The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (mdtodo files and git commit logs) and instructs the agent to analyze and summarize this content into a changelog entry. This creates a surface where malicious content in a todo item or commit message could attempt to influence the agent's behavior during the summarization step.
Ingestion points: changelog-context.sh reads tasks from mdtodo and commits from git log.
Boundary markers: None identified in the prompt to separate untrusted data from instructions.
Capability inventory: The skill has the ability to write to CHANGELOG.md, commit changes, and push to the remote repository using changelog-commit-push.sh.
Sanitization: No specific sanitization of the feature titles, task descriptions, or commit messages is performed before being presented to the agent for analysis.
[COMMAND_EXECUTION]: The skill executes local shell scripts and binaries (git, mdtodo) to gather repository context and perform git operations.
Evidence: changelog-context.sh runs mdtodo and git log. changelog-commit-push.sh executes git commit and push commands via a local library.
[DATA_EXPOSURE]: The script changelog-context.sh contains a load_dotenv function that parses and exports variables from a .env file in the repository root into the script's environment.
Evidence: load_dotenv "$ROOT_DIR/.env" in scripts/changelog-context.sh.

pflow-changelog