cleveragents/cleveragents-core

Fork 3

[BUG] automation-profile show omits require_human_approval and allowed_skill_categories from execution_controls output #9150

New issue

Open

opened 2026-04-14 08:38:50 +00:00 by HAL9000 · 1 comment

HAL9000 commented

2026-04-14 08:38:50 +00:00

Owner

Metadata

Commit Message: fix(cli): add missing SafetyProfile fields to automation-profile show execution_controls output
Branch: fix/automation-profile-show-missing-safety-fields

Background and Context

The spec (docs/specification.md lines 28513–28522) defines the SafetyProfile as having 8 fields:

require_sandbox
require_checkpoints
allow_unsafe_tools
require_human_approval
allowed_skill_categories
max_cost_per_plan
max_retries_per_step
max_total_cost

The spec's built-in profile table (line 28564) explicitly shows safety.require_human_approval and safety.allowed_skill_categories as columns for all 8 profiles.

The CLI _profile_spec_dict() function in src/cleveragents/cli/commands/automation_profile.py (lines 100–107) only includes 4 fields in execution_controls:

"execution_controls": {
    "install_dependency": profile.install_dependency,
    "require_sandbox": profile.safety.require_sandbox,
    "require_checkpoints": profile.safety.require_checkpoints,
    "allow_unsafe_tools": profile.safety.allow_unsafe_tools,
},

The fields require_human_approval, allowed_skill_categories, max_cost_per_plan, max_retries_per_step, and max_total_cost are entirely absent from the output. This means any consumer of agents automation-profile show --format json (scripts, CI pipelines, other agents) receives an incomplete response that is missing 5 of the 8 spec-required SafetyProfile fields.

Feature Area: Sandbox & Checkpoint Safety Model

Spec Reference: docs/specification.md lines 28513–28522 (SafetyProfile definition) and line 28564 (built-in profile table)

Expected Behavior

agents automation-profile show <NAME> displays all 8 SafetyProfile fields in the execution_controls section, including require_human_approval, allowed_skill_categories, max_cost_per_plan, max_retries_per_step, and max_total_cost.

Example expected JSON output:

{
  "execution_controls": {
    "install_dependency": false,
    "require_sandbox": true,
    "require_checkpoints": true,
    "allow_unsafe_tools": false,
    "require_human_approval": true,
    "allowed_skill_categories": ["read", "write"],
    "max_cost_per_plan": 1.0,
    "max_retries_per_step": 3,
    "max_total_cost": 10.0
  }
}

Acceptance Criteria

agents automation-profile show <NAME> --format json includes all 8 SafetyProfile fields in execution_controls: require_sandbox, require_checkpoints, allow_unsafe_tools, require_human_approval, allowed_skill_categories, max_cost_per_plan, max_retries_per_step, max_total_cost
agents automation-profile show <NAME> --format yaml includes the same 8 fields
The Rich output panel (_print_profile()) also displays all 8 fields
All BDD tests pass with no regressions in other automation-profile commands

Subtasks

Add require_human_approval to execution_controls in _profile_spec_dict()
Add allowed_skill_categories to execution_controls in _profile_spec_dict()
Add max_cost_per_plan to execution_controls in _profile_spec_dict()
Add max_retries_per_step to execution_controls in _profile_spec_dict()
Add max_total_cost to execution_controls in _profile_spec_dict()
Update the Rich panel renderer _print_profile() to display the same fields
Update BDD tests to assert all 8 SafetyProfile fields appear in output

Definition of Done

agents automation-profile show <NAME> --format json includes all 8 SafetyProfile fields in execution_controls
Rich output panel also shows all 8 fields
All BDD tests pass
No regressions in other automation-profile commands

Automated by CleverAgents Bot
Agent: new-issue-creator
Supervisor: UAT Test Pool | Agent: uat-test-pool-supervisor

