cleveragents/cleveragents-core

Fork 3

refactor(autonomy): rename automation profile task flags to match specification field names #902

New issue

Closed

opened 2026-03-13 23:47:58 +00:00 by freemo · 4 comments

freemo commented

2026-03-13 23:47:58 +00:00

Owner

Metadata

Commit Message: refactor(autonomy): rename automation profile task flags to spec names
Branch: feature/m5-automation-profile-fields

Background and Context

The specification defines 11 task-type confidence threshold flags on each automation profile. The code implements 11 threshold fields but with completely different names — using phase-transition semantics instead of the spec's task-type semantics.

#	Spec field name	Code field name (actual)
1	`decompose_task`	`auto_strategize`
2	`create_tool`	`auto_execute`
3	`select_tool`	`auto_apply`
4	`edit_code`	`auto_decisions_strategize`
5	`execute_command`	`auto_decisions_execute`
6	`create_file`	`auto_validation_fix`
7	`delete_content`	`auto_strategy_revision`
8	`access_network`	`auto_reversion_from_apply`
9	`install_dependency`	`auto_child_plans`
10	`modify_config`	`auto_retry_transient`
11	`approve_plan`	`auto_checkpoint_restore`

The SafetyProfile sub-model matches the spec exactly (all 8 fields).

The code's phase-transition approach (auto_strategize, auto_execute, auto_apply) may be a valid implementation strategy, but the field names must align with the spec for YAML config compatibility and documentation accuracy. All 8 built-in profiles (manual, review, supervised, cautious, trusted, auto, ci, full-auto) need their threshold values remapped.

Spec reference: Automation Profile section, ~lines 28216-28432

Acceptance Criteria

All 11 task flag field names match the specification exactly
All 8 built-in profiles have correct threshold values under the new field names
YAML automation profile configs use the spec field names
agents automation-profile show displays the spec field names
Profile resolution logic (plan > action > project > global) still works correctly
AutonomyController references updated field names
Database migration (if persisted) renames the columns

Subtasks

Rename all 11 fields in the AutomationProfile Pydantic model
Update all 8 built-in profile definitions with correct field names and values
Update YAML config schema for automation profiles
Update automation_profile.py CLI output formatting
Update AutonomyController / any service that references these fields
Update Alembic migration if columns are persisted
Tests (Behave): Verify profiles load correctly with new names
Verify coverage >= 97% via nox -s coverage_report
Run nox (all default sessions), fix any errors

Definition of Done

This issue is complete when:

All subtasks above are completed and checked off.
A Git commit is created where the first line of the commit message matches the Commit Message in Metadata exactly, followed by a blank line, then additional lines providing relevant details about the implementation.
The commit is pushed to the remote on the branch matching the Branch in Metadata exactly.
The commit is submitted as a pull request to master, reviewed, and merged before this issue is marked done.

## Metadata - **Commit Message**: `refactor(autonomy): rename automation profile task flags to spec names` - **Branch**: `feature/m5-automation-profile-fields` ## Background and Context The specification defines 11 task-type confidence threshold flags on each automation profile. The code implements 11 threshold fields but with **completely different names** — using phase-transition semantics instead of the spec's task-type semantics. | # | Spec field name | Code field name (actual) | |---|---|---| | 1 | `decompose_task` | `auto_strategize` | | 2 | `create_tool` | `auto_execute` | | 3 | `select_tool` | `auto_apply` | | 4 | `edit_code` | `auto_decisions_strategize` | | 5 | `execute_command` | `auto_decisions_execute` | | 6 | `create_file` | `auto_validation_fix` | | 7 | `delete_content` | `auto_strategy_revision` | | 8 | `access_network` | `auto_reversion_from_apply` | | 9 | `install_dependency` | `auto_child_plans` | | 10 | `modify_config` | `auto_retry_transient` | | 11 | `approve_plan` | `auto_checkpoint_restore` | The `SafetyProfile` sub-model matches the spec exactly (all 8 fields). The code's phase-transition approach (auto_strategize, auto_execute, auto_apply) may be a valid implementation strategy, but the field names must align with the spec for YAML config compatibility and documentation accuracy. All 8 built-in profiles (manual, review, supervised, cautious, trusted, auto, ci, full-auto) need their threshold values remapped. **Spec reference**: Automation Profile section, ~lines 28216-28432 ## Acceptance Criteria - [ ] All 11 task flag field names match the specification exactly - [ ] All 8 built-in profiles have correct threshold values under the new field names - [ ] YAML automation profile configs use the spec field names - [ ] `agents automation-profile show` displays the spec field names - [ ] Profile resolution logic (plan > action > project > global) still works correctly - [ ] AutonomyController references updated field names - [ ] Database migration (if persisted) renames the columns ## Subtasks - [ ] Rename all 11 fields in the `AutomationProfile` Pydantic model - [ ] Update all 8 built-in profile definitions with correct field names and values - [ ] Update YAML config schema for automation profiles - [ ] Update `automation_profile.py` CLI output formatting - [ ] Update `AutonomyController` / any service that references these fields - [ ] Update Alembic migration if columns are persisted - [ ] Tests (Behave): Verify profiles load correctly with new names - [ ] Verify coverage >= 97% via `nox -s coverage_report` - [ ] Run `nox` (all default sessions), fix any errors ## Definition of Done This issue is complete when: - All subtasks above are completed and checked off. - A Git commit is created where the **first line** of the commit message matches the Commit Message in Metadata exactly, followed by a blank line, then additional lines providing relevant details about the implementation. - The commit is pushed to the remote on the branch matching the **Branch** in Metadata exactly. - The commit is submitted as a **pull request** to `master`, reviewed, and **merged** before this issue is marked done.

