specsmith 0.16.0.dev592__tar.gz → 0.16.0.dev599__tar.gz

{specsmith-0.16.0.dev592/src/specsmith.egg-info → specsmith-0.16.0.dev599}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: specsmith
-Version: 0.16.0.dev592
+Version: 0.16.0.dev599
 Summary: AEE governance toolkit for AI-assisted development — session preflight gates, multi-agent dispatch, requirements↔test traceability, ESDB persistence, MCP server, and skills for Warp, Cursor, Claude Code, Copilot, Windsurf, and Aider.
 Author: Layer1Labs Silicon, Inc.
 License-Expression: MIT
@@ -110,6 +110,22 @@ flowchart LR
 - **Aider / Claude Code / Cursor:** agentic coding interfaces; SpecSmith is the policy and evidence layer around these clients.
 - **LangGraph / AutoGen:** orchestration frameworks; SpecSmith is a governance-first development layer with compliance-oriented traceability.
 
+### Governance efficiency benchmark
+
+We ran a [multi-condition benchmark](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) comparing specsmith governance against 11 alternatives (ungoverned, BMAD, Cursor rules, Copilot, Aider, Cline, Codex CLI, OpenSpec, Agile BDD/TDD, and context injection) across real coding tasks with gpt-4o-mini and gpt-5.5.
+
+| Condition | Pass Rate | Mean Tokens | Cost/run | Cost-of-Pass |
+|---|---|---|---|---|
+| Ungoverned (raw agent) | 0% on T1 | 44.6k | $0.0079 | ∞ |
+| Context injection (CLAUDE.md) | 100% | 43.7k | $0.0084 | $0.0084 |
+| BMAD-style structured prompting | 50% | 139.1k | $0.0262 | $0.0523 |
+| **specsmith LIGHT (preflight)** | **100%** | **21.1k** | **$0.0032** | **$0.0032** |
+| **specsmith FULL (governed)** | **100%** | **17.1k** | **$0.0026** | **$0.0026** |
+
+**Key findings:** specsmith FULL is the only condition to achieve 100% pass rate on the feature-addition task (T1). It uses 2.6× fewer tokens than ungoverned and produces a cost-of-pass 3.2× lower than the next-best alternative. With gpt-5.5, governance reduces cost-of-pass by **6.3×** ($0.028 vs $0.179).
+
+See the [full benchmark report](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) and [model comparison (gpt-4o-mini vs gpt-5.5)](https://specsmith.readthedocs.io/en/stable/model-comparison/).
+
 **v0.16.0 — stabilisation milestone:** 20+ new CLI commands (quickstart, expand, verify-integrations, governed-pr, transcript import, policy, recover, dashboard, drift-check, trace score, export/import, approve); new governance modules; ESDB SQLite backend with full test coverage; chronomemory 0.2.0 with all phases complete; Python 3.10–3.13 × Ubuntu + Windows all green; 1 607 tests passing.
 specsmith ships a full compliance and auditability layer aligned to the EU AI Act (2024/1689)
 and the NIST AI Risk Management Framework 1.0. Every agent action is cryptographically sealed,
@@ -431,7 +447,7 @@ across the AI lifecycle. specsmith addresses all four core functions:
 
 | NIST AI RMF Function | specsmith Mechanism |
 |---|---|
-| **GOVERN** — Policies & accountability | Governance rules (H1–H22), permissions profile, `scaffold.yml` policy |
+|| **GOVERN** — Policies & accountability | Governance rules, permissions profile, `scaffold.yml` policy |
 | **MAP** — Risk identification | AEE stress-test, belief graph, contradictions and uncertainty metrics |
 | **MEASURE** — Risk analysis | Confidence scoring, epistemic equilibrium, `specsmith epistemic-audit` |
 | **MANAGE** — Risk treatment | Kill-switch, escalation, bounded retry, safe-write backup, permissions deny-list |
@@ -1032,42 +1048,15 @@ Use cases: linguistics research, compliance pipelines, AI alignment, patent pros
 
