The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from git conflict markers which may contain malicious instructions designed to influence the agent's behavior during the resolution process.
Ingestion points: The agent is instructed in SKILL.md to "Inspect the replayed commit, base, and both sides" and uses the output of scripts/analyze-conflicts.sh which displays conflict sections directly from repository files.
Boundary markers: Absent. There are no explicit instructions or delimiters used to isolate untrusted conflict data or warn the agent to ignore instructions embedded within the rebased code.
Capability inventory: The skill possesses the capability to modify files during conflict resolution, stage changes (git add), continue the rebase process (git rebase --continue), and execute project-specific scripts (npm test, npm run lint).
Sanitization: Absent. Content from replayed commits is presented to the agent without filtering or validation.
[COMMAND_EXECUTION]: The skill relies on executing shell scripts and Git commands to perform its core functions.
Evidence: The skill includes and utilizes scripts/pre-rebase-backup.sh, scripts/analyze-conflicts.sh, and scripts/validate-merge.sh to automate repository management tasks.
Context: These operations are consistent with the skill's purpose as a Git rebase utility and are documented for the user's awareness.
[EXTERNAL_DOWNLOADS]: The skill provides references to official documentation for conflict resolution strategies.
Evidence: Documentation links to git-scm.com are included in references/automation.md and references/strategies.md.
Context: These links point to a well-known service providing authoritative documentation for the Git version control system.

git-rebase