Implement inline permission question widget for quick allow/reject
decisions within the conversation stream with file diff context.
ISSUES CLOSED: #997
Introduce DomainBaseModel in src/cleveragents/domain/models/base.py that
defines the single shared model_config (str_strip_whitespace, validate_assignment,
arbitrary_types_allowed=False, populate_by_name, use_enum_values) previously
duplicated verbatim across five domain model files.
Update 14 classes across five files to inherit from DomainBaseModel instead
of pydantic.BaseModel directly, removing all inline model_config duplication:
- aimodelscredentials/ai_models_credentials.py (ModelProviderOption)
- aimodelserrors/ai_models_errors.py (ModelError, FallbackResult)
- aimodelsproviders/ai_models_providers.py (ModelProviderExtraAuthVars,
ModelProviderConfigSchema)
- auth/auth.py (AuthHeader, TrialPlansExceededError, TrialMessagesExceededError,
BillingError, ApiError, ClientAccount, ClientAuth)
- planconfig/plan_config.py (PlanConfig, ConfigSetting)
Pure structural refactor — no behavioral changes. Models with different
configurations (acms, aimodels_custom, etc.) are left untouched.
Add BDD feature (22 scenarios) and Robot integration tests (8 test cases)
verifying inheritance, config correctness, and no-duplication invariants.
ISSUES CLOSED: #1941
Centralize E2E initialization in common suite setup so DB-dependent CLI commands no longer rely on per-suite or per-test init workarounds. This also sanitizes suite-home names to prevent invalid path-derived initialization failures in isolated Robot runs.
ISSUES CLOSED: #1023
Rewrites the A2aRequest and A2aResponse Pydantic models to use the field
names mandated by the JSON-RPC 2.0 specification, fixing a fundamental
protocol compliance issue that prevented external A2A-compliant clients
from communicating with the server.
Changes:
- A2aRequest: a2a_version→jsonrpc (fixed '2.0'), request_id→id,
operation→method; auth field removed (not in JSON-RPC 2.0)
- A2aResponse: a2a_version→jsonrpc, request_id→id, status+data→result
(success path), timing_ms removed; added _result_xor_error validator
enforcing mutual exclusion of result and error fields
- A2aLocalFacade.dispatch(): updated to use request.method, request.id,
result=data, error=A2aErrorDetail(...)
- A2aHttpTransport.send(): updated to use request.method
- CLI call sites (session.py, plan.py): updated A2aRequest(method=...)
and response.result / response.error field access
- All existing A2A Behave step files updated to new field names
- New 35-scenario Behave feature (a2a_jsonrpc_wire_format.feature)
covering serialisation, deserialisation, validation, and facade dispatch
- New 7-test Robot Framework suite (a2a_jsonrpc_wire_format.robot)
for end-to-end wire format verification
ISSUES CLOSED: #1501
Extended `_print_actor()` to render the full spec-required output for
`agents actor add`: the Actor Added panel now includes a Type field, and
three additional panels (Config, Capabilities, Tools) plus a success
status line are rendered when `show_add_panels=True`.
- Add `Type:` field to the Actor Added panel (from `config_blob["type"]`)
- Implement Config panel: Path, Hash, Options count, Nodes count, Edges count
- Implement Capabilities panel: bulleted list (rendered only when non-empty)
- Implement Tools panel: Rich table with Tool, Read-Only, Safe columns
(rendered only when non-empty; string tool entries default to yes/yes)
- Add `✓ OK Actor added` success status line after all panels
- Pass `config_path` and `show_add_panels=True` from `add()` command
- Add Behave BDD scenarios covering each new panel and the success line
- Add Robot Framework integration test verifying the full rich output
ISSUES CLOSED: #1499
Add missing provider: field to all actor examples in examples/actors/.
Fix llm_with_tools.yaml actor name from assistants/file_analyzer to
local/assistants-file_analyzer (custom actors must use local/ namespace
and cannot contain two slashes).
Add validate-all command to helper_actor_examples.py that uses ActorLoader
to validate all examples via business logic checks. Add integration test
Validate All Actor Examples Import Without Errors to actor_examples.robot
that confirms every example in examples/actors/ can be imported without
errors.
ISSUES CLOSED: #1504
Renames `plan lifecycle-list` to `plan list` and `plan lifecycle-apply` to `plan apply` to align with the specification's canonical command names. Removes legacy V2 plan commands that occupied those names.
- Renamed CLI command registrations from lifecycle-list/lifecycle-apply to list/apply
- Removed legacy V2 apply and list commands (~200 lines)
- Updated apply shortcut in main.py to delegate to v3 lifecycle
- Added defensive null check for plan existence in apply command
- Updated 63+ test, doc, and benchmark files for consistency
Closes#881
Co-authored-by: Jeffrey Phillips Freeman <the@jeffreyfreeman.me>
Co-committed-by: Jeffrey Phillips Freeman <the@jeffreyfreeman.me>
Remove the --mode/-m CLI flag from the validation attach command to align
with the specification, which defines validation mode as an inherent
property of the validation definition set at registration time via
validation add, not as a per-attachment override.
