Check (SPAE)

When to use

• STATE.json reads phase: check: verify the active PLAN.md task against repository evidence.

Goal

• Judge the active task's implementation against its Intent, Acceptance, and Verification; record every gap in FIX.md.
• Advance the workstream to phase: fix on a gap; else mark the task done and advance to phase: build or phase: verify.

Input

Read only what the check requires:

• .spae/current/STATE.json; references/STATE.md holds field reference, directives, and snapshots.
• .spae/current/PLAN.md: the active task in full, plus task titles and order.
• Repository source, tests, docs, configuration, and logs the active task touches, for analysis.

Workflow

1. GATE—Make no writes:
   • Parse .spae/current/STATE.json; self-heal missing metadata: version ("1.0"), workstream (target directory name), status ("active"); PERSIST writes the healed fields.
   • Self-heal state drift: treat a "todo" or missing tasks[active_task_id] as "in_progress" until PERSIST.
   • Halt on any Halt conditions entry.
2. ORIENT—State goal in one sentence:
   • PLAN.md stays unmodified; directly modify no source or tests; write nothing before PERSIST.
   • Read the active task in full, plus PLAN.md task titles and order.
3. PLAN—Declare the evidence map:
   • Read references/verification-guide.md and references/shell-command-guide.md.
   • Map each Acceptance outcome, each Verification step, and the Intent to the evidence that proves it.
   • Halt on any Halt conditions entry.
4. ACT—Gather evidence and label findings:
   • Run every task Verification step and the project checks.
   • Read the task diff with the scoped audit command in references/cleanup-guide.md.
   • Judge each Acceptance outcome and the Intent against step output, the diff, and source.
   • Label each finding with exactly one tag:
     • Gap—blocks a clear verdict; cite reproducible evidence, never a hypothetical risk or stylistic preference:
       • An unmet Acceptance outcome, a failing Verification step, or behavior diverging from the Intent.
       • A project-check failure the task diff causes.
       • Speculative abstraction, unrequested complexity, or defensive bloat the diff adds beyond the task Scope.
       • A task-written test breaking references/verification-guide.md.
       • Gap covers only task-introduced violations within the active task's declared scope: debug output, dead code, an orphan file, a commented-out block, a superseded duplicate, an unused dependency, or stray build/cache output this task's own diff introduces and leaves behind. Never flag pre-existing uncommitted files outside task scope as gaps.
     • Observation—never a verdict input: a concern outside task scope, unrequested edge-case coverage, a defensive-code suggestion, or a theoretical risk.
   • On any Gap, draft FIX.md per references/FIX.md.
   • Apply Directives and Constraints.
5. VERIFY—Loop over Verification criteria for the verdict and the FIX.md draft:
   • For each failure: return to ACT, complete the evaluation, then re-enter VERIFY.
   • Exit only when every criterion passes.
   • Halt only on a Halt conditions entry.
6. PERSIST—Write in order:
   • On a Gap verdict, write .spae/current/FIX.md, replacing any prior copy.
   • Write STATE.json per the references/STATE.md exit snapshot matching the verdict.
   • Re-read STATE.json; on a mismatch against Verification state criteria, rewrite and re-read before REPORT.
7. REPORT—Emit result following result directives and using result template:
   • Status block by re-read phase: fix→Gap found, build→Clear, tasks remain, verify→Clear, final task; a halt→Failed.
   • List each Gap, then each Observation, under Findings.

Directives

• Optimize for agent, token, and context efficiency.
• Refine and optimize FIX.md for agent consumption: agents read it for varied purposes, with no memory of this run; keep it self-contained, explicit, and token-dense; make each gap closable without PLAN.md.
• Consult references/prose-protocol.md before drafting FIX.md.

Constraints

• Write scope: write only .spae/current/FIX.md and .spae/current/STATE.json; forbid all other writes. Byproducts of a command run for evidence, such as a formatter's rewrites, never count as this phase's writes; never run a command for its write effect outside this scope. Read source code only for analysis.
• Phase boundary: judge exactly one task; hand off work to /fix, /build, or /verify.
• Version control and release: never execute git mutations (staging, commits, branches, tags, pushes) or release operations. Never stage or commit .spae/.
• Autonomy: never ask users for input or clarification mid-execution; halts and failures stop autonomously. Never perform or request human execution, an attended or interactive terminal, or human presence.
• State schema: never introduce STATE.json fields outside references/STATE.md.
• Halt conditions: halt without writing; report Failed with a diagnostic under Findings:
  • .spae/current missing or dangling.
  • STATE.json missing or holds unparseable JSON syntax.
  • STATE.json phase other than check.
  • PLAN.md missing, cursor.active_task_id missing, or the active task absent from PLAN.md.
  • tasks[active_task_id] reads "done".
  • A planned Verification step can't run for a cause outside the task diff: missing tooling, a broken external dependency, or a need for human presence.
• Scope control: ignore pre-existing uncommitted files and modifications outside the active task's declared scope; they yield no finding.

Verification

• Every Acceptance outcome, Verification step, and the Intent judged against cited evidence.
• Every defect matching a Gap item—diff bloat and cleanup violations included—carries Gap; every other finding carries Observation.
• Verdict reads clear only with zero Gap findings.
• On a Gap verdict, the FIX.md draft matches references/FIX.md: ## Intent verbatim from PLAN.md; every Gap with File, Expected, Actual, and reproducible Evidence; no Observation.
• STATE.json matches the references/STATE.md exit snapshot for the verdict.
• This phase wrote no file outside Write scope.

Result directives

• Minimum words. Maximum signal.
• Keep prose terse while ensuring clarity.
• Optimize prose for agent, token, and context efficiency.
• Split actions, findings, and summaries into terse bullet points.
• Use lists and sub-lists for long sentences and dense paragraphs.
• Emit result template as live markdown—never in a code fence.
• Output nothing outside template.
• Source every SPAE Status value from confirmed re-read, never from working memory of intent.

Result template

### Execution Summary

- **Actions**:
  - [Terse list of checks and verdict work]
- **Files**:
  - [Terse list of affected SPAE artifacts]
- **Findings**:
  - [Gaps or observations]
- **Summary**:
  - [Terse task outcome]

Emit exactly one status block below, matching the outcome.

Gap found

> **`SPAE` Status** • `[workstream-name]`
> **Result**: Gaps found, `FIX.md` written
> **Active Task**: `T-XXX` - [Task title]
>
> _Run `/fix` to close the gaps._

Clear, tasks remain

> **`SPAE` Status** • `[workstream-name]`
> **Result**: Clear, tasks remain
> **Progress**: Task [X] of [Y] ([Z] remaining)
> **Completed**: `T-XXX` - [Task title]
> **Next Task**: `T-YYY` - [Next task title]
>
> _Run `/build` to execute the next task._

Clear, final task

> **`SPAE` Status** • `[workstream-name]`
> **Result**: Clear, final task
> **Progress**: All [X] tasks completed
> **Completed**: `T-001` through `T-XXX`
> **Next Phase**: `/verify`
>
> _Run `/verify` to validate the implementation against the specification._

Failed

> **`SPAE` Status** • `[workstream-name]`
> **Result**: Failed
> **Active Task**: `T-XXX` - [Task title]