Free Tier Tools (9)

These 9 tools work without any API key. They provide core functionality for evaluating MidOS and basic knowledge retrieval.

Free Tier limits: 50 queries/day, 5 req/min. Knowledge results are truncated to 250 chars per chunk.

---

search_knowledge

Keyword search across the entire knowledge base.

Parameter	Type	Required	Default	Description
query	string	Yes	—	Search query (max 5,000 chars)
max_results	int	No	5	Number of results (1-100)

Returns: Matched chunks with titles, snippets, and relevance indicators.

Notes: Falls back to semantic search if no keyword matches found. Community tier returns truncated content (250 chars per chunk).

{
  "name": "search_knowledge",
  "arguments": {
    "query": "React Server Components best practices",
    "max_results": 10
  }
}

---

list_skills

Browse available skills with optional filtering by technology stack.

Parameter	Type	Required	Default	Description
filter	string	No	""	Text filter for skill names
stack	string	No	""	Tech stack filter, comma-separated (e.g., "python,react")

Returns: List of skills with names and descriptions. Frontmatter-aware ranking.

{
  "name": "list_skills",
  "arguments": {
    "stack": "python,fastapi"
  }
}

---

hive_status

Get knowledge base counts and system health metrics.

Parameter	Type	Required	Default	Description
(none)	—	—	—	—

Returns: { chunks_count, skills_count, eureka_count, truth_patches, vector_store_status }

{
  "name": "hive_status",
  "arguments": {}
}

---

project_status

Live MidOS system dashboard with comprehensive metrics.

Parameter	Type	Required	Default	Description
(none)	—	—	—	—

Returns: Dashboard markdown with KB stats, vector store health, research queue, tool guide, and quick-start info.

---

agent_handshake

Onboard a new agent with personalized tool recommendations. Call this first when connecting.

Parameter	Type	Required	Default	Description
model	string	No	""	Agent model name (e.g., "claude-sonnet-4-5-20250514")
context_window	int	No	0	Context window size in tokens
client	string	No	""	MCP client name (e.g., "cursor", "claude-desktop")
languages	string	No	""	Programming languages, comma-separated
frameworks	string	No	""	Frameworks in use, comma-separated
platform	string	No	""	OS/platform
project_goal	string	No	""	What the agent is working on (max 5,000 chars)

Returns: Context-aware tool catalog and setup configuration based on your specific environment.

{
  "name": "agent_handshake",
  "arguments": {
    "model": "claude-opus-4-20250514",
    "context_window": 200000,
    "client": "claude-code",
    "languages": "python,typescript",
    "frameworks": "fastapi,react",
    "project_goal": "Building a knowledge management system"
  }
}

---

where_was_i

Resume a previous session. Returns the last saved checkpoint with activity summary and next steps.

Parameter	Type	Required	Default	Description
client	string	No	""	Filter by client name
session_id	string	No	""	Specific session to resume

Returns: Last session activity, checkpoint data, and suggested next steps. Returns “no session found” if no prior sessions exist.

---

context_health

Analyze your context window utilization and get optimization advice.

Parameter	Type	Required	Default	Description
tokens_used	int	No	0	Tokens currently used
tokens_total	int	No	200000	Total context window size
active_files	int	No	0	Number of files in context

Returns: Zone assessment (GREEN < 30%, YELLOW 30-40%, RED > 40%) with specific optimization recommendations.

{
  "name": "context_health",
  "arguments": {
    "tokens_used": 85000,
    "tokens_total": 200000,
    "active_files": 12
  }
}

---

get_progress

Retrieve saved session progress (read-only).

Parameter	Type	Required	Default	Description
session_id	string	No	""	Session to retrieve
round_id	string	No	""	Specific round
latest	bool	No	true	Get most recent progress

Returns: Markdown-formatted progress checkpoint or “no progress found.”

---

agent_bootstrap

Deprecated — Use agent_handshake instead.

Returns a deprecation notice directing to the new agent_handshake tool.