Skip to content

Climb from Bronze to Silver

The toolkit’s validators turn the Silver climb into a literal to-do list you can read off tier-report.

node scripts/tier-report.mjs --json

Look at blocked.convergent - each entry is a S<n>: <message> reminder of what stands between this plugin and Silver.

For each S<n> entry, see the per-rule reference (../reference/silver-checks.md) for the exact remediation. Most fixes are small edits to library.json:

  • S1 missing agent-targets: add "agent-targets": ["claude", "codex"].
  • S2 missing prefix: add "prefix": "<short>-".
  • S3 components index drift: regenerate the manifest and align components.skills with what is on disk.

Re-run tier-report --json after each fix; the blocked.convergent list shrinks.

When blocked.convergent is empty (and S4/S5 are satisfied or do not apply), update library.json:

"tier": "convergent"

From that point node scripts/check.mjs starts gating on Convergent errors too (the declared-tier ceiling rises). If anything still flags, the climb is not actually done - keep fixing until the gate exits 0 at the new tier.

S6 (per-target format presence) is added in Phase 3B alongside the Codex-emission generators. Until then, Silver means the library.json and chain/workflow integrity are correct; the per-target emission proof arrives with the emitters.