Skip to content

docs: Update messaging-bots for smart defaults + auto-approve behavioral changes #220

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
MervinPraison wants to merge 1 commit into
base: main
Choose a base branch
from
Open

docs: Update messaging-bots for smart defaults + auto-approve behavioral changes #220

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions docs/features/bot-gateway.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,6 +26,8 @@ flowchart LR

Run all your bots from one command. The Gateway Server manages multiple bot connections and routes messages to the right AI agent.

**Parity with Bot()**: `praisonai gateway start` now applies the same smart defaults as `praisonai bot start`, resolving the previous "zero tools in daemon mode" issue. Both entry points produce identical behavior with safe tools auto-injected and auto-approval enabled by default.
Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The claim that praisonai gateway start now applies the same smart defaults as Bot() is not reflected in the implementation of praisonai/gateway/server.py. The gateway server instantiates bots using platform-specific adapters (e.g., TelegramBot) directly and does not appear to invoke the _apply_smart_defaults logic found in the Bot wrapper class. This means features like auto-injected safe tools and session history might be missing when running via the gateway.

Copy link
Copy Markdown

@coderabbitai coderabbitai Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟠 Major

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Verify whether gateway startup applies Bot smart defaults or channel-level defaults.
# Expected: the gateway agent initialization should call an equivalent of _apply_smart_defaults,
# or explicitly inject safe default tools and auto_approve_tools before registering agents.
rg -n -C10 'Agent\(|tools=agent_def\.get|auto_approve_tools|default_tools|_apply_smart_defaults' --iglob 'server.py'

Repository: MervinPraison/PraisonAIDocs

Length of output: 8570

🏁 Script executed:

# First, check the actual content of line 29 in bot-gateway.mdx
head -n 40 docs/features/bot-gateway.mdx | tail -n 15

Repository: MervinPraison/PraisonAIDocs

Length of output: 641

🏁 Script executed:

# Check full context of agent initialization in gateway server.py around line 1230-1290
sed -n '1230,1290p' praisonai/gateway/server.py

Repository: MervinPraison/PraisonAIDocs

Length of output: 2927

🏁 Script executed:

# Search for _apply_smart_defaults usage in gateway-related files
rg -n '_apply_smart_defaults' --glob '*gateway*' -A 3 -B 3

Repository: MervinPraison/PraisonAIDocs

Length of output: 53

🏁 Script executed:

# Search for auto_approve_tools in gateway files
rg -n 'auto_approve_tools' --glob '*gateway*' -A 2 -B 2

Repository: MervinPraison/PraisonAIDocs

Length of output: 53

🏁 Script executed:

# Check where agent_tools comes from in the agent creation context
sed -n '1200,1280p' praisonai/gateway/server.py | head -n 100

Repository: MervinPraison/PraisonAIDocs

Length of output: 4030

🏁 Script executed:

# Search for Bot class and _apply_smart_defaults in the codebase
rg -n 'class Bot|def _apply_smart_defaults' -A 5 --glob '*.py'

Repository: MervinPraison/PraisonAIDocs

Length of output: 7560

🏁 Script executed:

# Check what Bot() does for tools and auto_approve_tools
rg -n 'auto_approve_tools|safe.*tools|default.*tools' --glob 'bot.py' -A 2 -B 2

Repository: MervinPraison/PraisonAIDocs

Length of output: 3006

Remove or correct the smart defaults claim on line 29.

Line 29 claims praisonai gateway start applies the same smart defaults as praisonai bot start, but the gateway implementation (praisonai/gateway/server.py:1271-1280) does not call Bot._apply_smart_defaults(). The gateway only uses tools explicitly configured in YAML via tools=agent_tools if agent_tools else None, which means agents with zero configured tools still get zero tools—the "zero tools in daemon mode" issue is not resolved. Additionally, the gateway does not wire auto_approve_tools to an AutoApproveBackend as Bot does. Either implement these smart defaults in the gateway or correct the documentation to accurately describe the gateway's actual behavior.