The service layer (ToolRegistryService.attach_validation) no longer
accepts mode as a parameter; instead it reads the mode from the
validation's registered definition. The ToolRegistryRepository
_to_legacy_domain now exposes the mode field so the service can access it.
All tests that passed --mode to the CLI or service have been updated or
removed. The invalid-mode validation scenarios were removed since the
mode is no longer caller-supplied. Coverage remains at 98.7%.
ISSUES CLOSED: #913
Co-authored-by: Jeffrey Phillips Freeman <jeffrey.freeman@cleverthis.com>
Co-committed-by: Jeffrey Phillips Freeman <jeffrey.freeman@cleverthis.com>
Remove three top-level fields from CorrectionDryRunReport that duplicated data already present in the embedded CorrectionImpact object:
- excluded_decisions → now accessed via report.impact.excluded_decisions
- rollback_tier_depth → now accessed via report.impact.rollback_tier_depth
- child_plans_to_rollback → now accessed via report.impact.affected_child_plans
The redundant fields created a divergence risk: both models are frozen=True Pydantic models, so post-construction mutation is impossible, but nothing prevented constructing an instance where the top-level copies disagreed with the impact sub-object.
Approach: Option B (nest-only) — remove the duplicated top-level fields and update all consumers to use the impact object's fields instead.
Closes#1087
Co-authored-by: Jeffrey Phillips Freeman <the@jeffreyfreeman.me>
Co-committed-by: Jeffrey Phillips Freeman <the@jeffreyfreeman.me>
Remove four empty stub packages that contained only __init__.py with
__all__ = [] and no functional code, violating CONTRIBUTING.md §Commit
Completeness. All four were audited against docs/specification.md:
- src/cleveragents/runtime/ — Spec defines four layers (Domain,
Application, Infrastructure, Presentation) with no Runtime Layer.
Runtime concepts (LSP Runtime, actor runtime) belong in Infrastructure.
- src/cleveragents/domain/repositories/ — Spec mentions repository
interfaces in Domain but mandates no separate subpackage. Empty with
no protocols defined; existing models cover the domain.
- src/cleveragents/domain/plans/ — Plan domain models already exist in
domain/models/planconfig, planfiles, and plansettings. Empty duplicate.
- src/cleveragents/application/workflows/ — Application logic already
served by application/services/. Empty stub added no value.
Updated test references in features/steps/module_coverage_steps.py,
features/steps/coverage_extras_steps.py, features/architecture.feature,
and robot/architecture.robot to remove the deleted packages from import
verification and architecture validation lists.
ISSUES CLOSED: #948
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Robot Framework integration test for Specification Workflow Example 8:
Cloud Infrastructure Management with the supervised automation profile.
Test suite exercises:
- Custom resource type registration (local/terraform-state) via YAML
fixture with copy_on_write sandbox strategy and CLI args validation
- Custom skill creation (local/terraform-ops) with 3 anonymous tools
(terraform_plan, terraform_show, cloud_metrics) and skill composition
via includes (local/file-ops)
- Supervised profile gating: verifies both strategize-to-execute
(create_tool=1.0) and execute-to-apply (select_tool=1.0) transitions
require explicit human approval via should_auto_progress()
- Action creation with supervised profile, typed arguments (STRING,
FLOAT), and invariant propagation to plans via InvariantSource.ACTION
- Mocked infrastructure analysis producing right-sizing recommendations
for over-provisioned resources while skipping critical instances
- Invariant enforcement blocking modifications to resources tagged
critical:true while allowing non-critical changes
Fixture files provide deterministic mock Terraform state (3 AWS
resources), CloudWatch metrics (CPU at 12% avg on m5.2xlarge), and
sample HCL configuration.
ISSUES CLOSED: #772
Reviewed-by: freemo (reviewer-pool-1)
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Add Robot Framework integration test suite exercising Specification Workflow Example 6: documentation generation from codebase analysis using the trusted automation profile.
The test suite (6 test cases) validates:
- Context policy configuration with strategize/execute view-specific settings and view inheritance resolution
- Budget enforcement filtering ContextFragments by max_file_size and max_total_size limits
- Trusted profile auto-progress behavior (create_tool=0.0 triggers automatic Strategize→Execute; select_tool=1.0 gates Apply)
- Action creation with doc_types/output_dir arguments and 3 invariants matching the spec scenario
- Temp project sandbox with documentation generation into output directory and source-code modification invariant verification
- Full plan lifecycle through trusted profile with argument and invariant preservation
ISSUES CLOSED: #770
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
The CLI `session create` command created the session via SessionService.create()
(which commits via auto_commit=True), then called _facade_dispatch("session.create")
for A2A protocol bookkeeping. The facade handler unconditionally called
svc.create() on a second PersistentSessionService instance (a new Factory
resolution from the DI container with its own engine), creating a duplicate
session in the database.