freemo added the

labels

2026-03-13 23:48:36 +00:00

freemo added this to the v3.5.0 milestone

2026-03-13 23:48:40 +00:00

CoreRasurae was assigned by freemo

2026-03-14 04:27:12 +00:00

CoreRasurae referenced this issue from a commit

2026-03-24 03:43:04 +00:00

refactor(autonomy): rename automation profile task flags to spec names

~~CoreRasurae referenced this issue 2026-03-24 03:43:30 +00:00~~

refactor(autonomy): rename automation profile task flags to spec names #1147

CoreRasurae commented

2026-03-24 03:44:35 +00:00

Member

Implementation Notes

Summary

Renamed all 11 task-type confidence threshold fields in the AutomationProfile model from phase-transition semantics to the specification's task-type semantics. All 8 built-in profiles, CLI formatting, YAML schema, services, Alembic migration, and all Behave/Robot tests updated.

Files Modified (42 files)

Core: automation_profile.py, escalation.py, autonomy_controller.py, plan_lifecycle_service.py, CLI commands
Infrastructure: models.py (ORM columns), repositories.py (domain mapping)
Schema/Config: automation_profile.schema.yaml, all 9 YAML example profiles
Alembic: New migration m5_001_rename_automation_profile_fields.py
Tests: 15 Behave test files, 5 Robot test files, 3 benchmark files

Design Decisions

Kept original migration intact: a6_001_automation_profiles retains old column names; new m5_001 migration performs the rename using batch_alter_table.
Avoided false positives: Carefully avoided renaming auto_apply in unrelated contexts (ProjectSettings.auto_apply, PlanConfig.auto_apply, MigrationRunner.auto_apply) which use the same name but refer to different boolean settings.
Preserved threshold values: All 8 built-in profiles retain exact same numerical threshold values — only field names changed.
Updated CLI section headers: Changed from phase-transition groupings to unified "Task-Type Confidence Thresholds" section.

Quality Gate Results

Stage	Result
lint	PASSED
typecheck	PASSED (0 errors)
unit_tests	PASSED (462 features, 12,288 scenarios)
integration_tests	PASSED (1,699 tests)
coverage	98% (exceeds 97% threshold)