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In `@docs/features/bot-gateway.mdx` at line 29, Update the documentation sentence
that claims parity with Bot(): remove or correct the statement that "praisonai
gateway start applies the same smart defaults as praisionai bot start" and
instead state that the gateway does not call Bot._apply_smart_defaults() and
only uses tools passed as tools=agent_tools if agent_tools else None (so agents
with zero configured tools remain tool-less), and that the gateway does not wire
auto_approve_tools to an AutoApproveBackend; alternatively, if you prefer to
keep the parity claim implement the missing behavior by invoking
Bot._apply_smart_defaults() and wiring auto_approve_tools to an
AutoApproveBackend in the gateway code paths referenced, but for this change
simply edit the doc sentence to accurately reflect the gateway’s actual
behavior.


## Quick Start

<Steps>
Expand Down
100 changes: 97 additions & 3 deletions docs/features/messaging-bots.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -293,6 +293,84 @@ The streaming bridge hooks into the same `StreamEventEmitter` used by the web ch

---

## Smart Defaults

Bots ship with sensible defaults so you can start chatting immediately — no tool wiring required.

```mermaid
graph LR
A[🤖 Agent<br/>tools=None] --> SD[⚙️ Smart Defaults]
SD --> T[🔍 search_web<br/>📅 schedule_*<br/>🧠 memory<br/>📚 learning]
SD --> AA[✅ auto-approve<br/>for safe tools]
SD --> H[💬 session history<br/>last 20 messages]

classDef agent fill:#8B0000,stroke:#7C90A0,color:#fff
classDef config fill:#6366F1,stroke:#7C90A0,color:#fff
classDef tools fill:#189AB4,stroke:#7C90A0,color:#fff
classDef result fill:#10B981,stroke:#7C90A0,color:#fff

class A agent
class SD config
class T tools
class AA,H result
```

Both `praisonai bot start` and `praisonai gateway start` apply the same defaults:

| Default | Applied when | What you get |
|---------|--------------|--------------|
| Safe tools | Agent has no `tools` configured | `search_web`, `web_crawl`, `schedule_*`, `store_memory`/`search_memory`, `store_learning`/`search_learning` |
| Auto-approval | `auto_approve_tools=True` (default) | Tool calls run without CLI prompts — chat bots can't show them anyway |
Comment on lines +318 to +323
Copy link
Copy Markdown

@coderabbitai coderabbitai Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟠 Major

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Verify whether channel-level gateway config is parsed and applied to agent defaults.
# Expected: channel config fields auto_approve_tools/default_tools are read and passed into
# smart-default application before agent registration or bot/channel startup.
rg -n -C10 'auto_approve_tools|default_tools|channels|routes|routing|tools=agent_def\.get|_apply_smart_defaults' --iglob 'server.py' --iglob '*.py'

Repository: MervinPraison/PraisonAIDocs

Length of output: 50384

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# First, find the gateway/server.py file
fd -type f -name "server.py" | grep gateway

# Then examine the specific line ranges mentioned (1230-1280, 1279-1308)
if [ -f "praisonai/gateway/server.py" ]; then
  echo "=== Lines 1230-1280 ===" 
  sed -n '1230,1280p' praisonai/gateway/server.py | cat -n
  echo ""
  echo "=== Lines 1279-1310 ===" 
  sed -n '1279,1310p' praisonai/gateway/server.py | cat -n
fi

Repository: MervinPraison/PraisonAIDocs

Length of output: 241

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Search for where agents are created in gateway and how config is applied
# Look for agent instantiation patterns and config handling
rg -n "agent_defs|channel.*config|auto_approve|default_tools|_apply_smart_defaults" praisonai/gateway/server.py | head -100

Repository: MervinPraison/PraisonAIDocs

Length of output: 320

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check file size first
wc -l praisonai/gateway/server.py || echo "File not found"