The fix makes A2aLocalFacade._handle_session_create() idempotent: when a
session_id is already present in the params, it acknowledges the existing
session without creating a new one. The CLI dispatch params were also fixed
to use the correct key name (actor_name instead of actor).
Changes:
- src/cleveragents/a2a/facade.py: Early return in _handle_session_create when
session_id is already supplied, preventing duplicate session creation.
- src/cleveragents/cli/commands/session.py: Fixed param key from "actor" to
"actor_name" for consistency with the facade handler.
- features/a2a_facade_wiring.feature: Added idempotency scenario verifying
that session.create with an existing session_id does not call svc.create().
- features/steps/a2a_facade_wiring_steps.py: Added step asserting mock
SessionService.create was not called.
- features/tdd_session_create_persist.feature: Removed @tdd_expected_fail tag
now that the bug is fixed.
- robot/e2e/e2e_session_create_persist.robot: Removed tdd_expected_fail tag,
updated documentation.
- .semgrep.yml: Excluded wrapping.py from no-exec/no-compile-exec rules
(pre-existing sandboxed exec usage for tool transforms).
ISSUES CLOSED: #1141
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Add Behave BDD scenarios and Robot Framework integration tests that capture bug #1152: _enforce_hot_budget() permanently deletes hot-tier fragments via del self._hot[oldest_id] instead of demoting them to the warm tier. Similarly, evict_lru() permanently deletes via del store[fid].
The specification (§Plan Lifecycle ACMS Actions) describes a downward lifecycle ("Hot context archived to warm") that requires demotion, not destruction.
Two Behave scenarios tagged @tdd_expected_fail @tdd_issue @tdd_issue_1152:
- Budget-evicted fragment should be demoted to warm, not deleted
- evict_lru-evicted fragment should be demoted to warm, not deleted
Two Robot Framework tests mirror the Behave scenarios with a Python helper script for subprocess isolation.
ISSUES CLOSED: #1183
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Remove @tdd_expected_fail tags from test files for bugs #797 (actor list
triggers DB update) and #1028 (ACMS indexing pipeline not wired), both of
which are now closed and merged. The @tdd_bug and @tdd_bug_N tags remain
as permanent regression guards per CONTRIBUTING.md TDD workflow rules.
Files changed:
- features/tdd_actor_list_no_db_update.feature
- robot/tdd_actor_list_no_db_update.robot
- robot/e2e/tdd_acms_behavioral_validation.robot
ISSUES CLOSED: #1182
Add a dedicated TUI help overlay toggled by F1 and resolve its content from the current prompt mode for main-screen, slash-command, reference, and shell contexts.
Extend Behave and Robot coverage for the new help-panel widget, app wiring, context switching, and toggle behavior.
ISSUES CLOSED: #1013
Add content_hash(resource, *, algorithm='sha256') -> str to the
ResourceHandler protocol and all handler implementations:
- Protocol: new content_hash method on ResourceHandler (protocol.py)
- BaseResourceHandler: default impl hashes file content or directory
entry names; returns EMPTY_CONTENT_HASH sentinel for missing
resources
- GitCheckoutHandler: hashes git rev-parse HEAD through the requested
algorithm for consistent digest format
- FsDirectoryHandler: recursive walk hashing sorted relative paths
and file contents (content-only, ignores metadata)
- DevcontainerHandler: hashes devcontainer.json config file
- DatabaseResourceHandler: hashes connection string; for SQLite
file-based DBs, hashes the database file content
- _DefaultHandler: delegates to BaseResourceHandler
EMPTY_CONTENT_HASH sentinel is the SHA-256 of empty input
(e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855).
Hash algorithm is configurable via the algorithm parameter (default
sha256, accepts any hashlib.new()-compatible name).
Behave tests (10 scenarios): sentinel for missing/nonexistent,
determinism, different content produces different hash, fs-directory
recursive hash with change detection, git-checkout hash, configurable
algorithm (sha256 vs sha512), protocol compliance for all 4 handlers.
ISSUES CLOSED: #837
## Summary
- Fix automation profile resolution precedence at plan-creation time to follow plan > action > project > global.
- Narrow exception handling in `_resolve_plan_profile_ref` to catch only `(KeyError, ValueError, OSError)` with warning-level structured logging, replacing bare `except Exception:` per CONTRIBUTING.md §Exception Propagation.
- Fix TDD feature file tag/title inconsistency: align to `@tdd_bug @tdd_bug_1076` / "TDD Bug #1076" referencing the original bug ticket.
- Add BDD scenarios for plan-level profile override and config service error fallback.
- Harden lifecycle and Robot/Behave test paths for auto-progression and parallel CI stability (including pre-seeded DB initialization for `tdd_plan_explain_plan_id` helper).
- Update Robot helper timing/session patterns to reduce false negatives in long-running integration/E2E validations.
## Approach
The automation profile resolution is implemented in `PlanLifecycleService._resolve_plan_profile_ref`, which evaluates the four-level precedence chain (plan > action > project > global) at `use_action()` time. The resolved profile is stored as an `AutomationProfileRef` on the Plan with provenance tracking. The `ConfigService` is used for project-scoped and global config lookups, with narrowed exception handling that logs warnings on failure and falls back to the settings default.
