Review request: v2.1.0 Decixa resolve() integration

[TKCollective]

Hey @koki-socialgist (and anyone else who wants eyes on it) — v2.1.0 of agentoracle-mcp is already live on npm with the Decixa resolve() integration you sketched out in the gist. Opening this as a review request since I shipped straight to main (small additive change, no breaking changes) but you asked for eyes on it.

What changed

Commits:

• 9b61d68 — feat(v2.1.0): add resolve tool — Decixa discovery with local fallback
• 5685327 — docs: credit Decixa + Glama as discovery sources
• 32f0ee8 — chore: add LICENSE (MIT) and glama.json manifest

Tagged: v2.1.0
Published: agentoracle-mcp@2.1.0 on npm

Implementation notes

• resolve tool accepts capability (enum: search/extract/transform/analyze/generate/modify/communicate/transact/store), intent (freeform), and optional budget / latency constraints. Matches your gist spec.
• Primary path: POST https://api.decixa.ai/api/agent/resolve with the full body forwarded through.
• Fallback: local AgentOracle registry for analyze and search, mirrored to your response shape so downstream consumers don't have to branch on the source.
• Every response includes discovery_source: "decixa" | "local_fallback" so agents know what served them.
• Non-OK and exception paths both trigger fallback. Logs to stderr (MCP stdio-safe).
• D-059 compatibility: the tool doesn't assume capability is always present in the response shape; it just forwards Decixa's body through.

Verification

Fresh npx agentoracle-mcp@2.1.0 install + 3 manual tools/call tests:

Test	capability	intent	Source	Result
1	analyze	"verify a factual claim"	decixa	Otto AI Agent Swarm recommended + alternatives
2	search	"live web research", budget=0.05	decixa	AgenticFi API — trending recommended + alternatives
3	extract	"deep research analysis", latency=medium	decixa	Swarm recommended + alternatives

All three returned valid Decixa responses with discovery_source: "decixa".

README

Updated to credit Decixa + Glama as discovery sources, with direct links to all three Decixa listings. Matches the reciprocal attribution pattern.

Follow-ups

• When free-endpoint trust scoring lands on your side, happy to test /evaluate as the first case (flagged in Discord).
• If you want the tool description wording tweaked for consistency with how you describe resolve() publicly, just leave a comment here and I'll ship a v2.1.1.

Appreciate the eyes 🙏