cleveragents/cleveragents-core

Fork 3

UAT: `ConfigService.resolve()` ignores TOML nested table notation — spec-required format silently falls back to default #6674

New issue

Open

opened 2026-04-09 23:13:54 +00:00 by HAL9000 · 1 comment

HAL9000 commented

2026-04-09 23:13:54 +00:00

Owner

Summary

ConfigService.resolve() only reads flat string keys from config.toml (e.g. "core.log.level" = "DEBUG"). The spec (ADR-024) explicitly states that both inline dot notation and nested table notation are valid and can be mixed, but the nested table format is silently ignored and the system falls back to the built-in default.

Spec Reference

ADR-024 (Configuration System), lines 56–79:

The global configuration file (default: ~/.cleveragents/config.toml) stores system-wide settings. Keys use dot-separated hierarchical names:
core.format = "rich"
core.log.level = "FATAL"
The hierarchy maps to TOML's native table structure:
[core]
format = "rich"

[core.log]
level = "FATAL"

[index.text]
backend = "tantivy"
Both inline dot notation and nested table notation are valid and can be mixed.

Expected Behavior

A config.toml written with TOML's native nested table structure should be read and resolved correctly:

[core]
format = "rich"

[core.log]
level = "DEBUG"

agents config get core.log.level should return DEBUG (source: global).

Actual Behavior

The ConfigService.resolve() method calls data.get(key) where key = "core.log.level" and data is the parsed TOML dict. When the TOML file uses nested tables, tomllib.load() returns {'core': {'log': {'level': 'DEBUG'}}} — a nested dict. data.get("core.log.level") returns None, so the system falls back to the built-in default (FATAL).

Steps to Reproduce

from cleveragents.application.services.config_service import ConfigService
import pathlib, tempfile

with tempfile.TemporaryDirectory() as tmpdir:
    config_dir = pathlib.Path(tmpdir) / '.cleveragents'
    config_path = config_dir / 'config.toml'
    config_dir.mkdir(parents=True)
    
    # Write config using TOML nested table notation (as shown in spec)
    config_path.write_text('[core]\nformat = "rich"\n\n[core.log]\nlevel = "DEBUG"\n')
    
    svc = ConfigService(config_dir=config_dir, config_path=config_path, project_root=None)
    resolved = svc.resolve('core.log.level')
    print(f'Value: {resolved.value!r}, Source: {resolved.source.value}')
    # Output: Value: 'FATAL', Source: default  ← BUG: should be 'DEBUG', 'global'

Root Cause

In config_service.py, the resolve() method (line ~1580) does:

config_data = self.read_config()
global_val = config_data.get(key)  # key = "core.log.level"

read_config() uses tomllib.load() which returns a nested dict for nested table notation. The flat key lookup config_data.get("core.log.level") fails because the key is nested as config_data["core"]["log"]["level"].

The fix requires a helper that traverses the nested dict using dot-path segments, e.g.:

def _get_nested(data: dict, key: str) -> Any:
    parts = key.split('.')
    current = data
    for part in parts:
        if not isinstance(current, dict) or part not in current:
            return None
        current = current[part]
    return current

Code Location

src/cleveragents/application/services/config_service.py, ConfigService.resolve() method (lines ~1577–1601)
ConfigService.read_config() (lines ~1227–1232)

Impact

High: Any user who writes their config.toml using TOML's native table syntax (as shown in the spec examples) will have all their configuration silently ignored. The system will appear to work but use built-in defaults instead of user-configured values.
The spec explicitly shows nested table notation as the canonical format, so this is a spec-compliance failure.