 ---
 
-## Governance Rules (H1–H22)
-
-22 hard rules enforced by `specsmith validate` and `specsmith audit`.
-Full rule text: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/main/docs/governance/RULES.md)
-
-**H1–H14 — Core engineering and traceability rules:**
-- **H1** — No ledger entry = work not done.
-- **H2** — No proposal = no execution.
-- **H3** — All work must consider every target platform.
-- **H4** — No system-dependent assumptions; virtual environments required.
-- **H5** — No hidden service logic.
-- **H6** — If the task grows beyond the proposal, stop and re-propose.
-- **H7** — Every state change must be traceable and recorded.
-- **H8** — Architecture changes MUST update docs in the same work cycle.
-- **H9** — Every agent command must have a timeout.
-- **H10** — No hardcoded version strings outside `pyproject.toml`.
-- **H11** — Every loop must have a deadline; no unbounded blocking I/O.
-- **H12** — Platform-aware automation: sh/bash on Unix, `.cmd`/`.ps1` on Windows.
-- **H13** — Every proposal must declare its epistemic boundaries and assumptions.
-- **H14** — Documentation must be updated in the same work cycle as code changes.
-
-**H15–H22 — Anti-hallucination and epistemic stability (OEA framework):**
-
-Rules H15–H22 are derived from the *"Ontology-Epistemic-Agentic (OEA) Recursive
-Generative Stability"* study (Layer1Labs Research, 2026), which empirically validated
-the primary control mechanisms for preventing hallucination and semantic drift in
-production LLM systems:
-
-- **H15** — Epistemic scope bounding: no claims outside verified knowledge; say "unknown" rather than fabricate.
-- **H16** — Anti-drift recursion guard: max 5 autonomous generation steps before a human checkpoint.
-- **H17** — Calibration direction: express uncertainty, not false confidence.
-- **H18** — RAG retrieval filtering: validate context relevance (similarity ≥ 0.6) before injection.
-- **H19** — Synthetic contamination prevention: never mix synthetic and real data silently.
-- **H20** — Falsifiability required: cite sources or flag claims as `[HYPOTHESIS]`.
-- **H21** — Disclose all model-specific assumptions (context window, format, temperature).
-- **H22** — Cross-platform CI: green on one OS ≠ cross-platform coverage.
+## Governance Rules
+
+22 hard rules enforced by `specsmith validate` and `specsmith audit`. Rules 1–14 cover
+core engineering and traceability (ledger required, proposals required, platform-awareness,
+documentation currency, etc.). Rules 15–22 address anti-hallucination and epistemic
+stability, derived from the OEA (Ontological Epistemic Anchoring) research framework
+(Layer1Labs, 2026).
+
+Full rule reference: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/develop/docs/governance/RULES.md)
 
 ---
 

{specsmith-0.16.0.dev592 → specsmith-0.16.0.dev599}/README.md

@@ -32,6 +32,22 @@ flowchart LR
 - **Aider / Claude Code / Cursor:** agentic coding interfaces; SpecSmith is the policy and evidence layer around these clients.
 - **LangGraph / AutoGen:** orchestration frameworks; SpecSmith is a governance-first development layer with compliance-oriented traceability.
 
+### Governance efficiency benchmark
+
+We ran a [multi-condition benchmark](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) comparing specsmith governance against 11 alternatives (ungoverned, BMAD, Cursor rules, Copilot, Aider, Cline, Codex CLI, OpenSpec, Agile BDD/TDD, and context injection) across real coding tasks with gpt-4o-mini and gpt-5.5.
+
+| Condition | Pass Rate | Mean Tokens | Cost/run | Cost-of-Pass |
+|---|---|---|---|---|
+| Ungoverned (raw agent) | 0% on T1 | 44.6k | $0.0079 | ∞ |
+| Context injection (CLAUDE.md) | 100% | 43.7k | $0.0084 | $0.0084 |
+| BMAD-style structured prompting | 50% | 139.1k | $0.0262 | $0.0523 |
+| **specsmith LIGHT (preflight)** | **100%** | **21.1k** | **$0.0032** | **$0.0032** |
+| **specsmith FULL (governed)** | **100%** | **17.1k** | **$0.0026** | **$0.0026** |
+
+**Key findings:** specsmith FULL is the only condition to achieve 100% pass rate on the feature-addition task (T1). It uses 2.6× fewer tokens than ungoverned and produces a cost-of-pass 3.2× lower than the next-best alternative. With gpt-5.5, governance reduces cost-of-pass by **6.3×** ($0.028 vs $0.179).
+
+See the [full benchmark report](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) and [model comparison (gpt-4o-mini vs gpt-5.5)](https://specsmith.readthedocs.io/en/stable/model-comparison/).
+
 **v0.16.0 — stabilisation milestone:** 20+ new CLI commands (quickstart, expand, verify-integrations, governed-pr, transcript import, policy, recover, dashboard, drift-check, trace score, export/import, approve); new governance modules; ESDB SQLite backend with full test coverage; chronomemory 0.2.0 with all phases complete; Python 3.10–3.13 × Ubuntu + Windows all green; 1 607 tests passing.
 specsmith ships a full compliance and auditability layer aligned to the EU AI Act (2024/1689)
 and the NIST AI Risk Management Framework 1.0. Every agent action is cryptographically sealed,
@@ -353,7 +369,7 @@ across the AI lifecycle. specsmith addresses all four core functions:
 
