Git AI

Using the /ask Skill

Use past prompts as context to help engineers and agents understand and extend AI-generated code.

The /ask skill surfaces the intent, requirements, and decisions behind AI-generated code providing much richer context to agents and engineers than they can get from reading the source code.

Using /ask

Git AI installs the /ask skill to ~/.agents/skills/ and ~/.claude/skills/ at setup time. The skill is available in Cursor, Claude Code, Copilot, Codex, and other supported agents.

/ask Why didn't we use the SDK here?

With and without /ask

With agent session (/ask)Without Git AI
When Aidan was building telemetry, he instructed the agent not to block the CLI exit while flushing telemetry. Instead of using the Sentry SDK directly, the agent devised a pattern that writes events locally first via append_envelope(), then flushes them in the background via a detached subprocess. This keeps the hot path fast and ships telemetry asynchronously.src/commands/flush_logs.rs is a 5-line wrapper that delegates to src/observability/flush.rs (~700 lines). The commands/ layer handles CLI dispatch; observability/ handles Sentry, PostHog, metrics upload, and log processing. Parallel modules like flush_cas, flush_logs, flush_metrics_db follow the same thin-dispatch pattern.

Letting agents use the /ask skill

Agents make fewer mistakes and produce more maintainable code when they understand the requirements and decisions behind existing code. Add the /ask skill to agent instructions so agents query intent before planning changes.

Add to AGENTS.md (or CLAUDE.md, .cursorrules, etc.):

AGENTS.md
- In plan mode, always use the /ask skill to read the code and the original prompts that generated it. Understanding intent will help you write a better plan.

Agent session storage

Agent Sessions can be stored locally, in the Git AI Cloud, or in a self-hosted store — keeping repositories free of sensitive information, and preserving control over data.

Git AI Blame VSCode gutter showing agent session prompts

For individual users, Git AI stores agent session data locally in a SQLite database. Sessions are only visible to the author on the machine that created them.

Sharing with a team

Sharing the intent and rationale behind decisions in the codebase is essential for any team managing large AI-generated codebases.

If you're interested in sharing prompts amongst your teammates check out Git AI For TeamsSign up for early access. Large organizations with strict security requirements can self-host.