Repository: MervinPraison/PraisonAIDocs

Length of output: 103

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Read the specific lines mentioned in the review
echo "=== Lines 1230-1280 ===" 
sed -n '1230,1280p' praisonai/gateway/server.py

echo ""
echo "=== Lines 1279-1320 ===" 
sed -n '1279,1320p' praisonai/gateway/server.py

Repository: MervinPraison/PraisonAIDocs

Length of output: 4079

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Find the _create_bot method to understand how agents are instantiated
rg -n "_create_bot" praisonai/gateway/server.py -A 30 | head -80

Repository: MervinPraison/PraisonAIDocs

Length of output: 2445

🏁 Script executed:

#!/bin/bash
set -euo pivefail

# Search for Bot() instantiation to see what parameters are passed
rg -n "Bot\(" praisonai/gateway/server.py -B 3 -A 5 | head -100

Repository: MervinPraison/PraisonAIDocs

Length of output: 124

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Look for Bot() instantiation in bots module to see what parameters it accepts
rg -n "class.*Bot|def __init__" praisonai/bots/ -A 15 | grep -A 15 "auto_approve\|default_tools" | head -50

Repository: MervinPraison/PraisonAIDocs

Length of output: 53

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check what the config object passed to Bot() contains
# Search for how gateway config is structured and passed to bots
rg -n "self.config|config\.get|_apply_smart_defaults|auto_approve_tools|default_tools" praisonai/gateway/server.py

Repository: MervinPraison/PraisonAIDocs

Length of output: 1475

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Look at the Bot class initialization to see what parameters it accepts
rg -n "class Bot" praisonai/bots/__init__.py -A 30 | head -60

Repository: MervinPraison/PraisonAIDocs

Length of output: 53

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Find if there's any smart defaults application in the gateway initialization
rg -n "smart.*default|_apply_smart_defaults" praisonai/ -r 2>/dev/null | head -20

Repository: MervinPraison/PraisonAIDocs

Length of output: 53

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Find the Bot class definition to understand what parameters it accepts
find praisonai -name "*.py" -type f | xargs grep -l "^class Bot" | head -5

Repository: MervinPraison/PraisonAIDocs

Length of output: 204

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check the bots module structure
ls -la praisonai/bots/ 2>/dev/null || echo "Directory not found"

# Look for where auto_approve_tools might be defined
rg -n "auto_approve_tools|default_tools" praisonai/ --type py | head -30

Repository: MervinPraison/PraisonAIDocs

Length of output: 4615

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check if channel config in gateway.yaml gets passed to agent or bot
# Look at _create_bot signature and how ch_cfg (channel config) is used
sed -n '1413,1470p' praisonai/gateway/server.py

Repository: MervinPraison/PraisonAIDocs

Length of output: 2723

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check the Bot class __init__ to see what it does with config parameter
sed -n '1,250p' praisonai/bots/bot.py | cat -n

Repository: MervinPraison/PraisonAIDocs

Length of output: 10852

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check how gateway config is constructed and whether channel config is merged in
# Look at the initialization of gateway config and how it's passed down
rg -n "GatewayConfig|self.config.*=" praisonai/gateway/server.py | head -40

Repository: MervinPraison/PraisonAIDocs

Length of output: 406

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Check if channel config (ch_cfg) is ever merged with gateway config before Bot creation
sed -n '1300,1430p' praisonai/gateway/server.py | cat -n

Repository: MervinPraison/PraisonAIDocs

Length of output: 6704

Don't advertise gateway/channel-level smart defaults unless gateway config consumes them.

The documentation claims that channel-level auto_approve_tools / default_tools settings work in gateway with feature parity to Bot(). However, praisonai/gateway/server.py:1330-1390 shows gateway's start_channels() method creates a BotConfig from channel YAML (lines 1382-1387) that only extracts: token, allowed_users, allowed_channels, and mention_required. The channel config does not read or pass auto_approve_tools or default_tools fields to BotConfig. Users adding these fields to their channel config will find them silently ignored—the smart defaults never get applied because the configuration is never parsed from the YAML.