 | NIST AI RMF Function | specsmith Mechanism |
 |---|---|
-| **GOVERN** — Policies & accountability | Governance rules (H1–H22), permissions profile, `scaffold.yml` policy |
+|| **GOVERN** — Policies & accountability | Governance rules, permissions profile, `scaffold.yml` policy |
 | **MAP** — Risk identification | AEE stress-test, belief graph, contradictions and uncertainty metrics |
 | **MEASURE** — Risk analysis | Confidence scoring, epistemic equilibrium, `specsmith epistemic-audit` |
 | **MANAGE** — Risk treatment | Kill-switch, escalation, bounded retry, safe-write backup, permissions deny-list |
@@ -954,42 +970,15 @@ Use cases: linguistics research, compliance pipelines, AI alignment, patent pros
 
 ---
 
-## Governance Rules (H1–H22)
-
-22 hard rules enforced by `specsmith validate` and `specsmith audit`.
-Full rule text: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/main/docs/governance/RULES.md)
-
-**H1–H14 — Core engineering and traceability rules:**
-- **H1** — No ledger entry = work not done.
-- **H2** — No proposal = no execution.
-- **H3** — All work must consider every target platform.
-- **H4** — No system-dependent assumptions; virtual environments required.
-- **H5** — No hidden service logic.
-- **H6** — If the task grows beyond the proposal, stop and re-propose.
-- **H7** — Every state change must be traceable and recorded.
-- **H8** — Architecture changes MUST update docs in the same work cycle.
-- **H9** — Every agent command must have a timeout.
-- **H10** — No hardcoded version strings outside `pyproject.toml`.
-- **H11** — Every loop must have a deadline; no unbounded blocking I/O.
-- **H12** — Platform-aware automation: sh/bash on Unix, `.cmd`/`.ps1` on Windows.
-- **H13** — Every proposal must declare its epistemic boundaries and assumptions.
-- **H14** — Documentation must be updated in the same work cycle as code changes.
-
-**H15–H22 — Anti-hallucination and epistemic stability (OEA framework):**
-
-Rules H15–H22 are derived from the *"Ontology-Epistemic-Agentic (OEA) Recursive
-Generative Stability"* study (Layer1Labs Research, 2026), which empirically validated
-the primary control mechanisms for preventing hallucination and semantic drift in
-production LLM systems:
-
-- **H15** — Epistemic scope bounding: no claims outside verified knowledge; say "unknown" rather than fabricate.
-- **H16** — Anti-drift recursion guard: max 5 autonomous generation steps before a human checkpoint.
-- **H17** — Calibration direction: express uncertainty, not false confidence.
-- **H18** — RAG retrieval filtering: validate context relevance (similarity ≥ 0.6) before injection.
-- **H19** — Synthetic contamination prevention: never mix synthetic and real data silently.
-- **H20** — Falsifiability required: cite sources or flag claims as `[HYPOTHESIS]`.
-- **H21** — Disclose all model-specific assumptions (context window, format, temperature).
-- **H22** — Cross-platform CI: green on one OS ≠ cross-platform coverage.
+## Governance Rules
+
+22 hard rules enforced by `specsmith validate` and `specsmith audit`. Rules 1–14 cover
+core engineering and traceability (ledger required, proposals required, platform-awareness,
+documentation currency, etc.). Rules 15–22 address anti-hallucination and epistemic
+stability, derived from the OEA (Ontological Epistemic Anchoring) research framework
+(Layer1Labs, 2026).
+
+Full rule reference: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/develop/docs/governance/RULES.md)
 
 ---
 

{specsmith-0.16.0.dev592 → specsmith-0.16.0.dev599}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "specsmith"
-version = "0.16.0.dev592"
+version = "0.16.0.dev599"
 description = "AEE governance toolkit for AI-assisted development — session preflight gates, multi-agent dispatch, requirements↔test traceability, ESDB persistence, MCP server, and skills for Warp, Cursor, Claude Code, Copilot, Windsurf, and Aider."
 readme = "README.md"
 license = "MIT"