## Validation
- `nox -s lint` ✅
- `nox -s typecheck` ✅
- `nox -s unit_tests` ✅
- `nox -s coverage_report` ✅ (97%)
## Review Fix Round (Review #2963)
1. **Bare `except Exception:` → narrowed to `(KeyError, ValueError, OSError)` with warning logging** — addresses CONTRIBUTING.md §Exception Propagation violation
2. **Tag inconsistency fixed** — `@tdd_bug @tdd_bug_1076` with title "TDD Bug #1076"
3. **Added 2 new BDD scenarios** — plan-level override and config error fallback
4. **Rebased onto latest master** (532ea100)
Closes#854
Reviewed-on: cleveragents/cleveragents-core#1196
Co-authored-by: Brent Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent Edwards <brent.edwards@cleverthis.com>
## Summary
Add M6 parallel-scaling coverage for 10+ concurrent subplans:
- **15-subplan parallel scenario** with explicit peak-concurrency bound checks (`max_parallel=10`) and thread-safe concurrency tracking via `_build_executor()`.
- **Deep hierarchical decomposition** coverage (4+ levels) with adjusted leaf condition that only stops early when hitting `max_depth` or when the workset is trivially small (`min_files_per_subplan`).
- **Non-progress guard** in `_build_hierarchy` to prevent pathological recursion when clustering cannot meaningfully split the file set.
- **Small-project regression test** (< 50 files) verifying decomposition depth does not increase unexpectedly with the relaxed leaf condition.
- **ASV benchmark** for 15-subplan parallel execution with `max_parallel=10` to track scaling behavior.
### Removed from this PR
The `_build_hierarchy` child-linkage correctness fix (returning `node_id` from recursive calls instead of using `nodes[-1].node_id`) has been **removed** per review feedback — it is a separate bug fix and will be submitted as an independent issue/PR per CONTRIBUTING.md §Atomic Commits.
## Approach
- **Concurrency tracking:** The `_build_executor()` closure in step definitions detects `context.concurrency_counter` / `context.concurrency_lock` and performs thread-safe peak tracking in a try/finally block.
- **Leaf condition:** Replaced the `max_files_per_subplan` / `max_tokens_per_subplan` leaf check with a `min_files_per_subplan` check to allow deeper decomposition for large projects. Added a non-progress guard so clustering that cannot split the file set terminates immediately rather than recursing to `max_depth`.
- **Deterministic IDs:** `_ids_for_count()` preserves legacy fixed IDs for the first 5 subplans and generates additional deterministic IDs for scale scenarios.
## Validation
### Passing
- `nox -s lint` — all checks passed
- `nox -s typecheck` — 0 errors, 0 warnings
- `nox -s unit_tests` — 12,988 scenarios passed, 0 failed
- `nox -s coverage_report` — 97% (passes `--fail-under=97`)
Closes#855
Reviewed-on: cleveragents/cleveragents-core#1201
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Robot Framework integration test suite for Specification Workflow Example 5:
Database Schema Migration with Safety Nets. Exercises the review automation
profile with custom resource types, custom skills with database tools,
phased child plan execution, checkpointing, and rollback.
- 7 test cases covering: custom resource type registration, review profile
behavior, custom skill with 3 database tools, action creation with
review profile, checkpoint/rollback via managers, 5-phase sequential
SubplanService.spawn with fail-fast, and plan lifecycle through
strategize-to-execute
- Uses mocked LLM providers (CLEVERAGENTS_TESTING_USE_MOCK_AI=true)
- All Run Process calls have timeout=60s on_timeout=kill
- Fix environment-dependent test failures: mock shutil.disk_usage in
disk space and diagnostics --check Behave steps so tests pass
regardless of CI runner disk space; Robot diagnostics test tolerates
disk-space errors
ISSUES CLOSED: #769
Add --required and --informational as mutually exclusive boolean options
to the `agents validation add` CLI command. When specified, they override
the `mode` field from the YAML config file. This aligns the CLI with the
specification (specification.md line 22339) which states the validation
mode can be set "via --required/--informational on agents validation add".
The fix resolves the spec contradiction by:
- Implementing the flags in the CLI (per Core Concepts > Validation Mode)
- Updating the formal CLI reference to include the new flags
- Adding an exception to Design Principle #3 for validation add
- Removing the spurious positional name argument from the walkthrough
TDD tests from #1102 now run normally with @tdd_expected_fail removed.
New edge-case tests cover mutual exclusivity (both flags rejected).
Also excludes tool/wrapping.py from semgrep exec/compile rules since
that module intentionally uses exec() in a controlled sandbox for the
validation transform feature.
