Tool Invocation Reference

All 30 Manasvi built-in tools. For each tool: ID, a natural language prompt you can send to the agent, and the expected output shape.

See also: Built-in Tools Overview · Default Tool Sets · Demo Flows

Index

#	Tool ID	Category
1	tool.local-file-read	Filesystem (read)
2	tool.http-fetch	Web
3	tool.web-search	Web
4	tool.memory-note-write	Memory
5	tool.shell-command	Runtime
6	tool.approval-request	Governance
7	tool.exec	Runtime
8	tool.process	Runtime
9	tool.code-execution	Runtime
10	tool.bash	Runtime
11	tool.fs-write-file	Filesystem (write)
12	tool.fs-append-file	Filesystem (write)
13	tool.fs-apply-patch	Filesystem (write)
14	tool.fs-rename-file	Filesystem (write)
15	tool.sessions-list	Sessions
16	tool.sessions-history	Sessions
17	tool.session-status	Sessions
18	tool.sessions-send	Sessions
19	tool.sessions-yield	Sessions
20	tool.sessions-spawn	Sessions
21	tool.subagents	Sessions
22	tool.memory-search	Memory
23	tool.memory-get	Memory
24	tool.x-search	Web
25	tool.browser	UI
26	tool.canvas	UI
27	tool.cron	Automation
28	tool.gateway	Automation
29	tool.message	Messaging
30	tool.nodes	Nodes
31	tool.agents-list	Agents

Core Tools (B4)

tool.local-file-read

Category: Filesystem (read) | Approval: not required | Mutability: read_only

Prompt to invoke:

"Read the file at ./src/config.ts and summarise what it configures."

Expected output:

{
  "path": "./src/config.ts",
  "content": "import { z } from 'zod';\n\nexport const configSchema = ...",
  "encoding": "utf8",
  "bytes": 1420,
  "truncated": false,
  "provenance": {
    "source": "filesystem",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

tool.http-fetch

Category: Web | Approval: not required | Mutability: read_only

Prompt to invoke:

"Fetch https://httpbin.org/json and tell me what's in the response."

Expected output:

{
  "url": "https://httpbin.org/json",
  "status": 200,
  "contentType": "application/json",
  "body": "{ \"slideshow\": { \"title\": \"Sample Slide Show\" } }",
  "truncated": false,
  "provenance": {
    "source": "remote-http",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

tool.web-search

Category: Web | Approval: not required | Mutability: read_only

Prompt to invoke:

"Search for recent news about TypeScript 5.5 and summarise the top results."

Expected output:

{
  "query": "TypeScript 5.5 news",
  "results": [
    {
      "title": "TypeScript 5.5 Release Notes",
      "url": "https://devblogs.microsoft.com/typescript/...",
      "snippet": "TypeScript 5.5 introduces inferred type predicates..."
    },
    {
      "title": "What's new in TypeScript 5.5",
      "url": "https://...",
      "snippet": "..."
    }
  ],
  "provenance": {
    "source": "web-search",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

tool.memory-note-write

Category: Memory | Approval: not required | Mutability: mutating

Prompt to invoke:

"Remember that the API rate limit is 1000 requests per hour."

Expected output:

{
  "noteId": "note:1k3x",
  "namespace": "tenant-local/workspace-local/notes/session-abc",
  "persisted": true,
  "trustClassification": "USER_OWNED",
  "createdAt": "2026-05-03T10:00:00.000Z"
}

tool.shell-command

Category: Runtime | Approval: required | Mutability: mutating

Prompt to invoke:

"Run ls -la /tmp/session-data to list what's in that directory."

Expected output (after operator approval):

{
  "exitCode": 0,
  "stdout": "total 16\ndrwxr-xr-x  4 user  staff  128 May  3 10:00 .\n...",
  "stderr": "",
  "durationMs": 38,
  "command": "ls",
  "args": ["-la", "/tmp/session-data"]
}

tool.approval-request

Category: Governance | Approval: not required (creates approval request) | Mutability: mutating

Prompt to invoke:

"Submit an approval request for deleting all files in /tmp/build-cache."

Expected output:

{
  "approvalRequestId": "approval:req789",
  "state": "pending",
  "summary": "Delete all files in /tmp/build-cache",
  "intentId": "intent:del456",
  "expiresAt": "2026-05-03T10:15:00.000Z",
  "createdAt": "2026-05-03T10:00:00.000Z"
}

Runtime Tools (B5)

tool.exec

Category: Runtime | Approval: required | Mutability: mutating

Prompt to invoke:

"Execute node --version to confirm which Node.js version is running in the sandbox."

Expected output (after operator approval):

{
  "exitCode": 0,
  "stdout": "v22.3.0\n",
  "stderr": "",
  "durationMs": 52,
  "command": "node",
  "args": ["--version"],
  "sandboxMode": "no_network_compute"
}

tool.process

Category: Runtime | Approval: required | Mutability: mutating

Prompt to invoke:

"List the currently running sandbox processes."

