The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard Git commands to inspect repository state, manage the index, and continue interrupted operations. It also references local scripts and tools for managing artifact storage settings.
Evidence: Commands such as git rev-parse, git status, git diff, and git merge --continue are used in SKILL.md. The skill also references a local script scripts/durable_artifacts.py and helper tools like choose and resolve in the reference files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from files containing Git conflict markers and metadata, which are potentially untrusted sources controlled by external contributors.
Ingestion points: The agent reads file contents during conflict resolution, as well as Git metadata from git status, git diff, and "primary sources" like commit messages and PR descriptions.
Boundary markers: The skill requires the agent to map each conflict hunk to an "intent basis" from verified sources and follow a strict "Contract" to prevent guessing or following unauthorized instructions.
Capability inventory: The skill has the ability to modify project files and advance Git repository state through staging and continuing operations.
Sanitization: The instructions explicitly require the agent to verify that conflict markers are authentic and not merely formatting artifacts (e.g., Markdown separators), reducing the risk of accidental obedience to embedded text.

tk-merge-conflict