PR

PR #1147 — feature/m5-automation-profile-fields → master
Commit: 2d03b809

## Implementation Notes ### Summary Renamed all 11 task-type confidence threshold fields in the `AutomationProfile` model from phase-transition semantics to the specification's task-type semantics. All 8 built-in profiles, CLI formatting, YAML schema, services, Alembic migration, and all Behave/Robot tests updated. ### Files Modified (42 files) - **Core**: `automation_profile.py`, `escalation.py`, `autonomy_controller.py`, `plan_lifecycle_service.py`, CLI commands - **Infrastructure**: `models.py` (ORM columns), `repositories.py` (domain mapping) - **Schema/Config**: `automation_profile.schema.yaml`, all 9 YAML example profiles - **Alembic**: New migration `m5_001_rename_automation_profile_fields.py` - **Tests**: 15 Behave test files, 5 Robot test files, 3 benchmark files ### Design Decisions 1. **Kept original migration intact**: `a6_001_automation_profiles` retains old column names; new `m5_001` migration performs the rename using `batch_alter_table`. 2. **Avoided false positives**: Carefully avoided renaming `auto_apply` in unrelated contexts (`ProjectSettings.auto_apply`, `PlanConfig.auto_apply`, `MigrationRunner.auto_apply`) which use the same name but refer to different boolean settings. 3. **Preserved threshold values**: All 8 built-in profiles retain exact same numerical threshold values — only field names changed. 4. **Updated CLI section headers**: Changed from phase-transition groupings to unified "Task-Type Confidence Thresholds" section. ### Quality Gate Results | Stage | Result | |---|---| | lint | PASSED | | typecheck | PASSED (0 errors) | | unit_tests | PASSED (462 features, 12,288 scenarios) | | integration_tests | PASSED (1,699 tests) | | coverage | **98%** (exceeds 97% threshold) | ### PR PR #1147 — `feature/m5-automation-profile-fields` → `master` Commit: `2d03b809`

CoreRasurae added

and removed

labels

2026-03-24 03:45:05 +00:00

CoreRasurae referenced this issue from a pull request that will close it,

2026-03-24 03:46:39 +00:00

refactor(autonomy): rename automation profile task flags to spec names #1147

CoreRasurae commented

2026-03-27 10:25:14 +00:00

Member

Applied review-driven fixes for #902 on feature/m5-automation-profile-fields and validated with nox.

What was updated:

Hardened AutomationProfile.from_config unknown-field handling so legacy top-level keys are rejected instead of silently defaulting.
Added additionalProperties: false to docs/schema/automation_profile.schema.yaml for top-level profile validation consistency.
Added BDD regression coverage for legacy threshold keys in automation-profile add config handling.
Aligned docs/spec references from legacy auto_* threshold wording to the renamed task-type threshold fields (where relevant to automation profile thresholds).
Updated CHANGELOG.md with an Unreleased entry for these fixes.

Validation (per repo constraints):

TEST_PROCESSES=9 nox -s lint ✅
TEST_PROCESSES=9 nox -s typecheck ✅
TEST_PROCESSES=9 nox -s unit_tests ✅
TEST_PROCESSES=9 nox -s integration_tests ✅

Notes:

Coverage sessions were intentionally not run in this pass.
No new commit was created in this step.

Applied review-driven fixes for #902 on `feature/m5-automation-profile-fields` and validated with nox. What was updated: - Hardened `AutomationProfile.from_config` unknown-field handling so legacy top-level keys are rejected instead of silently defaulting. - Added `additionalProperties: false` to `docs/schema/automation_profile.schema.yaml` for top-level profile validation consistency. - Added BDD regression coverage for legacy threshold keys in `automation-profile add` config handling. - Aligned docs/spec references from legacy `auto_*` threshold wording to the renamed task-type threshold fields (where relevant to automation profile thresholds). - Updated `CHANGELOG.md` with an Unreleased entry for these fixes. Validation (per repo constraints): - `TEST_PROCESSES=9 nox -s lint` ✅ - `TEST_PROCESSES=9 nox -s typecheck` ✅ - `TEST_PROCESSES=9 nox -s unit_tests` ✅ - `TEST_PROCESSES=9 nox -s integration_tests` ✅ Notes: - Coverage sessions were intentionally not run in this pass. - No new commit was created in this step.

CoreRasurae commented

2026-03-27 15:01:43 +00:00

Member

Applied a follow-up validation/fix pass for #902 on local branch feature/m5-automation-profile-fields and cross-checked against docs/specification.md.

Validated review findings and outcomes:

Docs schema/runtime parity for guards — valid, fixed.
- Added top-level guards object schema to docs/schema/automation_profile.schema.yaml (while preserving additionalProperties: false at top level).
Missing regression guard for schema/runtime parity on guards — valid, fixed.
- Added BDD scenario and step coverage:
  - features/automation_profile_cli_coverage.feature
  - features/steps/automation_profile_cli_coverage_steps.py
- New scenario validates the same guarded profile config against both docs schema (jsonschema) and runtime model (AutomationProfile.from_config).
Migration header consistency — valid, fixed.
- Aligned migration docstring header Revises: with actual metadata in:
  - alembic/versions/m5_001_rename_automation_profile_fields.py

Documentation and changelog updates:

Updated CHANGELOG.md Unreleased entry for #902 to include these follow-up fixes.

