HAL9000/temp
0
0
Fork 0
forked from HAL9000/cleveragents-core
Code Pull Requests Activity

docs(spec): document ReconciliationBlockedError and invariant reconciliation failure behavior at phase transitions

Browse Source 
Adds documentation for the reconciliation failure behavior that was
implemented in PR #5614 but not reflected in the specification:

- ReconciliationBlockedError blocks phase transitions when invariants fail
- INVARIANT_VIOLATED event emitted on reconciliation failure
- Reconciliation runs at Strategize, Execute, and Apply phase starts
- CORRECTION_APPLIED event triggers automatic re-reconciliation
- builtin/invariant-reconciliation actor name documented

ISSUES CLOSED: #5942
This commit is contained in:
HAL 9000
2026-04-09 14:10:08 +00:00
parent 33d5c0b244
commit bbbd39c6ee
1 changed files with 12 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/specification.md
+12
View File
@@ -19746,6 +19746,18 @@ The lookup order is: plan → project → global config. The first Invariant Rec
Each effective invariant is then recorded as an `invariant_enforced` decision in the plan's decision tree. This makes invariants visible, auditable, and correctable through the standard decision correction mechanism.
**Reconciliation failure behavior**: When invariant reconciliation fails at a phase transition, the transition is **blocked** — the plan cannot proceed to the next phase until the invariants are corrected. The system raises a `ReconciliationBlockedError` and emits an `INVARIANT_VIOLATED` event on the event bus.
Reconciliation runs at the start of three phase transitions:
- **Strategize start** (`start_strategize()`) — reconciles all applicable invariants before strategy decisions are recorded
- **Execute start** (`execute_plan()`) — re-checks invariants before execution begins
- **Apply start** (`apply_plan()`) — re-checks invariants before changes are applied to the project
After a correction is applied (`agents plan correct`), reconciliation automatically re-runs via a `CORRECTION_APPLIED` event subscription (best-effort; does not block correction completion). This allows the plan to resume if the correction resolves the invariant violation.
The built-in reconciliation actor is registered as `builtin/invariant-reconciliation`. It is used when no custom invariant actor is configured at the plan, project, or global level.
**Non-overridable global invariants**: A global invariant may be marked `non_overridable: true`. When set, this invariant takes precedence over all lower-scope invariants regardless of the normal precedence chain -- even plan-level invariants cannot override it. This is intended for system-wide safety constraints that must never be relaxed (e.g., "Never commit secrets to version control"). Non-overridable invariants are set via `agents invariant add --global --non-overridable "<constraint>"`. The `non_overridable` flag is only meaningful on `GLOBAL`-scoped invariants; it is ignored on project, action, and plan invariants.
**Child plan inheritance**: When a top-level plan spawns child plans, the parent's effective invariant view (already reconciled) is passed down to each child plan. Child plans do not re-run reconciliation — they inherit the parent's resolved view.