Skip to content

English | Русский

SENAR v1.5 Core — Compliance Matrix

Date: 2026-06-13 | Auditors: 6+ independent review cycles | Framework: TAUSIK v1.5.1

Quality Gates

GateRequirementStatusEnforcementEvidence
QG-0Goal required✅ ImplementedHard blockgate_qg0_check.py check_qg0_start() — ServiceError (via service_gates.GatesMixin._check_qg0_start delegator)
QG-0AC required✅ ImplementedHard blockgate_qg0_check.py check_qg0_start() — ServiceError (via service_gates.GatesMixin._check_qg0_start delegator)
QG-0Negative scenario in AC✅ ImplementedHard blockgate_negative_scenario.py NEGATIVE_SCENARIO_KEYWORDS + has_negative_scenario() (30+ en+ru); enforced inside gate_qg0_check.check_qg0_start()
QG-0Scope warning✅ ImplementedWarninggate_qg0_check.py check_qg0_start() — scope + scope_exclude stderr
QG-0Security surface detection✅ ImplementedWarninggate_qg0_check.py SECURITY_KEYWORDS + SECURITY_AC_KEYWORDS (re-exported by service_gates for backward-compat)
QG-2AC verified with evidence✅ ImplementedHard blockgate_ac_check.py verify_ac() — flag + notes + per-criterion. NO --force bypass. (via service_gates.GatesMixin._verify_ac delegator)
QG-2Plan steps complete✅ ImplementedHard blockgate_ac_check.py verify_plan_complete() — JSON plan check (via service_gates.GatesMixin._verify_plan_complete delegator)
QG-2Scoped pytest gate✅ ImplementedHard blockservice_verification.py — basename match tests/test_<file>.py per relevant_files (no fallback to full suite when files supplied)
QG-2Verify cache (10 min TTL)✅ ImplementedSkip-on-hitverification_runs table — same files_hash + green = skip; security paths bypass cache
QG-2Quality gates (pytest/ruff)✅ ImplementedHard blockgate_runner.py + service_gates.py _run_quality_gates()
QG-2Verification checklist (4 tiers)✅ ImplementedWarninggate_ac_check.py check_verification_checklist() + determine_checklist_tier() auto-tier — v1.5 also runs service_ac_evidence.build_report() to surface per-AC coverage gaps and missing test refs (via service_gates.GatesMixin._check_verification_checklist delegator)
QG-2Root cause for defects✅ ImplementedWarningservice_task.py task_done() — keyword check
QG-2Knowledge capture✅ ImplementedWarningservice_task.py task_done() — memory/decision count

Result: 13/13 implemented. Enforcement levels match SENAR spec.

Rules

RuleDescriptionStatusEnforcementEvidence
1Task before code✅ ImplementedHard (hook)hooks/task_gate.py blocks Write/Edit without active task
2Scope boundaries✅ ImplementedWarningscope + scope_exclude warned on start for medium/complex
3Verify against criteria✅ ImplementedHardQG-0 + QG-2 combined enforcement
5Verification checklist✅ ImplementedWarning4-tier auto-detection (lightweight/standard/high/critical)
7Root cause for defects✅ ImplementedWarningKeyword detection in notes
8Knowledge capture✅ ImplementedWarningmemory/decision count + --no-knowledge opt-out
9.1No code without task✅ ImplementedHard (hook)Same as Rule 1
9.2Session time limit (180 min active)✅ ImplementedHard blockBounded gap-based active time (Σ min(Δ, threshold), default threshold 10 min — long AFK clipped to threshold, v14b-session-active-time). service_gates.py blocks task_start at >180 min active; status shows "X min active / Y min wall"; session extend and session recompute available. Threshold configurable via session_idle_threshold_minutes.
9.3Checkpoint every 30-50 calls✅ ImplementedWarning (auto)MCP counter in meta table, warning at 40 calls, reset on handoff
9.4Document dead ends✅ ImplementedInstruction + toolingdead_end() + skill instructions + /end check
9.5Periodic audit✅ ImplementedWarningaudit_check/mark + /start integration

Result: 11/11 implemented.

Gaps and Plan to Close

GapPlanPriority
Rule 2: scope_exclude not checked✅ FIXED — warning added for medium/complex tasksDone
Rule 9.3: No automated checkpoint counter✅ FIXED — MCP counter + warning at 40 calls + reset on handoffDone

Metrics

MetricStatusEvidence
Throughput (tasks/session)✅ Implementedbackend_queries.py get_metrics() combined query
Lead Time (avg hours)✅ Implementedbackend_queries.py get_metrics() — julianday * 24
FPSR (first pass %)✅ Implementedbackend_queries.py get_metrics() — attempts=1
DER (defect escape %)✅ Implementedbackend_queries.py get_metrics() — DISTINCT defect_of
Dead End Rate (%)✅ Implementedbackend_queries.py get_metrics() — memory type=dead_end
Cost per Task (hours by complexity)✅ Implementedbackend_queries.py get_metrics() — GROUP BY complexity

Result: 6/6 implemented. All calculations verified correct.

Section 5.1: Explorations

FeatureStatusEvidence
explore_start (time-bounded, 30 min default)✅ Implementedservice_knowledge.py exploration_start() — clamps 1-480 min
explore_current (elapsed + over_limit)✅ Implementedservice_knowledge.py exploration_current() — UTC elapsed calc
explore_end (findings capture)✅ Implementedservice_knowledge.py exploration_end() — summary + optional task

Result: 3/3 implemented.

Additional Features (beyond SENAR Core)

FeatureStatusEvidence
Multi-language gates✅ Implementedproject_config.py — 25 default stacks + custom_stacks override
MCP coverage 124 tools✅ Implementedtools.py + tools_extra.py — (117 project + 7 brain)
Batch execution (/run)✅ Implementedplan_parser.py + /run skill
Structured logs (task_logs + FTS5)✅ Implementedbackend_schema.py + service_task.py:task_log
Fake test detection✅ Implemented/review skill — 10 patterns
Skills system✅ Implemented13 core skills + 20 official/vendor on demand (bundles via tausik skill bundle) — service_skills.py + tausik-skills repo
Hooks system✅ Implemented20 Python hooks + 1 shell pre-commit across PreToolUse / PostToolUse / SessionStart / SessionEnd / Stop / UserPromptSubmit
Roles registry✅ ImplementedHybrid: SQLite metadata + harness/roles/{role}.md profile; CRUD CLI + 6 MCP tools
Doctor health check✅ Implementedtausik doctor + tausik_doctor MCP — 4 groups (venv/DB/MCP/skills) + drift
Zero-defect skill✅ Implemented/zero-defect (Maestro-inspired): read-before-write, verify-before-claim, never-hallucinate-APIs

Overall Score

CategoryImplementedPartialMissingScore
Quality Gates (13)1300100%
Rules (11)1100100%
Metrics (6)600100%
Explorations (3)300100%
Total (33)3300100%

SENAR v1.3 Core compliance: 100%. All gaps closed.