The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from pull requests and branches, which could theoretically contain malicious instructions embedded in code or comments.
Ingestion points: Git diff outputs and application source files located in src/frontend/** (SKILL.md).
Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore embedded instructions within the ingested code.
Capability inventory: The skill utilizes shell commands (git, npx, uv), can modify files on the filesystem to apply fixes, and performs local network operations for scanning (http://localhost:3000).
Sanitization: No sanitization or validation of the ingested code content is specified before processing.
[COMMAND_EXECUTION]: The skill executes several shell-based development tools as part of its workflow.
Evidence: Uses git diff, git ls-files, and gh pr diff to identify changes. It also runs npx playwright test, npx jest, and uv run to execute accessibility scans and unit tests.
[EXTERNAL_DOWNLOADS]: The skill references and downloads dependencies from well-known services.
Evidence: Suggests installing Playwright and Chromium using uv run --with playwright playwright install chromium. Playwright is a well-known browser automation library from Microsoft.

ibm-a11y-pr-remediation