ISSUES CLOSED: #1038
## Summary
- tighten `plan correct` active-plan fallback so it only runs for isolated `CLEVERAGENTS_HOME` mismatch cases and never when explicit DB env overrides are configured
- narrow fallback exception handling in `_resolve_active_plan_id()` to expected DB/path/service failures; unexpected errors now surface instead of being silently swallowed
- add BDD regression coverage for both safeguards in `features/consolidated_plan_misc.feature` + `features/steps/plan_cli_legacy_r2_steps.py`
## Validation
- `nox -e lint`: PASS
- `nox -e typecheck`: PASS
- `nox -e unit_tests`: PASS
- `nox -e integration_tests`: FAIL in current branch baseline (29 failing Robot integration tests in this environment)
- `nox -e e2e_tests`: FAIL in current branch baseline (45 failing E2E tests in this environment)
- `nox -e coverage_report`: PASS (97%)
Closes#1025
Reviewed-on: cleveragents/cleveragents-core#1184
Reviewed-by: Jeffrey Phillips Freeman <jeffrey.freeman@cleverthis.com>
Co-authored-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Co-committed-by: Brent E. Edwards <brent.edwards@cleverthis.com>
Aligned v3_plans table with specification DDL:
1. Added effective_profile_snapshot column (TEXT NOT NULL) for
storing frozen JSON snapshot of automation profile at plan
creation time. Added Pydantic field_validator ensuring the
value is well-formed JSON. Validator catches RecursionError
for deeply nested JSON, consistent with automation_profile
deserialization hardening. Validator error message uses
length-only to avoid potential information disclosure.
Documented that the default "{}" exists for backward
compatibility; new plans should explicitly set the snapshot.
2. Made root_plan_id NOT NULL — root plans self-reference their
own plan_id, child plans reference the root ancestor. Added
explicit ondelete="RESTRICT" FK policy for consistency with
other FKs in the model. Documented known FK policy drift
between ORM model (RESTRICT) and migrated databases (retained
SET NULL) in the migration; data integrity is preserved by
the NOT NULL constraint regardless. Moved root_plan_id
self-reference resolution into a PlanIdentity model_validator
so the domain model is consistent with the DB NOT NULL
constraint before and after persistence (previously the
resolution only happened in from_domain(), creating an
asymmetry where root_plan_id was None in-memory but non-null
after round-tripping through the database).
3. Made automation_profile NOT NULL with default "balanced".
4. Documented intentional deviation: phase default is "action"
(code) vs "strategize" (spec) because the Action phase was
added as a pre-Strategize setup step.
5. Created Alembic migration with backfill logic for existing
rows. Root-ancestor backfill uses level-by-level propagation
with a parent-readiness guard to correctly resolve plans at
arbitrary hierarchy depth (3+ levels). Added safety bound
(max 100 iterations) with logged error on exhaustion to guard
against cycles in parent_plan_id. Merged batch_alter_table
operations to avoid redundant full-table copies in SQLite
batch mode. Migration backfill also handles empty-string
automation_profile values. Documented downgrade limitation
(backfill is not reversible). Orphan-row fallback now logs
affected row count at WARNING level. Migration cycle-detection
now logs affected plan_id values before the orphan fallback
overwrites them. All migration SQL uses sa.text() for
consistency with SQLAlchemy best practices.
6. Hardened automation_profile deserialization in to_domain() to
catch ValueError (invalid StrEnum provenance), Pydantic
ValidationError, and RecursionError (deeply nested JSON) in
addition to JSONDecodeError and KeyError, preventing
unreadable plans from corrupted DB rows. Applied the same
defensive deserialization pattern to effective_profile_snapshot
in to_domain(): corrupted JSON falls back to '{}' with a
WARNING log instead of crashing the read path. Added TypeError
to the effective_profile_snapshot exception list in to_domain()
for consistency with the Pydantic validator. Logging of
unparseable values uses length only to avoid potential
information disclosure.
7. Used explicit None check (is not None) instead of truthiness
for root_plan_id resolution in from_domain(), for
effective_profile_snapshot in to_domain(), and in
_serialize_automation_profile() for consistency.
8. Documented intentional column naming conventions vs spec DDL
(e.g. automation_profile vs automation_profile_name, *_actor
vs *_actor_name, processing_state vs state, v3_plans vs
plans). Documented the semantic difference: automation_profile
stores either a bare name or structured JSON with provenance,
whereas the spec automation_profile_name stores a plain name.
9. Fixed benchmark plan constructors
(plan_phase_migration_bench.py) that were missing the now-
required root_plan_id and effective_profile_snapshot fields.
10. Replaced defensive getattr() with direct attribute access for
effective_profile_snapshot in from_domain() and update(),
since the field is now defined on the Plan domain model.
11. Fixed Any type annotation in test helper _make_plan() to use
AutomationProfileRef | None for proper type safety.
12. Added BDD scenarios for PlanIdentity self-reference
resolution, NULL effective_profile_snapshot constraint
enforcement, valid-JSON-missing-profile_name-key
deserialization, invalid-JSON and empty-string snapshot
rejection by Pydantic validator, and corrupted
effective_profile_snapshot DB fallback in to_domain().
13. Extracted default automation profile name to a module-level
constant (DEFAULT_AUTOMATION_PROFILE) to reduce sentinel
duplication across models.py and repositories.py.
