AletaIndex is Needs Work to agents.
Discry independently scored how well an AI agent can discover and understand the AletaIndex API from what’s public — not whether it’s usable. Below: every signal we checked, what’s costing the score, and what to change.
SCORED UNDER RUBRIC 1.2 · A full re-launch under Discry Score 2.5 — a new behavioral instrument, not comparable to these scores — is in progress.
Discovery
45% of score · 36/100Comprehension
55% of score · 91/100What we found
- An agent that starts at the brand domain hits a wall: aletaindex.com is a Webflow marketing site with no llms.txt, no sitemap.xml, no .well-known/mcp.json, and no on-site API documentation — the only calls to action are a demo link and a 'Get in touch' email, so an agent cannot self-serve from here.
- An agent that searches by name lands on the real surface immediately: the API and docs live at github.com/AletaIndex/aletaindex-fin-narratives (and the API host aletaindex-narrative.com), listed on Glama, Smithery, PyPI and awesome-mcp-servers — offering one-command MCP install (uvx narrative-intelligence-mcp) or one-click via directory.
- A publicly accessible OpenAPI 3.1.0 spec at aletaindex-narrative.com/openapi.json lets an agent auto-generate a typed client covering the full endpoint surface — a strong signal most scanned APIs lack.
- Comprehension is A-grade: endpoint descriptions are task-oriented ('Returns the full narrative hierarchy...'), examples use realistic tickers/dates across curl, Python and TypeScript, and the MCP quickstart plus 'Common Patterns' guides walk an agent end-to-end. Capability boundaries (tier limits, max 10 tickers/query, 50 holdings, credit model) are explicit.
- The main discovery weakness beyond the bare brand domain: cross-doc naming drift — the README's quick example uses narrative_id / sentiment_label / dominance_score while the canonical api-reference uses id / sentiment.label, and daily-topic dates appear as both event_date and time — which could trip an agent that trusts the README over the reference.
What to change
Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.
- 01Add an llms.txt (and llms-full.txt) to aletaindex.com that points to the GitHub docs, the OpenAPI spec, and the MCP install instructions — the single cheapest fix to lift discovery, since the good docs already exist elsewhere.
- 02Link the developer surface from aletaindex.com itself: add a 'Developers' / 'API' nav item pointing to the GitHub repo and aletaindex-narrative.com. Today the brand domain and the API/docs domain are disconnected, so an agent on aletaindex.com has no path to the docs.
- 03Publish a sitemap.xml and a real robots.txt on aletaindex.com (both are currently missing/empty), listing the key pages so crawlers and agents can enumerate the site.
- 04Fix cross-document field-name consistency so the README example matches the api-reference schema (id/title/sentiment.label) and daily-topic dates use one name (event_date vs time) — agents that copy from the README should hit the exact shape the API returns.
- 05Add an AGENTS.md to the repo root for coding-agent context (setup, auth, tool names, gotchas); it is a rising standard and is currently absent.
Execution coverage · INFORMATIONAL, UNSCORED
Whether an agent can actually complete a call and recover from errors is the deeper Audit layer — documented here, but not part of the Discry Score.
Auth is a single API key via the X-API-Key header, clearly documented in every guide and the OpenAPI spec. Error format is documented as an HTTP status table (401/403/422/429/500) and the OpenAPI spec defines a machine-parseable 422 validation schema. Rate limits are documented as a credit-based tier model (429 on exceed) though no rate-limit header names are given. Pagination is documented (paginate/article_limit/article_offset). Idempotency is not mentioned.