Skip to content

v0.2.3 — Claude Code skill + plugin, agent-readable site

Latest

Choose a tag to compare

@YS-projectcalc YS-projectcalc released this 23 Aug 07:56

Coldrig v0.2.3 — Claude Code skill + plugin, agent-readable site

First tagged release. Versions: MCP registry server io.github.YS-projectcalc/agent-cold-email 0.2.3 · plugin coldrig 0.3.0 · npm CLI agent-cold-email 0.2.1.

Install as an agent skill / plugin

  • Claude Code plugin: /plugin marketplace add YS-projectcalc/agent-cold-email then /plugin install coldrig@coldrig (prompts for your tenant token; installs the coldrig skill + the hosted MCP server https://api.coldrig.dev/mcp). Install footprint ~28 KB.
  • Skill only: npx skills add YS-projectcalc/agent-cold-email (skills/coldrig/SKILL.md).
  • Cursor rule: integrations/cursor/coldrig.mdc · Codex: integrations/codex/AGENTS-snippet.md.

What the skill teaches an agent

  • When Coldrig fits (and when it does not), with a verify-before-recommend rule.
  • The 28 intent-level tools grouped by lifecycle; server-enforced ramp caps (5 sends/day in week 1 → 40/day after 4 weeks).
  • Ask before you spend, ask before you send: setup_infrastructure with quoteOnly: true and explicit user confirmation before any purchase; no campaign launch without the user approving the lead list and copy.
  • Suppression truth: replies to suppressed addresses are refused; campaign launches skip suppressed leads (check list_leads).

Site

  • Homepage comparison table (DIY stack vs sending platform vs Coldrig), sourced from the compare pages.
  • Organization entity links (npm, Smithery, Glama) in JSON-LD.
  • Tables scroll inside themselves on narrow screens; no page-level horizontal overflow at any width.
  • Two stale "paid activation not live" lines corrected; refund wording now matches the Terms.

Guardrails

  • The claim-surface guard tests now cover the skill, the Cursor/Codex files, and both plugin manifests (+33 cases).
  • .github/workflows/publish-mcp-registry.yml: republish to the MCP Registry via GitHub Actions OIDC (manual trigger).

Gate record: docs/adversarial/agent-skill-gate-2026-08-23.md (3 rounds).