14. Centralised automation-profile serialisation into
LifecyclePlanModel._serialize_automation_profile() to
eliminate duplication between from_domain() and
LifecyclePlanRepository.update().
15. Fixed to_domain() root_plan_id type cast from str | None
to str, reflecting the NOT NULL column constraint.
16. Added PlanIdentity model_validator that resolves None
root_plan_id to plan_id at domain construction time, ensuring
the domain model honours the spec DDL NOT NULL constraint
regardless of persistence state. Simplified from_domain()
root resolution accordingly.
ISSUES CLOSED: #921
Implement three-scope configuration resolution (local > project > global)
with deep merge. Add ConfigScope enum, project-root discovery, and
config.local.toml file support.
Changes:
- Add LOCAL to ConfigLevel enum and new ConfigScope enum (GLOBAL, PROJECT, LOCAL)
- Implement discover_project_root() walking up from CWD for cleveragents.toml/.cleveragents
- Implement config.local.toml file loading in ConfigService
- Implement three-scope deep merge via _deep_merge() helper
- Add read_project_config(), read_local_config(), read_merged_config() methods
- Add write_scoped_config() for writing to any scope file
- Update set_value() to accept optional scope parameter
- Update resolve() with six-level precedence chain (CLI > env > local > project > global > default)
- Update CLI config set with --scope flag (global/project/local)
- Add config.local.toml to .gitignore and DEFAULT_IGNORE_PATTERNS
- Add Behave BDD scenarios for three-scope resolution, deep merge, and project-root discovery
ISSUES CLOSED: #937
Register context as a sub-app of the actor Typer group so commands are
accessible under the canonical specification path `agents actor context`.
The top-level `agents context` group remains as a deprecated alias
(with Typer deprecated=True) for one release cycle.
Key changes:
- context.app registered under actor.app in main.py
- `rm` subcommand renamed to `remove`; `rm` kept as hidden deprecated
alias that prints a deprecation warning before delegating
- Programmatic rm_command() deprecated in favor of remove_command()
- Help text updated to reference canonical `agents actor context` paths
- Behave and Robot tests updated to use canonical paths and renamed
subcommand; assertion for context help lowercased for flexibility
- All nox sessions pass: lint, typecheck, unit_tests (495 features,
12731 scenarios), coverage at 97%
ISSUES CLOSED: #888
Implement the full correction-checkpoint rollback pipeline:
- Workspace snapshots: CheckpointService.create_workspace_snapshot()
creates diff-based checkpoints before decision execution, storing
only changed file paths in metadata.extra["diff_paths"]
- CorrectionService.revert_decisions(): new high-level entry point
that creates a correction, computes impact, invokes checkpoint
rollback, and archives artifacts in a single call
- Physical artifact archival: CheckpointService.archive_artifacts()
moves files to .cleveragents/archived_artifacts/ instead of just
flagging metadata. CorrectionService._archive_decision_artifacts()
delegates to this during revert execution
- Selective rollback: CheckpointService.selective_rollback() wraps
rollback_to_checkpoint with atomic semantics — captures HEAD before
rollback and recovers on failure
- Diff-based storage: _compute_diff_snapshot() computes changed paths
between checkpoints via git diff; snapshots store diff manifest and
SHA-256 hash in metadata
- CLI: plan rollback now accepts --to-checkpoint <id> in addition to
the positional checkpoint_id argument; uses selective_rollback for
atomic execution
- DI wiring: Container now injects checkpoint_service into
CorrectionService; CLI correct command uses container-provided
service instead of ad-hoc instance (fixes bug #986)
- Checkpoint model: pre_decision added to allowed checkpoint_type
values
- TDD: Removed @tdd_expected_fail from wiring test feature since the
DI bug is now fixed
ISSUES CLOSED: #943
Implement three new actor context subcommands as specified in the v3
CLI specification:
- `agents actor context remove [--yes|-y] (--all|-a|<NAME>)`:
Remove a named context or all contexts with interactive confirmation.
Supports --all for bulk removal and --yes to skip prompts.
- `agents actor context export (--output|-o) <FILE> <NAME>`:
Export a named context to JSON or YAML (detected from file extension).
Includes sha256 integrity checksum in output metadata.
- `agents actor context import [--update] (--input|-i) <FILE> [<NAME>]`:
Import a context from JSON or YAML. Name is inferred from file metadata
if omitted. Refuses to overwrite existing contexts unless --update is
specified.
All commands:
- Support 6 output formats (json, yaml, plain, table, rich, color) via
the shared format_output framework
- Use ContextManager for persistence (messages, metadata, state,
global_context)
- Follow existing error handling patterns (typer.Exit for user errors)
- Are wired as `agents actor context` subcommand group via add_typer
New files:
- src/cleveragents/cli/commands/actor_context.py (command module)
- features/actor_context_cmds.feature (Behave BDD scenarios)
- features/steps/actor_context_cmds_steps.py (BDD step definitions)
- robot/actor_context_export_import.robot (Robot integration test)
ISSUES CLOSED: #869
Remove the <NAME> positional argument from `actor add` command. The actor
name is now derived from the `name` field inside the config YAML file, per
the specification: `agents actor add (--config|-c) <FILE> [--update]`.