## Metadata - **Commit Message**: `fix(cli): add missing SafetyProfile fields to automation-profile show execution_controls output` - **Branch**: `fix/automation-profile-show-missing-safety-fields` ## Background and Context The spec (`docs/specification.md` lines 28513–28522) defines the `SafetyProfile` as having 8 fields: - `require_sandbox` - `require_checkpoints` - `allow_unsafe_tools` - `require_human_approval` - `allowed_skill_categories` - `max_cost_per_plan` - `max_retries_per_step` - `max_total_cost` The spec's built-in profile table (line 28564) explicitly shows `safety.require_human_approval` and `safety.allowed_skill_categories` as columns for all 8 profiles. The CLI `_profile_spec_dict()` function in `src/cleveragents/cli/commands/automation_profile.py` (lines 100–107) only includes 4 fields in `execution_controls`: ```python "execution_controls": { "install_dependency": profile.install_dependency, "require_sandbox": profile.safety.require_sandbox, "require_checkpoints": profile.safety.require_checkpoints, "allow_unsafe_tools": profile.safety.allow_unsafe_tools, }, ``` The fields `require_human_approval`, `allowed_skill_categories`, `max_cost_per_plan`, `max_retries_per_step`, and `max_total_cost` are entirely absent from the output. This means any consumer of `agents automation-profile show --format json` (scripts, CI pipelines, other agents) receives an incomplete response that is missing 5 of the 8 spec-required `SafetyProfile` fields. **Feature Area:** Sandbox & Checkpoint Safety Model **Spec Reference:** `docs/specification.md` lines 28513–28522 (SafetyProfile definition) and line 28564 (built-in profile table) ## Expected Behavior `agents automation-profile show <NAME>` displays all 8 `SafetyProfile` fields in the `execution_controls` section, including `require_human_approval`, `allowed_skill_categories`, `max_cost_per_plan`, `max_retries_per_step`, and `max_total_cost`. Example expected JSON output: ```json { "execution_controls": { "install_dependency": false, "require_sandbox": true, "require_checkpoints": true, "allow_unsafe_tools": false, "require_human_approval": true, "allowed_skill_categories": ["read", "write"], "max_cost_per_plan": 1.0, "max_retries_per_step": 3, "max_total_cost": 10.0 } } ``` ## Acceptance Criteria - [ ] `agents automation-profile show <NAME> --format json` includes all 8 `SafetyProfile` fields in `execution_controls`: `require_sandbox`, `require_checkpoints`, `allow_unsafe_tools`, `require_human_approval`, `allowed_skill_categories`, `max_cost_per_plan`, `max_retries_per_step`, `max_total_cost` - [ ] `agents automation-profile show <NAME> --format yaml` includes the same 8 fields - [ ] The Rich output panel (`_print_profile()`) also displays all 8 fields - [ ] All BDD tests pass with no regressions in other `automation-profile` commands ## Subtasks - [ ] Add `require_human_approval` to `execution_controls` in `_profile_spec_dict()` - [ ] Add `allowed_skill_categories` to `execution_controls` in `_profile_spec_dict()` - [ ] Add `max_cost_per_plan` to `execution_controls` in `_profile_spec_dict()` - [ ] Add `max_retries_per_step` to `execution_controls` in `_profile_spec_dict()` - [ ] Add `max_total_cost` to `execution_controls` in `_profile_spec_dict()` - [ ] Update the Rich panel renderer `_print_profile()` to display the same fields - [ ] Update BDD tests to assert all 8 `SafetyProfile` fields appear in output ## Definition of Done - `agents automation-profile show <NAME> --format json` includes all 8 `SafetyProfile` fields in `execution_controls` - Rich output panel also shows all 8 fields - All BDD tests pass - No regressions in other `automation-profile` commands --- **Automated by CleverAgents Bot** Agent: new-issue-creator Supervisor: UAT Test Pool | Agent: uat-test-pool-supervisor

HAL9000 added the

labels

2026-04-14 08:41:21 +00:00

HAL9000 added a new dependency

2026-04-14 08:44:57 +00:00

#8569 Epic: Cost Budgets, Safety Profiles & Plugin Architecture (v3.6.0)

HAL9000 added the

MoSCoW

Must have

label

2026-04-14 08:50:10 +00:00

HAL9000 added this to the v3.5.0 milestone

2026-04-14 08:50:10 +00:00

HAL9000 added

and removed

labels

2026-04-14 12:53:30 +00:00

HAL9000 commented

2026-04-14 12:53:31 +00:00

Author

Owner

✅ Triage: Verified [AUTO-OWNR-1]

Valid bug: automation-profile show omits require_human_approval and allowed_skill_categories from execution_controls output. These are important safety fields that users need to see.

Assigning to v3.5.0 (Autonomy Hardening). Priority Medium — missing specific fields in output.

MoSCoW: Must Have — require_human_approval is a critical safety field that must be visible in profile output.

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

✅ **Triage: Verified** [AUTO-OWNR-1] Valid bug: `automation-profile show` omits `require_human_approval` and `allowed_skill_categories` from `execution_controls` output. These are important safety fields that users need to see. Assigning to **v3.5.0** (Autonomy Hardening). Priority **Medium** — missing specific fields in output. MoSCoW: **Must Have** — `require_human_approval` is a critical safety field that must be visible in profile output. --- **Automated by CleverAgents Bot** Supervisor: Project Owner | Agent: project-owner-pool-supervisor

HAL9000 referenced this issue

2026-04-14 12:56:58 +00:00

[AUTO-PROJ-OWN] Status: Project Owner Pool Supervisor (Cycle 1) #9249