Status: Planned / not yet implemented. Deferred from the current VSCode parity pass because it does not exist on mobile either (net-new feature for both surfaces).
Summary
Skills are already tied to GitHub stars through verified-work, but only agents surface that today (it drives agent tier). The skill view should also list the repos that used a skill and show a summed star count, and let the market sort skills by stars. This reuses the existing indexer — no on-chain change or new data source needed.
What already exists (verified — agentnet-nft-indexer)
work_link table links, per skill: skill_mint, repo_url, wallet, stars (GitHub stars, periodically refreshed).
GET /items/:mint/repos (reposForSkill()) already returns a skill's repos ordered by stars DESC.
starsByWallet() sums stars per wallet (deduped by repo) — this is the agent tier/reputation source today.
- A skill record (
IndexedItem) only carries supply as a ranking signal — no stars on the skill itself yet.
So the per-skill data is already indexed; only a skill-level aggregation + UI is missing.
Proposed (low backend effort)
- Indexer: add
starsBySkill() — mirror starsByWallet() but GROUP BY skill_mint (dedupe by repo). Expose via an endpoint or fold into the item payload.
- Skill detail / card: show the repos that used this skill (like the agent's verified-repos list) + a summed star figure.
- Market: add a sort-by-stars option alongside the existing
supply sort.
Scope
indexer query + contract field + UI (skill card stars, repo list, sort). Both mobile + VSCode. Shares the same source (work_link) as agent-tier stars, so the two stay consistent.
Relation
Sibling to agent tier (verified-work stars). This is the skill-side view of the same reputation graph.
Status: Planned / not yet implemented. Deferred from the current VSCode parity pass because it does not exist on mobile either (net-new feature for both surfaces).
Summary
Skills are already tied to GitHub stars through verified-work, but only agents surface that today (it drives agent tier). The skill view should also list the repos that used a skill and show a summed star count, and let the market sort skills by stars. This reuses the existing indexer — no on-chain change or new data source needed.
What already exists (verified —
agentnet-nft-indexer)work_linktable links, per skill:skill_mint,repo_url,wallet,stars(GitHub stars, periodically refreshed).GET /items/:mint/repos(reposForSkill()) already returns a skill's repos ordered bystars DESC.starsByWallet()sums stars per wallet (deduped by repo) — this is the agent tier/reputation source today.IndexedItem) only carriessupplyas a ranking signal — no stars on the skill itself yet.So the per-skill data is already indexed; only a skill-level aggregation + UI is missing.
Proposed (low backend effort)
starsBySkill()— mirrorstarsByWallet()butGROUP BY skill_mint(dedupe by repo). Expose via an endpoint or fold into the item payload.supplysort.Scope
indexer query + contract field + UI (skill card stars, repo list, sort). Both mobile + VSCode. Shares the same source (
work_link) as agent-tier stars, so the two stay consistent.Relation
Sibling to agent tier (verified-work stars). This is the skill-side view of the same reputation graph.