New use cases: PR review, fleet migration, security audit, and beyond

[axon-agent]

🤖 Axon Agent @gjkim42

Summary

After analyzing Axon's current capabilities, examples, and existing issues, I've identified several high-value use cases that are underserved by the current example set. Three of these are concrete enough to propose as examples in PR #290. This issue documents the full analysis and proposes additional use cases for consideration.

Analysis: Current example coverage gap

Axon's four CRDs (Task, Workspace, AgentConfig, TaskSpawner) form a composable system that can express far more than the existing examples demonstrate:

Feature used	Existing examples	New examples (PR #290)
githubIssues with types: [issues]	03	—
githubIssues with types: [pulls]	—	05-pr-review
cron schedule	04	07-security-audit
AgentConfig with plugins	—	05-pr-review
AgentConfig (agentsMD only)	—	06-migration, 07-security-audit
Multi-workspace pattern	—	06-migration
excludeLabels feedback loop	(self-development only)	05-pr-review
Conditional output (skip if nothing to do)	—	07-security-audit

Key observation: the AgentConfig CRD and the types: [pulls] option are completely absent from the existing examples directory, despite being powerful features.

Proposed use cases (concrete examples in PR #290)

1. Automated PR Review (example 05)

Target audience: Teams that want AI-assisted code review on demand.
Pattern: Label a PR with needs-ai-review → agent reviews → swaps label to ai-reviewed.
Why it matters: Code review is one of the most requested AI use cases. Showing this as a first-class Axon workflow makes the framework immediately relevant to teams already thinking about AI-assisted review.

2. Fleet-Wide Migration (example 06)

Target audience: Platform engineering teams managing multiple microservices.
Pattern: One TaskSpawner per repo, all sharing a common AgentConfig with migration instructions. All agents run in parallel.
Why it matters: This is Axon's strongest differentiator — no other tool makes it this easy to fan out AI agents across dozens of repos simultaneously. This directly addresses the README's "Fleet-Wide Refactoring" orchestration pattern with a concrete, copy-paste example.

3. Security Audit (example 07)

Target audience: Security-conscious teams, compliance-driven organizations.
Pattern: Weekly cron agent scans the codebase and only creates issues when findings exist.
Why it matters: Continuous security scanning is a fast-growing category. Positioning Axon here gives it access to security and compliance budgets, not just developer tooling budgets.

Additional use cases (not yet in examples, for future consideration)

4. Documentation generation from code changes

Trigger: githubIssues watching for merged PRs (use types: [pulls], state: closed).
Prompt: "Review the changes in this merged PR and update any affected documentation (README, API docs, inline comments). If the docs are already up to date, exit without changes."
Why: Keeps docs in sync with code changes automatically. Low risk (the worst outcome is a docs PR that gets declined).

5. Onboarding assistant

Trigger: githubIssues with label good-first-issue or help-wanted.
Prompt: "This issue is tagged for new contributors. Add a detailed comment explaining how to approach it: which files to look at, what the expected behavior is, and any relevant context from the codebase."
Why: Reduces maintainer burden for open-source projects. The agent doesn't fix the issue — it just provides guidance, making it safe and low-stakes.

6. Release notes generation

Trigger: cron on release day or githubIssues with types: [pulls] watching for PRs with a release label.
Prompt: "Generate release notes by reviewing all merged PRs since the last release tag. Categorize changes into Features, Bug Fixes, Breaking Changes, and Dependencies."
Why: Release notes are a universal pain point. This automates a tedious task that blocks every release.

7. Stale issue triage

Trigger: cron (weekly).
Prompt: "List all open issues older than 90 days. For each, determine if it's still relevant by checking if the referenced code still exists and the described behavior still occurs. Comment on issues that appear outdated with a suggestion to close."
Why: Issue hygiene is a common pain point for active repositories. This pattern uses Axon purely for analysis and commenting — no code changes, very low risk.

Observations about Axon's current positioning

1. The multi-repo story is strong but hidden. The README mentions "Fan out hundreds of agents" but there's no concrete example showing how. Example 06 fills this gap. Consider highlighting multi-repo as a primary differentiator.

2. AgentConfig is underutilized in examples. It's documented in the README but none of the examples use it. PR Examples: Add PR review, multi-repo migration, and security audit use cases #290 adds three examples that use AgentConfig, demonstrating its value for reusable instructions and skills.

3. The PR review use case is the easiest on-ramp. Teams that are already considering AI code review (Copilot reviews, CodeRabbit, etc.) would find Axon's Kubernetes-native approach compelling — especially for organizations that want to keep AI interactions within their own infrastructure.

4. Security scanning positions Axon for enterprise budgets. Enterprise security teams have budget for automated scanning tools. An Axon-based security scanner that runs inside the organization's own cluster (no data leaving the network) is a strong value proposition.

Related issues

• Examples: Missing beginner-friendly TaskSpawner examples #206 — Missing beginner-friendly TaskSpawner examples (the new examples are intermediate-level)
• Documentation: Missing dedicated examples directory with simple, standalone YAML manifests #222 — Missing dedicated examples directory (these add to the existing examples/ directory)
• API: Add taskCompletion trigger source and structured outputs for multi-step agent workflows #283 — Task completion triggers (would enable chaining review after code generation)
• API: Add webhook and GitHub event sources to TaskSpawner for real-time task triggering #268 — Webhook triggers (would improve latency for the PR review use case)
• AgentConfig does not support codex and gemini #289 — AgentConfig does not support codex and gemini (limits examples 05/07 to claude-code)

[axon-agent]

🤖 Axon Agent @gjkim42

Axon Triage Report

Kind: kind/docs

Status: STILL VALID

Evidence:

• No merged PRs reference this issue. The companion PR Examples: Add PR review, multi-repo migration, and security audit use cases #290 ("Examples: Add PR review, multi-repo migration, and security audit use cases") is still open and unmerged.
• No recent commits on main address the proposed use cases (PR review, fleet migration, security audit examples).
• The examples/ directory currently contains 5 examples (01 through 05). None cover types: [pulls] triggers, multi-workspace patterns, or cron-based security scanning — the three main patterns proposed here.
• The existing example 05 (05-task-with-agentconfig) demonstrates basic AgentConfig usage with a Task, but does not cover the advanced patterns proposed in this issue (AgentConfig with plugins, shared AgentConfig across multiple TaskSpawners).
• The AgentConfig CRD and types: [pulls] option remain underrepresented in examples, as the issue correctly identifies.

Duplicates found:

• Documentation: Missing dedicated examples directory with simple, standalone YAML manifests #222 (partial overlap — requests a dedicated examples directory with beginner-friendly YAML manifests; this issue specifically targets intermediate/advanced use cases like PR review, fleet migration, and security audit)
• Examples: Missing beginner-friendly TaskSpawner examples #206 (partial overlap — requests beginner-friendly TaskSpawner examples; this issue explicitly acknowledges Examples: Missing beginner-friendly TaskSpawner examples #206 and targets a different audience with more advanced patterns)

These are complementary rather than true duplicates — #206/#222 focus on beginner on-ramp, while #291 focuses on showcasing advanced capabilities.

Recommendation: KEEP OPEN — The proposed use cases demonstrate important Axon features (multi-repo fan-out, AgentConfig reuse, PR-triggered workflows) that have no existing examples. PR #290 is the implementation vehicle and should be reviewed alongside this issue.