Updated all BDD and integration tests to pass name via config file.
ISSUES CLOSED: #914
Align the structured output envelope (JSON/YAML modes) with the specification:
add `status` field, rename `elements` to `data`, rename `metadata` to
`messages`, and fix the fallback chain to include the table step
(rich→table→color→plain).
ISSUES CLOSED: #884
Add --execution-env-priority flag to 'project context set' command,
enabling project-level execution environment priority per spec WF17.
- Add execution_env_priority field to ContextConfig domain model
- Validate flag value against ExecutionEnvPriority enum (fallback/override)
- Persist in context_policy_json, preserving existing execution_environment
- Display in 'project context show' Execution Environment section
- Merge with existing blob to avoid overwriting previously set fields
Tests: 9 Behave scenarios, 26 steps.
ISSUES CLOSED: #1079
Renamed all 11 task-type confidence threshold fields in AutomationProfile
from phase-transition semantics to spec-defined task-type semantics.
Updated all 8 built-in profiles, CLI formatting, YAML schema, services,
and all Behave/Robot tests referencing the old field names.
Post-review fixes:
- Fixed 24 stale old field names in M6 fixture files
(automation_profiles.json, autonomy_guardrails.json)
- Added model_validator(mode='before') to detect legacy field names
and raise actionable ValueError with rename mapping
- Added semantic bridge comments in PlanLifecycleService mapping
task-type thresholds to phase-transition gates
- Added threshold_field to structured log messages for observability
- Restored categorised CLI automation-profile show output to match
spec (Phase Transitions / Decision Automation / Self-Repair /
Execution Controls) instead of flat list
- Added missing access_network field to spec show output examples
(Rich, Plain, JSON, YAML variants)
- Aligned ADR-017 profile fields table to all 11 fields with
descriptions matching spec Automatable Tasks table
- Aligned automation_profiles.md threshold descriptions with spec
- Added spec section references in phase_reversion.md, error_recovery.md,
and plan_execute.md for field naming context
- Extended repository roundtrip test to assert all 11 threshold fields
- Fixed benchmark _make_profile() passing safety fields as top-level
kwargs instead of via SafetyProfile sub-model (incompatible with
extra="forbid")
- Aligned CLI JSON/YAML output structure for automation-profile show
with the specification grouped format (phase_transitions,
decision_automation, self_repair, execution_controls)
- Moved safety boolean fields into the Execution Controls section
of Rich output per spec examples
- Reverted auto profile description to "Fully automatic except apply"
per specification (line 16703, line 28406)
- Improved bridge comments in test steps with semantic context for
threshold-to-gate mappings
ISSUES CLOSED: #902
E2E test for Specification Workflow Example 18: Container with Remote
Repo Clone using the trusted automation profile. Exercises the full
container-instance workflow per the spec:
- Registers a container-instance resource with --clone-into flag (new
CLI flag added to resource add for REPO_URL:CONTAINER_PATH).
- Two-step project creation: project create + project link-resource.
- Dynamic LLM actor selection (Anthropic/OpenAI) matching available
API keys.
- Plan use with --execution-environment container,
--execution-env-priority fallback, and --automation-profile trusted.
- Full plan lifecycle: strategize, execute, diff, apply, status.
- Positive assertions: non-empty output, plan ID presence, terminal
state verification, container/clone evidence logging.
- Negative assertions: Traceback and INTERNAL error checks.
- ULID regex uses Crockford base32 character set.
- Unique suffix for resource/project names (parallel CI safe).
- Skip If No LLM Keys guard for graceful degradation.
- CHANGELOG.md updated.
ISSUES CLOSED: #764
Implements E2E test for devcontainer-driven development workflow.
Tests project setup with .devcontainer config, resource registration,
and full plan lifecycle with supervised automation profile.
Changes from Cycle 5 review:
1. Replaced custom WF16_ALLOW_XFAIL_SKIP escape hatch with the project
standard tdd_expected_fail tag system. Test is now tagged
[Tags] tdd_expected_fail tdd_bug tdd_bug_762 and the listener
inverts failures in CI until devcontainer features are fully wired.
2. Added diagnostic WF16 Test Teardown keyword that captures plan
status JSON on failure, mirroring the WF05 teardown pattern.
3. Added stderr checks to all Traceback/INTERNAL guard assertions
so tracebacks appearing only in stderr are also detected.
4. Rebased branch onto latest master (e4c01492).
5. Documented timeout budget: added comment explaining theoretical
worst-case (~46min) vs realistic execution under the 35min limit.
6. Added fail-fast warning in AC-3 check noting that AC-4, AC-5, and
AC-6 are dependent on AC-3 and are likely to also fail.
7. Reordered plan apply arguments to --yes --format json <plan_id>
for spec consistency.
