Cisco is Needs Work to agents.
Discry independently scored how well an AI agent can discover and understand the Cisco 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 · 62/100Comprehension
55% of score · 71/100What we found
- Cisco has no llms.txt, llms-full.txt, or .well-known/mcp.json — agents relying on these standard discovery signals will not find Cisco APIs automatically
- Multiple AGENTS.md files exist across CiscoDevNet repos (Local-RAG-AI-Assistant, cisco-secure-access-python-sdk) showing early adoption of agent-first practices
- The Meraki API has excellent rate limit documentation with recovery guidance, burst allowances, and best practices — a model for agent-friendly operational docs
- Strong MCP ecosystem presence with 8+ community servers on Glama and PulseMCP covering Meraki, Webex, ISE, and Catalyst Center
- OpenAPI spec is publicly maintained at github.com/meraki/openapi with regular releases (v1.70.0 as of May 2026)
What to change
Prioritized by impact on discoverability. You (or your docs platform) deploy these — Discry never touches your API.
- 01Add llms.txt at developer.cisco.com covering the major API families (Meraki, Webex, Catalyst Center) with structured descriptions and links to quickstart guides
- 02Add llms-full.txt with comprehensive markdown covering core API operations across Cisco's developer platform
- 03Create .well-known/mcp.json referencing the community MCP servers or publish an official Cisco MCP server
- 04Move sitemap.xml to the standard /sitemap.xml path — currently only referenced via robots.txt at non-standard /data/sitemaps/ paths
- 05Add error recovery guidance beyond HTTP status codes — the Meraki error page lacks actionable fix steps for most error scenarios
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.
Meraki API documents API key and OAuth 2.0 auth with clear comparison table. Error responses are JSON with errors array. Rate limits well-documented at 10 req/s per org with burst allowance. RFC5988 Link header pagination with startingAfter/endingBefore tokens. No idempotency key support documented.