Ops Tier Tools (5)

These 5 tools require a midos_sk_ops_* or midos_sk_mod_* API key. They unlock the most valuable knowledge layers and advanced search capabilities.

Ops Tier ($49.90/mo): 2,000 requests/day, 60 req/min. Includes 30 agent-driven research tasks/month, AOTC + Security + Infra skill packs, pool signals, and agent coordination.

---

get_eureka

Get a breakthrough improvement document (validated discovery with ROI data).

Parameter	Type	Required	Default	Description
name	string	Yes	—	EUREKA name (case-insensitive, fuzzy matched)

Returns: Full EUREKA document in markdown (up to 10,000 chars). Each EUREKA includes: Problem → Solution → Test Results → ROI.

{
  "name": "get_eureka",
  "arguments": {
    "name": "response cache"
  }
}

---

get_truth

Get a verified truth patch (correction to common misconceptions).

Parameter	Type	Required	Default	Description
name	string	Yes	—	Truth patch name (case-insensitive, fuzzy matched)

Returns: Full truth patch document in markdown. Format: Problem → Evidence → Correction.

{
  "name": "get_truth",
  "arguments": {
    "name": "preflight dedup protocol"
  }
}

---

semantic_search

LanceDB vector search with Gemini embeddings and optional reranking.

Parameter	Type	Required	Default	Description
query	string	Yes	—	Search query (max 5,000 chars)
top_k	int	No	5	Number of results (1-100)
stack	string	No	""	Filter by tech stack
search_mode	string	No	"hybrid"	"vector", "keyword", or "hybrid"
rerank	bool	No	true	Apply relevance reranking

Returns: Ranked results with normalized relevance scores (0.0-1.0).

{
  "name": "semantic_search",
  "arguments": {
    "query": "How to implement the repository pattern in Python with async SQLAlchemy",
    "top_k": 10,
    "search_mode": "hybrid",
    "rerank": true
  }
}

Search modes:

• "vector": Pure semantic similarity using Gemini embeddings (3072-d)
• "keyword": Grep-based keyword matching
• "hybrid" (default): Combines both for best results

---

hybrid_search

Grep-first search with automatic semantic fallback.

Parameter	Type	Required	Default	Description
query	string	Yes	—	Search query (max 5,000 chars)
top_k	int	No	10	Number of results (1-100)
rerank	bool	No	true	Apply relevance reranking

Returns: Keyword results first; if zero matches, falls back to vector search automatically.

When to use: When you want fast keyword matching but don’t want to miss results if the exact terms aren’t in the knowledge base.

---

episodic_search

Search past agent sessions and experiences.

Parameter	Type	Required	Default	Description
query	string	Yes	—	What to search for (max 5,000 chars)
limit	int	No	5	Number of results (1-100)

Returns: Matching episodic memories from past agent runs. Useful for recalling how similar problems were solved before.

{
  "name": "episodic_search",
  "arguments": {
    "query": "authentication middleware debugging",
    "limit": 5
  }
}