Also applies to: 1279-1308

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In `@docs/features/messaging-bots.mdx` around lines 318 - 323, The gateway's
start_channels() currently builds a BotConfig from channel YAML but only
extracts token, allowed_users, allowed_channels, and mention_required, so
channel-level auto_approve_tools and default_tools are ignored; update
start_channels() to parse auto_approve_tools and default_tools from the channel
YAML and pass them into the BotConfig constructor (the same fields used by
Bot()), ensure BotConfig accepts these fields if not already, and add a small
unit/integration test that a channel YAML with auto_approve_tools/default_tools
results in a BotConfig with those values.

| Session history | Agent has no `memory` configured | Last 20 messages remembered per user, zero-dep |
Comment on lines +320 to +324
Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

It should be noted that these smart defaults currently only apply to individual Agent instances. The implementation in praisonai/bots/bot.py (lines 150-152) explicitly excludes AgentTeam and AgentFlow from receiving these automatic configurations.


### Opting out

| Goal | How |
|------|-----|
| Run with **zero** tools | Set `tools: []` explicitly in YAML, or pass `tools=[]` to `Agent` |
Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The documentation suggests that passing tools=[] allows running with zero tools. However, the implementation in praisonai/bots/bot.py (lines 181-182) uses getattr(agent, 'tools', None) or [] and then checks if not current_tools:. This logic treats an empty list the same as None, triggering the automatic injection of default tools. As a result, there is currently no way to explicitly opt out of default tools using an empty list.

| Require manual approval | Set `auto_approve_tools: false` in the channel config |
| Override the tool list | Set `default_tools: [...]` under the channel — destructive tools are still filtered |

<Warning>
Destructive tools (`execute_command`, `delete_file`, `write_file`, `shell_command`) are **never auto-injected**, even if you add them to `default_tools`. Wire them explicitly on the agent and add a chat-level approval flow.
Comment on lines +322 to +335
Copy link
Copy Markdown

@coderabbitai coderabbitai Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟠 Major

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Verify BotConfig defaults and smart-default semantics.
# Expected for these docs to be accurate:
# - auto_approve_tools defaults to True
# - default_tools excludes execute_command and other dangerous tools
# - _apply_smart_defaults distinguishes missing tools from explicit tools=[]
# - dangerous tools are filtered before auto-injection
rg -n -C4 'default_tools|auto_approve_tools' --iglob 'config.py'
rg -n -C8 'def _apply_smart_defaults|if not agent\.tools|agent\.tools\.extend|auto_approve_tools|DEFAULT_DANGEROUS_TOOLS' --iglob 'bot.py' --iglob 'registry.py'

Repository: MervinPraison/PraisonAIDocs

Length of output: 9242

🏁 Script executed:

# See full _apply_smart_defaults method and related tool injection/filtering logic
rg -n -A30 'def _apply_smart_defaults' praisonai/bots/bot.py

# Check if tools are extended or filtered during smart defaults
rg -n -A20 'agent\.tools\s*=' praisonai/bots/bot.py

# Look for dangerous tool filtering logic
rg -n 'DANGEROUS_TOOLS|dangerous.*tool|filter.*tool' --iglob 'bot.py'

Repository: MervinPraison/PraisonAIDocs

Length of output: 3418

🏁 Script executed:

sed -n '136,210p' praisonai/bots/bot.py

Repository: MervinPraison/PraisonAIDocs

Length of output: 3833

🏁 Script executed:

# Find tool profiles definition
rg -n -A20 'resolve_profiles|def resolve_profiles' --iglob 'profiles.py'

# Check all references to default_tools from BotConfig
rg -n 'default_tools' --iglob 'bot.py' --iglob 'adapter' --iglob 'gateway'

# Verify if BotConfig.default_tools is used in the channel config
rg -n -B5 -A5 'default_tools.*:' docs/features/messaging-bots.mdx

Repository: MervinPraison/PraisonAIDocs

Length of output: 6018

🏁 Script executed:

# Find BUILTIN_PROFILES definition
rg -n -A100 'BUILTIN_PROFILES' praisonaiagents/tools/profiles.py | head -150

Repository: MervinPraison/PraisonAIDocs

Length of output: 5306

Update documented defaults and correct tool opt-out semantics.