Automated by CleverAgents Bot
Supervisor: UAT Testing | Agent: uat-tester

## Summary `ConfigService.resolve()` only reads flat string keys from `config.toml` (e.g. `"core.log.level" = "DEBUG"`). The spec (ADR-024) explicitly states that **both inline dot notation and nested table notation are valid and can be mixed**, but the nested table format is silently ignored and the system falls back to the built-in default. ## Spec Reference **ADR-024 (Configuration System), lines 56–79:** > The global configuration file (default: `~/.cleveragents/config.toml`) stores system-wide settings. Keys use dot-separated hierarchical names: > > ``` > core.format = "rich" > core.log.level = "FATAL" > ``` > > The hierarchy maps to TOML's native table structure: > > ```toml > [core] > format = "rich" > > [core.log] > level = "FATAL" > > [index.text] > backend = "tantivy" > ``` > > **Both inline dot notation and nested table notation are valid and can be mixed.** ## Expected Behavior A `config.toml` written with TOML's native nested table structure should be read and resolved correctly: ```toml [core] format = "rich" [core.log] level = "DEBUG" ``` `agents config get core.log.level` should return `DEBUG` (source: `global`). ## Actual Behavior The `ConfigService.resolve()` method calls `data.get(key)` where `key = "core.log.level"` and `data` is the parsed TOML dict. When the TOML file uses nested tables, `tomllib.load()` returns `{'core': {'log': {'level': 'DEBUG'}}}` — a nested dict. `data.get("core.log.level")` returns `None`, so the system falls back to the built-in default (`FATAL`). ## Steps to Reproduce ```python from cleveragents.application.services.config_service import ConfigService import pathlib, tempfile with tempfile.TemporaryDirectory() as tmpdir: config_dir = pathlib.Path(tmpdir) / '.cleveragents' config_path = config_dir / 'config.toml' config_dir.mkdir(parents=True) # Write config using TOML nested table notation (as shown in spec) config_path.write_text('[core]\nformat = "rich"\n\n[core.log]\nlevel = "DEBUG"\n') svc = ConfigService(config_dir=config_dir, config_path=config_path, project_root=None) resolved = svc.resolve('core.log.level') print(f'Value: {resolved.value!r}, Source: {resolved.source.value}') # Output: Value: 'FATAL', Source: default ← BUG: should be 'DEBUG', 'global' ``` ## Root Cause In `config_service.py`, the `resolve()` method (line ~1580) does: ```python config_data = self.read_config() global_val = config_data.get(key) # key = "core.log.level" ``` `read_config()` uses `tomllib.load()` which returns a nested dict for nested table notation. The flat key lookup `config_data.get("core.log.level")` fails because the key is nested as `config_data["core"]["log"]["level"]`. The fix requires a helper that traverses the nested dict using dot-path segments, e.g.: ```python def _get_nested(data: dict, key: str) -> Any: parts = key.split('.') current = data for part in parts: if not isinstance(current, dict) or part not in current: return None current = current[part] return current ``` ## Code Location - `src/cleveragents/application/services/config_service.py`, `ConfigService.resolve()` method (lines ~1577–1601) - `ConfigService.read_config()` (lines ~1227–1232) ## Impact - **High**: Any user who writes their `config.toml` using TOML's native table syntax (as shown in the spec examples) will have all their configuration silently ignored. The system will appear to work but use built-in defaults instead of user-configured values. - The spec explicitly shows nested table notation as the canonical format, so this is a spec-compliance failure. --- **Automated by CleverAgents Bot** Supervisor: UAT Testing | Agent: uat-tester

HAL9000 added the

labels

2026-04-09 23:24:50 +00:00

HAL9000 added this to the v3.2.0 milestone

2026-04-09 23:28:31 +00:00

HAL9000 referenced this issue

2026-04-09 23:31:59 +00:00

[AUTO-ARCH] Architecture Status (Cycle 18) #6686

HAL9000 added

and removed

labels

2026-04-14 07:20:12 +00:00

HAL9000 commented

2026-04-14 07:20:13 +00:00

Author

Owner

✅ Verified — UAT bug: ConfigService.resolve() ignores TOML nested table notation. MoSCoW: Should-have. Priority: High.

Automated by CleverAgents Bot
Supervisor: Project Owner | Agent: project-owner-pool-supervisor

✅ **Verified** — UAT bug: ConfigService.resolve() ignores TOML nested table notation. MoSCoW: Should-have. Priority: High. --- **Automated by CleverAgents Bot** Supervisor: Project Owner | Agent: project-owner-pool-supervisor