LunarCommand
OrganizationClaude Code skills for prod-local agent development: query real traces and logs from your local stack, plan before you build, and review deeper than a diff-scoped bot.
Categories
Indexed Skills (6)
hyperdx
Query and read logs and traces from HyperDX using Lucene syntax via the bundled hdx_query.sh CLI. Use this skill whenever the user asks to search logs, fetch errors, debug a service, investigate incidents, check log output, inspect traces, or look up anything in HyperDX — even if they don't say "HyperDX" explicitly. Trigger on phrases like "check the logs", "show me errors from", "what's happening with service X", "look up logs for", "any 500s recently", "show me the traces", or "what does HyperDX say about". Always use this skill when log querying or incident investigation is involved. IMPORTANT: always query logs by running the bundled hdx_query.sh script — never attempt to query HyperDX via curl, the API directly, or any other method.
langfuse
Interact with Langfuse and access its documentation. Use when needing to (1) query or modify Langfuse data programmatically — traces, prompts, scores, sessions, observations, (2) look up Langfuse documentation, concepts, integration guides, or SDK usage, or (3) understand how any Langfuse feature works. Works against self-hosted and cloud Langfuse, on both the legacy v1 REST API (server major <= 3) and the v2/v3 read APIs (server major >= 4).
pr-review
Use this skill when asked to review PR comments, address CoPilot feedback, respond to pull request review threads, or resolve PR conversations. Triggers on phrases like "review the PR comments", "address CoPilot feedback", "fix the PR", "go through PR comments", or any mention of pull request review threads needing attention. Always use this skill when a PR number or URL is mentioned alongside a request to review or respond to comments.
adversarial-review
Adversarial, multi-lens code review that GENERATES findings (not triage). Use before opening or merging a PR, or whenever asked to "adversarially review", "try to break", "deeply review", or "self-review" a change. Assembles whole-system context plus the repo's operational invariants, runs specialized lenses that look for what breaks (not what the code does), verifies each finding by refutation before surfacing it, and reports ranked by severity. For a single file it runs the lenses inline; for PR-scale scope it escalates to the bundled multi-agent workflow so the lenses run as genuinely independent agents.
feature-planning
Use this skill whenever a user wants to plan or implement a new feature, whether they reference a requirements file or describe the feature directly in chat. Triggers include phrases like "I've created a requirements file", "review the reqs for", "new feature requirements", "requirements in /_reqs/", any mention of a .md requirements document, OR a prose description of a feature the user wants planned or built (e.g. "I want to build X that does Y"). Always use this skill when the user provides either a requirements filename or a written feature description and wants Claude to review, plan, and implement — even if they don't say "skill" or "plan" explicitly.
mutation-test
Proves a test actually checks something, by breaking the behaviour it claims to cover and confirming it goes red. Use when a test, fixture, assertion or guard has been reported as working and you want evidence rather than a green run, or to check a claim that some code is dead, unused or unreachable. Triggers on "mutation test", "mutation testing", "prove it fails", "are these tests real", "does this test actually assert anything", "is that assertion live", "did you verify it's not vacuous", "how do you know it's checking anything", or any claim that a test passes being offered as evidence the behaviour is correct. Also use it proactively before reporting harness or fixture work as done.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.