8. Decomposed long ready_for_apply Evaluate expression into three
intermediate booleans (is_apply_phase, is_execute_complete,
is_already_applied) for readability.
9. Added shared_session.close() calls in resource_dag.robot before
each final print() to ensure explicit session cleanup.
10. Cleaned up stale xfail/skip classification references in log
messages.
Changes from earlier cycles (3-4) retained in this commit:
- Terminal state validation, AC output combination, JSON format flags,
suite-level init, Force Tags, automation profile fallback, ULID regex
IGNORECASE, AC-5 Evaluate decomposition, defensive re-check pattern.
ISSUES CLOSED: #762
Add explicit indexing timeout bounds and propagate timeout_seconds through the repo indexing service, utility walker, and CLI entrypoint so large repositories fail predictably instead of hanging.
Add 10K-scale Behave/Robot verification and a dedicated benchmark path to validate that large-project indexing completes, persists status correctly, and remains measurable for regressions.
Address review feedback by splitting oversized Behave and Robot support files into focused modules, trimming repo_indexing_service.py back under the 500-line limit, and rebasing the branch onto current master.
ISSUES CLOSED: #851
# Conflicts:
# CHANGELOG.md
# robot/helper_m5_e2e_verification.py
Integrate ACMS execute-phase context assembly into LLMExecuteActor and inject assembled context into execute prompts with resilient fallback when assembly fails or returns empty output.
Wire the plan CLI executor to use an ACMS-backed execute context assembler, add Behave coverage for context injection/fallback/empty context, and extend Robot M5 verification helpers to assert execute-phase ACMS context usage.
ISSUES CLOSED: #850
Add phase-wise context analysis summaries for strategize, execute, and apply, and integrate them into `agents project context inspect` and `agents project context simulate` with per-phase resource counts, size and token utilization, and narrowing diagnostics.\n\nAdd Behave and Robot coverage for empty, single-resource, multi-resource, and budget-constrained context analysis, and tighten the M5 smoke coverage after review feedback with exact summary assertions and a missing-path regression scenario for `agents context add`.\n\nISSUES CLOSED: #849
Implements E2E test for Specification WF17: Explicit Container with
Directory Mount. The test exercises the full workflow from the spec:
- Registers git-checkout resource and container-instance resource
(with --image flag; --mount not yet available on CLI)
- Creates project, links both resources via project link-resource
- Sets execution environment to container via project context set
- Uses --execution-env-priority override on plan use (per ADR-043)
- Creates action with dynamically selected actor (Anthropic or OpenAI)
- Runs full plan lifecycle: use → strategize → execute → diff → apply → status
- Uses trusted automation profile for auto-execution of tool invocations
Quality improvements over initial implementation:
- Custom suite setup with workspace init (agents init --force --yes)
- UUID-based unique suffix for resource/project names (CI parallelism)
- Dynamic actor selection matching available API keys (M6 pattern)
- Skip If No LLM Keys guard for graceful degradation
- Crockford Base32 ULID regex ([0-9A-HJ-NP-Z]{26})
- Checked return codes on all git subprocess calls
- INTERNAL error checks alongside Traceback checks
- --format json on all CLI commands for deterministic output
- Domain-specific assertions for plan output verification
- Diff output non-empty assertion
- Security note documenting trusted profile auto-execution decision
- CHANGELOG entry added
ISSUES CLOSED: #763
Implements E2E test for Workflow Example 12: large-scale hierarchical
feature implementation with the supervised automation profile.
Key test behaviors:
- Sets up 4 projects (protos, api, worker, frontend) with per-project
invariants, matching the specification.
- Creates action YAML with spec-required fields: estimation_actor,
invariant_actor, automation_profile: cautious, and action-level
invariants.
- Passes all 4 projects to plan use (spec Step 3).
- Dynamically selects actor based on available API key (Anthropic or
OpenAI) for flexible CI/local execution.
- Uses Skip If No LLM Keys for graceful degradation in keyless CI.
- Verifies hierarchical decomposition via plan tree (decision_id count,
children field check).
- Selects a non-root decision for plan correct --mode append to
exercise targeted correction rather than root invalidation.
- Verifies correction acknowledgment via output content assertions.
- Verifies phased lifecycle-apply with --yes flag and phase field
parsing.
- Final status uses --format json with Safe Parse Json Field to verify
terminal state (phase and processing_state populated).
- UUID-suffixed names prevent UNIQUE constraint collisions in parallel CI.
- Crockford Base32 ULID regex (excludes I, L, O, U) consistent with
m1_acceptance.robot.
- All commands check for Traceback and INTERNAL error markers.
- Force Tags E2E in Settings (m6 convention).
- Timeout set to 35 minutes for real LLM execution headroom.
Note: plan prompt (spec Step 4 — supervised-profile user intervention)
is not yet implemented as a CLI command. A documentation note in the
test indicates this should be added once the command is available.
ISSUES CLOSED: #758
freemo
brent.edwards
hamza.khyari
CoreRasurae
hurui200320
aditya