These lines document auto_approve_tools=True as the default and tools: [] as an explicit opt-out, but the implementation contradicts both:

  • auto_approve_tools default is False: praisonaiagents/bots/config.py:60 sets auto_approve_tools: bool = False, not True. Docs line 1279 incorrectly claim true is the default for chat bots.
  • tools: [] does NOT opt out: praisonai/bots/bot.py:177–178 treats tools=None and tools=[] identically via if not current_tools:, so setting tools: [] still triggers smart defaults injection, not prevention.
  • default_tools in config is unused: BotConfig.default_tools (which includes execute_command) is never referenced in _apply_smart_defaults. The code instead uses resolve_profiles("web", "schedule", "memory", "learning") which only includes safe tools (search_web, schedule_*, store_memory, search_learning). The promise that destructive tools are never auto-injected is correct but misleading—it's by design (safe profiles only), not by filtering.

Channel-level default_tools overrides (docs line 332) have no effect. Remove the claim about setting default_tools under the channel, and clarify that disabling smart defaults entirely requires pre-configuring the agent before passing it to Bot().

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In `@docs/features/messaging-bots.mdx` around lines 322 - 335, Update the docs to
reflect the actual behavior: state that BotConfig.auto_approve_tools
(auto_approve_tools) defaults to False, not True; remove the claim that setting
default_tools at the channel level overrides injected tools
(BotConfig.default_tools is not used by _apply_smart_defaults); and clarify that
tools: [] does not opt out of smart defaults because Bot._apply_smart_defaults
treats empty lists the same as None, so to disable smart-default injection you
must pre-configure the agent (set tools explicitly on the Agent before passing
it to Bot) or change the code to treat None vs [] differently.

</Warning>

<Info>
Upgrading from an older release? `auto_approve_tools` used to default to `False`. If your bot relied on manual approval, set `auto_approve_tools: false` explicitly.
</Info>

```python
from praisonaiagents import Agent
from praisonai.bots import Bot

# Zero config — gets search_web, schedule_*, memory, learning, and auto-approval
agent = Agent(name="assistant", instructions="Help the user")
bot = Bot("telegram", agent=agent)
bot.run()
```

```bash
# Same result — no YAML tool list needed
praisonai bot telegram --token $TELEGRAM_BOT_TOKEN
```

```mermaid
sequenceDiagram
participant U as 👤 User (Telegram)
participant B as 🤖 Bot
participant A as 🧠 Agent
participant T as 🔍 search_web

U->>B: "What's the latest on Llama 4?"
B->>A: Forward message (no tool prompt)
A->>T: auto-approved call
T-->>A: Results
A-->>B: Answer
B-->>U: 📰 Here's the latest...
```

