Stellar Experimental
Official@stellar-experimental
Experiments at the frontier of the Stellar Development Foundation. (Repos not covered by SDF Bug Bounty)
Agent Skills by Stellar Experimental
Showing 8 vetted skills indexed across 1 GitHub repositories.
stellar-experimentalcloudflare-observability-review
Investigate production Cloudflare Workers logs, traces, and Ray IDs for MCP request forensics.
stellar-experimentalgolden-truth
Verifies and edits eval golden answers through multi-source triangulation and provenance tracking.
stellar-experimentalimprovements-pipeline
Manage upstream service-improvement findings through filing, verification, and resolution workflows.
stellar-experimentalaudit-reviewability
Audit and repair code, comments, documentation, and tests for reviewability debt.
stellar-experimentalrun-evals
Orchestrates full eval rounds on the stellar-raven MCP server from CLI agents.
stellar-experimentaltruth-maintenance
Coordinates multi-lane maintenance rounds across drift, evals, goldens, and upstream issue tracking.
stellar-experimentalretrieval-system-audit
Audit and improve MCP server retrieval across live service operations and skills.
stellar-experimentallive-drift-resolution
Resolve live service drift issues by regenerating catalog artifacts and classifying upstream changes.
Frequently Asked Questions About Stellar Experimental
FAQPage SchemaWhat tasks can I perform with Stellar Experimental's skills?▼
You can run eval rounds on stellar-raven-codemode, audit retrieval across every exposed service operation, edit the golden Q&A corpus with provenance, investigate Cloudflare Workers production telemetry, clean AI-generated code for reviewability, and resolve live service drift against upstream Stellar surfaces.
Who are these skills designed for?▼
They target engineers and coding agents (Codex, Claude Code, or other CLI agents) maintaining the stellar-raven-codemode MCP service — specifically those running evals, curating golden truth, auditing search ranking, and handling upstream findings, issues, and PR follow-up.
How do eval rounds work in practice?▼
An orchestrating agent picks instruments (routing gate, QA headline, agentic, plan, live-data), spawns answering and judge agents, records results, reviews every answer and verdict, triages failures to root cause, and files evidence-backed upstream findings in improvements/ — findings, not scores, are the primary artifact.
What is live drift resolution and when is it triggered?▼
A daily CI diffs the committed catalog against live upstream surfaces (Lumenloop, Stellar Light/Scout, Stellar Docs) and files a drift issue. The skill guides regenerating artifacts, classifying drift as policy change, routing-baseline change, or mechanical bump, dual-verifying, deploying when authorized, and closing.
What prerequisites or dependencies do these skills require?▼
They require access to the stellar-raven-codemode repository and its live deployment on Cloudflare Workers, plus the improvements/ intake pipeline, golden corpus files, and upstream Stellar services. Note these experimental repos are not covered by the SDF Bug Bounty.