Expected output (after operator approval):

{
  "processes": [
    {
      "pid": 1234,
      "command": "node",
      "args": ["worker.js"],
      "status": "running",
      "cpuPercent": 2.1,
      "memoryMb": 48
    }
  ],
  "total": 1
}

tool.code-execution

Category: Runtime | Approval: required | Mutability: mutating

Prompt to invoke:

"Calculate the sum of the first 100 prime numbers using Python."

Expected output (after operator approval):

{
  "exitCode": 0,
  "stdout": "24133\n",
  "stderr": "",
  "language": "python",
  "durationMs": 142,
  "sandboxMode": "no_network_compute"
}

tool.bash

Category: Runtime | Approval: required | Mutability: mutating

Prompt to invoke:

"Run a bash script that counts the number of .ts files in the src/ directory."

Expected output (after operator approval):

{
  "exitCode": 0,
  "stdout": "47\n",
  "stderr": "",
  "durationMs": 88,
  "sandboxMode": "no_network_compute"
}

Filesystem Write Tools (B5)

tool.fs-write-file

Category: Filesystem (write) | Approval: required by default | Mutability: mutating

Prompt to invoke:

"Save this analysis to output/summary.txt: 'Monthly traffic increased 12% in April.'"

Expected output:

{
  "path": "output/summary.txt",
  "operation": "write",
  "dryRun": false,
  "wouldChange": true,
  "changed": true,
  "approved": true,
  "hashBefore": null,
  "hashAfter": "sha256:...",
  "sizeBefore": 0,
  "sizeAfter": 44
}

tool.fs-append-file

Category: Filesystem (write) | Approval: required by default | Mutability: mutating

Prompt to invoke:

"Append this line to notes/todo.md: - follow up with policy review."

Expected output:

{
  "path": "notes/todo.md",
  "operation": "append",
  "dryRun": false,
  "wouldChange": true,
  "changed": true,
  "approved": true,
  "hashBefore": "sha256:...",
  "hashAfter": "sha256:...",
  "sizeBefore": 120,
  "sizeAfter": 151
}

tool.fs-apply-patch

Category: Filesystem (write) | Approval: required by default | Mutability: mutating

Prompt to invoke:

"Apply this patch to update the README title: --- a/README.md\n+++ b/README.md\n@@ -1 +1 @@\n-# Old Title\n+# New Title"

Expected output (after operator approval):

{
  "path": "README.md",
  "operation": "patch",
  "dryRun": false,
  "wouldChange": true,
  "changed": true,
  "approved": true,
  "hashBefore": "sha256:...",
  "hashAfter": "sha256:...",
  "sizeBefore": 980,
  "sizeAfter": 984
}

tool.fs-rename-file

Category: Filesystem (write) | Approval: required by default | Mutability: mutating

Prompt to invoke:

"Rename docs/old-name.md to docs/new-name.md."

Expected output (after operator approval):

{
  "fromPath": "docs/old-name.md",
  "toPath": "docs/new-name.md",
  "operation": "rename",
  "dryRun": false,
  "wouldChange": true,
  "changed": true,
  "approved": true
}

Session Tools (B5)

tool.sessions-list

Category: Sessions | Approval: not required | Mutability: read_only

Prompt to invoke:

"List all currently active sessions."

Expected output:

{
  "sessions": [
    {
      "sessionId": "session:abc123",
      "status": "active",
      "tenantId": "tenant-local",
      "workspaceId": "workspace-local",
      "createdAt": "2026-05-03T09:00:00.000Z",
      "iteration": 3
    }
  ],
  "total": 1
}

tool.sessions-history

Category: Sessions | Approval: not required | Mutability: read_only

Prompt to invoke:

"Show me the message history for session session:abc123."

Expected output:

{
  "sessionId": "session:abc123",
  "messages": [
    {
      "role": "user",
      "content": "Search for TypeScript news",
      "timestamp": "2026-05-03T09:01:00.000Z"
    },
    {
      "role": "assistant",
      "content": "Here are the top results...",
      "timestamp": "2026-05-03T09:01:05.000Z"
    }
  ],
  "total": 2
}

tool.session-status

Category: Sessions | Approval: not required | Mutability: read_only

Prompt to invoke:

"What is the current status and risk level of session session:abc123?"

Expected output:

{
  "sessionId": "session:abc123",
  "status": "active",
  "riskLevel": "low",
  "iteration": 3,
  "lastActivityAt": "2026-05-03T09:01:05.000Z",
  "approvalsPending": 0
}

tool.sessions-send

Category: Sessions | Approval: not required | Mutability: mutating

Prompt to invoke:

"Send the message 'The build has completed successfully' to session session:abc123."