---

## Socket Mode vs Webhook

PraisonAI bots support two connection modes:
Expand Down Expand Up @@ -345,7 +423,7 @@ config = BotConfig(
reply_in_thread=False, # Reply in threads (default: inline)
thread_threshold=500, # Auto-thread if response > N chars (0 = disabled)
group_policy="mention_only", # Group chat policy: respond_all, mention_only, command_only
auto_approve_tools=False, # Auto-approve tool executions (skip confirmation)
auto_approve_tools=True, # Auto-approve safe tool executions (default: True for chat bots)
)
```

Expand All @@ -363,8 +441,8 @@ config = BotConfig(
| `reply_in_thread` | `bool` | `False` | Always reply in threads |
| `thread_threshold` | `int` | `500` | Auto-thread responses longer than N chars (0 = disabled) |
| `group_policy` | `str` | `"mention_only"` | Group chat behavior: `respond_all`, `mention_only`, or `command_only` |
| `default_tools` | `list` | `["execute_command", ...]` | Default tools enabled for all bots |
| `auto_approve_tools` | `bool` | `False` | Skip tool execution confirmation (for trusted environments) |
| `default_tools` | `list[str]` | `["search_web", "web_crawl", "schedule_add", "schedule_list", "schedule_remove", "store_memory", "search_memory", "store_learning", "search_learning"]` | Safe tools auto-injected when the agent has no tools configured. Destructive tools (`execute_command`, `delete_file`, `write_file`, `shell_command`) are excluded from auto-injection even if listed. |
Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

There is a discrepancy between the default_tools listed here and the implementation in praisonaiagents/bots/config.py (line 52), which still includes "execute_command". Additionally, the smart default injection logic in praisonai/bots/bot.py (lines 186-191) uses hardcoded profiles and does not seem to respect the default_tools attribute from the configuration.

| `auto_approve_tools` | `bool` | `True` | Skip confirmation for safe tool execution. Chat bots can't show CLI approval prompts, so this defaults to `True`. Set `False` to require approval (only useful if you wire a chat-level approval flow). |
Copy link
Copy Markdown

@gemini-code-assist gemini-code-assist Bot Apr 22, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high

The documentation states that auto_approve_tools defaults to True. However, the BotConfig dataclass in praisonaiagents/bots/config.py (line 60) still defaults to False. Furthermore, the logic in praisonai/bots/bot.py (line 155) only enables auto-approval if a configuration is explicitly provided and has this flag set to True. This inconsistency could lead to tool calls hanging silently while waiting for a CLI prompt that the user cannot see.

| `retry_attempts` | `int` | `3` | Number of retry attempts for failed operations |
| `polling_interval` | `float` | `1.0` | Interval for polling mode (seconds) |

Expand Down Expand Up @@ -1198,6 +1276,10 @@ channels:
slack:
token: "${SLACK_BOT_TOKEN}"
app_token: "${SLACK_APP_TOKEN}"
auto_approve_tools: true # NEW (default: true for chat bots)
# default_tools: # NEW — override the safe tool list per channel
# - search_web
# - store_memory
routing:
dm: "personal"
channel: "support"
Expand All @@ -1223,6 +1305,8 @@ channels:
praisonai gateway --config gateway.yaml
```

The gateway now produces identical results to `Bot()` — agents get the same safe tools and auto-approval in both entry points.

```mermaid
graph LR
T[Telegram]:::platform --> GW[Gateway]:::tool
Expand Down Expand Up @@ -1265,6 +1349,8 @@ agent:
praisonai bot start --config bot.yaml
```

> **Tip:** You can omit `tools:` entirely — the bot auto-injects safe defaults (`search_web`, schedule, memory, learning). Keep `tools: []` only if you want the bot to run with zero tools.

<Tip>
The `.env` file in the current directory is auto-loaded, so you can store tokens there and reference them with `${VAR_NAME}` syntax.
</Tip>
Expand Down Expand Up @@ -1345,6 +1431,14 @@ Web mode uses a reverse-engineered protocol. Your number may be banned by Meta.
<Accordion title="Handle rate limits gracefully">
Configure `retry_attempts` and implement exponential backoff for API rate limits.
</Accordion>

<Accordion title="Safe default tools only">
The default tool list is intentionally safe (`search_web`, `schedule_*`, memory/learning). Tools like `execute_command` require explicit opt-in and should be paired with an approval backend. See [Approval Protocol](/features/approval-protocol).
</Accordion>

<Accordion title="Set approval flow for auto-approve disabled">
Only set `auto_approve_tools: false` if you've wired a chat-level approval flow (e.g. `SlackApproval`). Otherwise tool calls will hang silently waiting for a CLI prompt the user cannot see.
</Accordion>
</AccordionGroup>

## Multi-Agent Configuration
Expand Down