Platform Management

ID	platform_management
Category	Platform
Features	None
Dependencies	session_file_system

Tools to manage Everruns entities programmatically. Read, create, update, and delete harnesses, agents, and sessions — and interact with sessions by sending messages.

Platform Documentation

This capability mounts the Everruns platform documentation at /workspace/docs as a virtual read-only filesystem. The documentation is embedded at compile time from the repository docs/ directory (markdown files only) and served from memory — no database writes per session.

Agents can browse and search the docs using standard file tools (read_file, list_directory, grep) and virtual bash commands (cat, ls, grep).

Key sections:

• /workspace/docs/getting-started/ — Introduction, concepts, architecture
• /workspace/docs/features/ — SDK, CLI, UI, events, harnesses, capabilities
• /workspace/docs/capabilities/ — Per-capability reference
• /workspace/docs/integrations/ — External integrations (Slack, Daytona, etc.)
• /workspace/docs/advanced/ — Budgets, compaction, embedding, network access
• /workspace/docs/sre/ — Environment variables, runbooks

Tools

read_capabilities

Discover available capabilities (built-in, MCP servers, skills). Returns a single capability when id is given, otherwise a filtered list.

Parameter	Type	Required	Description
id	string	no	Capability ID to get a single capability
search	string	no	Case-insensitive filter by name, description, category, or ID

read_harnesses

Read harnesses by ID or list all. Returns full detail (incl. system_prompt) when id is given.

Parameter	Type	Required	Description
id	string	no	Harness ID to get a single harness

manage_harnesses

Harness mutations: create, update, delete, copy.

Parameter	Type	Required	Description
operation	enum	yes	create, update, delete, copy
harness_id	string	conditional	Required for update, delete, copy
name	string	conditional	Required for create
system_prompt	string	conditional	Required for create
description	string	no	Optional description
capabilities	string[]	no	Capabilities to enable

read_agents

Read agents by ID or list all. Returns full detail (incl. system_prompt) when id is given.

Parameter	Type	Required	Description
id	string	no	Agent ID to get a single agent

manage_agents

Agent mutations: create, update, delete.

Parameter	Type	Required	Description
operation	enum	yes	create, update, delete
agent_id	string	conditional	Required for update, delete
name	string	conditional	Required for create
system_prompt	string	conditional	Required for create
description	string	no	Optional description
capabilities	string[]	no	Capabilities to enable

read_sessions

Read sessions by ID or list/filter.

Parameter	Type	Required	Description
id	string	no	Session ID to get a single session
agent_id	string	no	Optional filter by agent
limit	integer	no	Max results for list (default: 20)

manage_sessions

Session mutations: create, delete.

Parameter	Type	Required	Description
operation	enum	yes	create, delete
session_id	string	conditional	Required for delete
harness_id	string	no	Harness ID for the session (defaults to Generic)
agent_id	string	no	Optional for create
title	string	no	Optional for create

session_send_message

Send a user message to a session, triggering a turn.

Parameter	Type	Required	Description
session_id	string	yes	Target session ID
content	string	yes	Message content

session_read_messages

Read messages from a session.

Parameter	Type	Required	Description
session_id	string	yes	Target session ID
limit	integer	no	Max messages (default: 10)

session_read_response

Wait for session to finish processing and return the response.

Parameter	Type	Required	Description
session_id	string	yes	Target session ID
timeout_secs	integer	no	Timeout (default: 120). Set to 0 to check status without waiting.

Use Cases

• Meta-agents — an agent that creates and configures other agents
• Orchestration — spawn sessions across multiple agents and collect results
• Self-configuration — agent inspects available capabilities before setting up new agents
• Automated testing — programmatically create sessions and verify agent responses

Example

Agent creates a specialized sub-agent and interacts with it:

Agent:
  → read_capabilities({ search: "file" })
  ← { capabilities: [{ id: "session_file_system", name: "File System", ... }], count: 1 }

  → manage_agents({
      operation: "create",
      name: "Code Reviewer",
      system_prompt: "You review code for bugs and style issues.",
      capabilities: ["session_file_system", "virtual_bash"]
    })
  ← { "agent_id": "agt_01xyz...", "ui_link": "/agents/agt_01xyz..." }

  → manage_sessions({
      operation: "create",
      harness_id: "hrs_01abc...",
      agent_id: "agt_01xyz..."
    })
  ← { "session_id": "ses_01def...", "ui_link": "/sessions/ses_01def.../chat" }

  → session_send_message({
      session_id: "ses_01def...",
      content: "Review this Python function: def add(a, b): return a + b"
    })

Notes

• All tool results include ui_link fields for navigating to the web interface
• read_capabilities searches across built-in, MCP server, and skill capabilities
• Session interaction follows the same turn lifecycle as the UI

See Also

• Concepts — Harness, Agent, Session model
• Agent Skills — skill discovery
• API Reference — full REST API
• Capabilities Overview