Quality gates run (as requested, no coverage session):

TEST_PROCESSES=9 nox -s lint ✅
TEST_PROCESSES=9 nox -s typecheck ✅
TEST_PROCESSES=9 nox -s unit_tests ✅
TEST_PROCESSES=9 nox -s integration_tests ✅

No reviewed items were deferred in this pass; all validated findings were applied without conflicting with issue scope or specification intent.

Applied a follow-up validation/fix pass for #902 on local branch `feature/m5-automation-profile-fields` and cross-checked against `docs/specification.md`. Validated review findings and outcomes: 1. **Docs schema/runtime parity for `guards`** — **valid**, fixed. - Added top-level `guards` object schema to `docs/schema/automation_profile.schema.yaml` (while preserving `additionalProperties: false` at top level). 2. **Missing regression guard for schema/runtime parity on `guards`** — **valid**, fixed. - Added BDD scenario and step coverage: - `features/automation_profile_cli_coverage.feature` - `features/steps/automation_profile_cli_coverage_steps.py` - New scenario validates the same guarded profile config against both docs schema (`jsonschema`) and runtime model (`AutomationProfile.from_config`). 3. **Migration header consistency** — **valid**, fixed. - Aligned migration docstring header `Revises:` with actual metadata in: - `alembic/versions/m5_001_rename_automation_profile_fields.py` Documentation and changelog updates: - Updated `CHANGELOG.md` Unreleased entry for #902 to include these follow-up fixes. Quality gates run (as requested, no coverage session): - `TEST_PROCESSES=9 nox -s lint` ✅ - `TEST_PROCESSES=9 nox -s typecheck` ✅ - `TEST_PROCESSES=9 nox -s unit_tests` ✅ - `TEST_PROCESSES=9 nox -s integration_tests` ✅ No reviewed items were deferred in this pass; all validated findings were applied without conflicting with issue scope or specification intent.

CoreRasurae referenced this issue from a commit

2026-03-27 15:39:57 +00:00

refactor(autonomy): rename automation profile task flags to spec names

CoreRasurae referenced this issue

2026-03-27 17:30:19 +00:00

refactor(autonomy): rename automation profile task flags to spec names #1147

CoreRasurae referenced this issue from a commit

2026-03-27 17:59:22 +00:00

refactor(autonomy): rename automation profile task flags to spec names

CoreRasurae commented

2026-03-27 22:39:24 +00:00

Member

Post-Review Fix Pass (Review #2)

Applied fixes from the code review on PR #1147, validated against docs/specification.md and issue scope.

Fixes Applied

ID	Severity	Fix
M-1	Medium	Restored categorised CLI `automation-profile show` output (Phase Transitions / Decision Automation / Self-Repair / Execution Controls) to match specification output examples. The prior commit had flattened all thresholds into a single "Task-Type Confidence Thresholds" section.
M-2	Medium	Added missing `access_network` field to specification `show` output examples (Rich, Plain, JSON, YAML variants). This was pre-existing but within scope since the output sections were already being modified.
M-4	Medium	Added 4 missing fields (`create_file`, `install_dependency`, `modify_config`, `approve_plan`) to ADR-017 profile fields table, bringing it to all 11 spec-defined fields.
M-5	Medium	Extended repository roundtrip test (`repositories_uncovered_lines_steps.py`) to assert all 11 threshold fields after upsert (was only asserting 3).
H-4	High	Aligned `automation_profiles.md` threshold descriptions with the specification's Automatable Tasks table (§28327–28339).
H-5	High	Added spec section references (`§ Automatable Tasks`) in `phase_reversion.md`, `error_recovery.md`, and `plan_execute.md` to provide field naming context.

