The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its reliance on external project data to drive decision-making loops.
Ingestion points: The agent is instructed in references/prompt-templates.md and scripts/render_crank_prompt.py to ingest untrusted data from git log and gh pr list.
Boundary markers: There are no delimiters or specific instructions provided to the agent to treat this external data as untrusted or to ignore any embedded instructions within it.
Capability inventory: The skill has high autonomy ("execute-forward" policy) and the ability to execute shell commands (git, gh) and perform file system modifications through the worker agent's actions.
Sanitization: No sanitization or validation of the content retrieved from git or GitHub is performed before it is processed by the agent.
[COMMAND_EXECUTION]: The skill facilitates the execution of local development tools to automate status gathering.
It utilizes shell commands including git rev-parse, git log, git status, git stash, and gh pr list to satisfy the "Finish Contract" metrics defined in SKILL.md.

project-crank-engine