: Use of ripgrep (rg) with fixed-string matching to search for change identifiers within proposal files.
Evidence: Invocation of a project-specific openspec command-line tool for archiving change directories.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting data from project artifacts.
Ingestion points: Reads content from proposal.md files and authoritative task artifacts within the openspec/changes directory to verify dependencies and task completion.
Boundary markers: The skill does not define specific delimiters for separating ingested file content from its own instructions, although it uses strict template formats for output.
Capability inventory: The skill has permissions to search the file system, compute file hashes, write new markdown files, and execute the openspec archival command.
Sanitization: The instructions mandate the normalization of informal requests (e.g., "abandoned" or "wontfix") into canonical dispositions, which serves as a validation step for ingested data.

change-disposition