Findings Not Applied (with justification)

ID	Severity	Reason
H-2, H-3	High	The semantic confusion in ADR-006 and ADR-017 (task-type names describing phase-transition behaviors) is inherent in the specification's design — the spec explicitly maps these names to those behaviors. Per CONTRIBUTING.md "Specification-First Development", the code/docs align with the spec. The ADR descriptions are correct.
M-3	Medium	Schema descriptions already match the spec's "Description" column. Adding runtime behavior to the schema would make it verbose and diverge from the spec format.
L-1	Low	Test-only: legacy validator tests 3/11 names but the dict-lookup detection logic is exercised by any single key. Low risk.
L-2	Low	Test-only: CI profile values assumed but not asserted. Does not affect correctness.
L-5	Low	ADR-007 text matches the spec's description of `edit_code`. No change needed.

Quality Gates

TEST_PROCESSES=9 nox -s lint — PASS
TEST_PROCESSES=9 nox -s typecheck — PASS (0 errors)
TEST_PROCESSES=9 nox -s unit_tests — PASS (463 features, 12,293 scenarios)
TEST_PROCESSES=9 nox -s integration_tests — PASS (1,699 tests)

Commit

Amended commit 4dbb6ef9 on feature/m5-automation-profile-fields.

## Post-Review Fix Pass (Review #2) Applied fixes from the code review on PR #1147, validated against `docs/specification.md` and issue scope. ### Fixes Applied | ID | Severity | Fix | |----|----------|-----| | **M-1** | Medium | Restored categorised CLI `automation-profile show` output (Phase Transitions / Decision Automation / Self-Repair / Execution Controls) to match specification output examples. The prior commit had flattened all thresholds into a single "Task-Type Confidence Thresholds" section. | | **M-2** | Medium | Added missing `access_network` field to specification `show` output examples (Rich, Plain, JSON, YAML variants). This was pre-existing but within scope since the output sections were already being modified. | | **M-4** | Medium | Added 4 missing fields (`create_file`, `install_dependency`, `modify_config`, `approve_plan`) to ADR-017 profile fields table, bringing it to all 11 spec-defined fields. | | **M-5** | Medium | Extended repository roundtrip test (`repositories_uncovered_lines_steps.py`) to assert all 11 threshold fields after upsert (was only asserting 3). | | **H-4** | High | Aligned `automation_profiles.md` threshold descriptions with the specification's Automatable Tasks table (§28327–28339). | | **H-5** | High | Added spec section references (`§ Automatable Tasks`) in `phase_reversion.md`, `error_recovery.md`, and `plan_execute.md` to provide field naming context. | ### Findings Not Applied (with justification) | ID | Severity | Reason | |----|----------|--------| | **H-2, H-3** | High | The semantic confusion in ADR-006 and ADR-017 (task-type names describing phase-transition behaviors) is inherent in the specification's design — the spec explicitly maps these names to those behaviors. Per CONTRIBUTING.md "Specification-First Development", the code/docs align with the spec. The ADR descriptions are correct. | | **M-3** | Medium | Schema descriptions already match the spec's "Description" column. Adding runtime behavior to the schema would make it verbose and diverge from the spec format. | | **L-1** | Low | Test-only: legacy validator tests 3/11 names but the dict-lookup detection logic is exercised by any single key. Low risk. | | **L-2** | Low | Test-only: CI profile values assumed but not asserted. Does not affect correctness. | | **L-5** | Low | ADR-007 text matches the spec's description of `edit_code`. No change needed. | ### Quality Gates - `TEST_PROCESSES=9 nox -s lint` — PASS - `TEST_PROCESSES=9 nox -s typecheck` — PASS (0 errors) - `TEST_PROCESSES=9 nox -s unit_tests` — PASS (463 features, 12,293 scenarios) - `TEST_PROCESSES=9 nox -s integration_tests` — PASS (1,699 tests) ### Commit Amended commit `4dbb6ef9` on `feature/m5-automation-profile-fields`.

CoreRasurae referenced this issue from a commit

2026-03-27 22:46:13 +00:00

refactor(autonomy): rename automation profile task flags to spec names

CoreRasurae referenced this issue

2026-03-27 23:01:48 +00:00

refactor(autonomy): rename automation profile task flags to spec names #1147

CoreRasurae referenced this issue

2026-03-27 23:21:11 +00:00

refactor(autonomy): rename automation profile task flags to spec names #1147

CoreRasurae referenced this issue from a commit

2026-03-27 23:21:47 +00:00

refactor(autonomy): rename automation profile task flags to spec names