Expected output:

{
  "sessionId": "session:abc123",
  "messageId": "msg:xyz456",
  "delivered": true,
  "sentAt": "2026-05-03T10:05:00.000Z"
}

tool.sessions-yield

Category: Sessions | Approval: not required | Mutability: mutating

Prompt to invoke:

"Yield a result back to the parent session with the payload { status: 'done', count: 42 }."

Expected output:

{
  "sessionId": "session:abc123",
  "yielded": true,
  "payload": { "status": "done", "count": 42 },
  "yieldedAt": "2026-05-03T10:06:00.000Z"
}

tool.sessions-spawn

Category: Sessions | Approval: required | Mutability: mutating

Prompt to invoke:

"Spawn a new session using the agent-def:summariser agent definition to process this document in the background."

Expected output (after operator approval):

{
  "sessionId": "session:new789",
  "agentDefinitionId": "agent-def:summariser",
  "status": "active",
  "parentSessionId": "session:abc123",
  "spawnedAt": "2026-05-03T10:07:00.000Z"
}

tool.subagents

Category: Sessions | Approval: required | Mutability: mutating

Prompt to invoke:

"Spawn a subagent using agent definition agent-def:researcher to find papers about transformer architectures and return a summary."

Expected output (after operator approval):

{
  "subagentSessionId": "session:sub101",
  "agentDefinitionId": "agent-def:researcher",
  "status": "active",
  "spawnedAt": "2026-05-03T10:08:00.000Z"
}

Memory Tools (B5)

tool.memory-search

Category: Memory | Approval: not required | Mutability: read_only

Prompt to invoke:

"Search my saved notes for anything about deadlines this week."

Expected output:

{
  "namespace": "tenant-local/workspace-local/notes/user-alice",
  "query": "deadline",
  "results": [
    {
      "noteId": "note:1k3xab",
      "note": "Project alpha milestone due 2026-05-07",
      "noteType": "fact",
      "trustClassification": "USER_OWNED",
      "tags": ["deadline", "project-alpha"],
      "createdAt": "2026-05-01T08:00:00.000Z"
    }
  ],
  "total": 1
}

tool.memory-get

Category: Memory | Approval: not required | Mutability: read_only

Prompt to invoke:

"Retrieve the note with ID note:1k3xab and show me its full content."

Expected output:

{
  "noteId": "note:1k3xab",
  "namespace": "tenant-local/workspace-local/notes/user-alice",
  "note": "Project alpha milestone due 2026-05-07",
  "noteType": "fact",
  "trustClassification": "USER_OWNED",
  "tags": ["deadline", "project-alpha"],
  "createdAt": "2026-05-01T08:00:00.000Z",
  "provenance": {
    "source": "memory-store",
    "trustClassification": "USER_OWNED"
  }
}

Web Tools (B5)

tool.x-search

Category: Web | Approval: not required | Mutability: read_only

Prompt to invoke:

"Search X (Twitter) for recent posts about the Anthropic Claude API and summarise what people are saying."

Expected output:

{
  "query": "Anthropic Claude API",
  "results": [
    {
      "id": "tweet:1234567890",
      "text": "Just tried the new Claude API — the tool use is impressively reliable.",
      "author": "@dev_example",
      "postedAt": "2026-05-03T08:30:00.000Z",
      "url": "https://x.com/dev_example/status/1234567890"
    }
  ],
  "provenance": {
    "source": "x-api",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

UI Tools (B5)

tool.browser

Category: UI | Approval: required | Mutability: mutating

Prompt to invoke:

"Open https://example.com, take a screenshot, and describe what the page looks like."

Expected output (after operator approval):

{
  "operation": "screenshot",
  "url": "https://example.com",
  "screenshotBase64": "iVBORw0KGgo...",
  "width": 1280,
  "height": 800,
  "provenance": {
    "source": "headless-browser",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

tool.canvas

Category: UI | Approval: may require | Mutability: mutating

Prompt to invoke:

"Render a markdown summary of today's deployment status to the operator dashboard canvas."

Expected output:

{
  "operation": "render",
  "format": "markdown",
  "bytesWritten": 312,
  "canvasId": "canvas:main",
  "renderedAt": "2026-05-03T10:10:00.000Z"
}

Automation Tools (B5)

tool.cron

Category: Automation | Approval: required (for create/delete) | Mutability: mutating

Prompt to invoke:

"Schedule a daily job at 8am to search for 'AI safety news' and write a note with the results."

Expected output (after operator approval):

{
  "operation": "create",
  "jobId": "cron:job-abc",
  "name": "Daily AI safety digest",
  "schedule": "0 8 * * *",
  "status": "active",
  "nextRunAt": "2026-05-04T08:00:00.000Z",
  "createdAt": "2026-05-03T10:00:00.000Z"
}

tool.gateway

Category: Automation | Approval: required | Mutability: mutating

Prompt to invoke:

"Invoke the erp-system gateway endpoint to fetch the latest inventory levels."

Expected output (after operator approval):

{
  "endpointId": "erp-system",
  "status": 200,
  "responseBody": { "inventory": [{ "sku": "WIDGET-001", "qty": 450 }] },
  "durationMs": 210,
  "provenance": {
    "source": "operator-gateway",
    "trustClassification": "EXTERNAL_UNTRUSTED"
  }
}

Messaging Tools (B5)

tool.message

Category: Messaging | Approval: may require | Mutability: mutating

Prompt to invoke:

"Send a message to the telegram:ops-alerts channel saying the build succeeded for version 1.4.2."

Expected output:

{
  "messageId": "msg:tg-abc",
  "channel": "telegram:ops-alerts",
  "status": "sent",
  "sentAt": "2026-05-03T10:11:00.000Z"
}

Node Tools (B5)

tool.nodes

Category: Nodes | Approval: not required (read ops) | Mutability: read_only

Prompt to invoke:

"List all available execution nodes and their current status."

Expected output:

{
  "operation": "list",
  "nodes": [
    {
      "nodeId": "node:primary",
      "status": "online",
      "capabilities": ["python", "node", "bash"],
      "load": 0.32,
      "region": "local"
    },
    {
      "nodeId": "node:gpu-01",
      "status": "online",
      "capabilities": ["python", "cuda"],
      "load": 0.71,
      "region": "local"
    }
  ],
  "total": 2
}

Agent Tools (B5)

tool.agents-list

Category: Agents | Approval: not required | Mutability: read_only

Prompt to invoke:

"Show me all available agent definitions I can use or spawn."

Expected output:

{
  "agents": [
    {
      "agentDefinitionId": "agent-def:summariser",
      "name": "Summariser",
      "capabilities": ["web.search", "memory.write"],
      "toolIds": ["tool.web-search", "tool.memory-note-write"],
      "status": "active",
      "version": "1.0.0"
    },
    {
      "agentDefinitionId": "agent-def:researcher",
      "name": "Researcher",
      "capabilities": ["web.search", "http.fetch"],
      "toolIds": ["tool.web-search", "tool.http-fetch"],
      "status": "active",
      "version": "1.0.0"
    }
  ],
  "total": 2
}

Notes on output trust

All outputs from remote/external sources carry provenance.trustClassification: "EXTERNAL_UNTRUSTED". The agent will cite this when using external results in its response. Memory records written by the user carry USER_OWNED and that classification is preserved through all subsequent reads.

Tools requiring approval return a pendingApproval: true response immediately; the tool result arrives only after the operator acts on the approval request in the admin dashboard.

Index​

Core Tools (B4)​

tool.local-file-read​

tool.http-fetch​

tool.web-search​

tool.memory-note-write​

tool.shell-command​

tool.approval-request​

Runtime Tools (B5)​

tool.exec​

tool.process​

tool.code-execution​

tool.bash​

Filesystem Write Tools (B5)​

tool.fs-write-file​

tool.fs-append-file​

tool.fs-apply-patch​

tool.fs-rename-file​

Session Tools (B5)​

tool.sessions-list​

tool.sessions-history​

tool.session-status​

tool.sessions-send​

tool.sessions-yield​

tool.sessions-spawn​

tool.subagents​

Memory Tools (B5)​

tool.memory-search​

tool.memory-get​

Web Tools (B5)​

tool.x-search​

UI Tools (B5)​

tool.browser​

tool.canvas​

Automation Tools (B5)​

tool.cron​

tool.gateway​

Messaging Tools (B5)​

tool.message​

Node Tools (B5)​

tool.nodes​

Agent Tools (B5)​

tool.agents-list​

Notes on output trust​

Index

Core Tools (B4)

tool.local-file-read

tool.http-fetch

tool.web-search

tool.memory-note-write

tool.shell-command

tool.approval-request

Runtime Tools (B5)

tool.exec

tool.process

tool.code-execution

tool.bash

Filesystem Write Tools (B5)

tool.fs-write-file

tool.fs-append-file

tool.fs-apply-patch

tool.fs-rename-file

Session Tools (B5)

tool.sessions-list

tool.sessions-history

tool.session-status

tool.sessions-send

tool.sessions-yield

tool.sessions-spawn

tool.subagents

Memory Tools (B5)

tool.memory-search

tool.memory-get

Web Tools (B5)

tool.x-search

UI Tools (B5)

tool.browser

tool.canvas

Automation Tools (B5)

tool.cron

tool.gateway

Messaging Tools (B5)

tool.message

Node Tools (B5)

tool.nodes

Agent Tools (B5)

tool.agents-list

Notes on output trust