{specsmith-0.16.0.dev592 → specsmith-0.16.0.dev599/src/specsmith.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: specsmith
-Version: 0.16.0.dev592
+Version: 0.16.0.dev599
 Summary: AEE governance toolkit for AI-assisted development — session preflight gates, multi-agent dispatch, requirements↔test traceability, ESDB persistence, MCP server, and skills for Warp, Cursor, Claude Code, Copilot, Windsurf, and Aider.
 Author: Layer1Labs Silicon, Inc.
 License-Expression: MIT
@@ -110,6 +110,22 @@ flowchart LR
 - **Aider / Claude Code / Cursor:** agentic coding interfaces; SpecSmith is the policy and evidence layer around these clients.
 - **LangGraph / AutoGen:** orchestration frameworks; SpecSmith is a governance-first development layer with compliance-oriented traceability.
 
+### Governance efficiency benchmark
+
+We ran a [multi-condition benchmark](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) comparing specsmith governance against 11 alternatives (ungoverned, BMAD, Cursor rules, Copilot, Aider, Cline, Codex CLI, OpenSpec, Agile BDD/TDD, and context injection) across real coding tasks with gpt-4o-mini and gpt-5.5.
+
+| Condition | Pass Rate | Mean Tokens | Cost/run | Cost-of-Pass |
+|---|---|---|---|---|
+| Ungoverned (raw agent) | 0% on T1 | 44.6k | $0.0079 | ∞ |
+| Context injection (CLAUDE.md) | 100% | 43.7k | $0.0084 | $0.0084 |
+| BMAD-style structured prompting | 50% | 139.1k | $0.0262 | $0.0523 |
+| **specsmith LIGHT (preflight)** | **100%** | **21.1k** | **$0.0032** | **$0.0032** |
+| **specsmith FULL (governed)** | **100%** | **17.1k** | **$0.0026** | **$0.0026** |
+
+**Key findings:** specsmith FULL is the only condition to achieve 100% pass rate on the feature-addition task (T1). It uses 2.6× fewer tokens than ungoverned and produces a cost-of-pass 3.2× lower than the next-best alternative. With gpt-5.5, governance reduces cost-of-pass by **6.3×** ($0.028 vs $0.179).
+
+See the [full benchmark report](https://specsmith.readthedocs.io/en/stable/efficiency-benchmark/) and [model comparison (gpt-4o-mini vs gpt-5.5)](https://specsmith.readthedocs.io/en/stable/model-comparison/).
+
 **v0.16.0 — stabilisation milestone:** 20+ new CLI commands (quickstart, expand, verify-integrations, governed-pr, transcript import, policy, recover, dashboard, drift-check, trace score, export/import, approve); new governance modules; ESDB SQLite backend with full test coverage; chronomemory 0.2.0 with all phases complete; Python 3.10–3.13 × Ubuntu + Windows all green; 1 607 tests passing.
 specsmith ships a full compliance and auditability layer aligned to the EU AI Act (2024/1689)
 and the NIST AI Risk Management Framework 1.0. Every agent action is cryptographically sealed,
@@ -431,7 +447,7 @@ across the AI lifecycle. specsmith addresses all four core functions:
 
 | NIST AI RMF Function | specsmith Mechanism |
 |---|---|
-| **GOVERN** — Policies & accountability | Governance rules (H1–H22), permissions profile, `scaffold.yml` policy |
+|| **GOVERN** — Policies & accountability | Governance rules, permissions profile, `scaffold.yml` policy |
 | **MAP** — Risk identification | AEE stress-test, belief graph, contradictions and uncertainty metrics |
 | **MEASURE** — Risk analysis | Confidence scoring, epistemic equilibrium, `specsmith epistemic-audit` |
 | **MANAGE** — Risk treatment | Kill-switch, escalation, bounded retry, safe-write backup, permissions deny-list |
@@ -1032,42 +1048,15 @@ Use cases: linguistics research, compliance pipelines, AI alignment, patent pros
 
 ---
 
-## Governance Rules (H1–H22)
-
-22 hard rules enforced by `specsmith validate` and `specsmith audit`.
-Full rule text: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/main/docs/governance/RULES.md)
-
-**H1–H14 — Core engineering and traceability rules:**
-- **H1** — No ledger entry = work not done.
-- **H2** — No proposal = no execution.
-- **H3** — All work must consider every target platform.
-- **H4** — No system-dependent assumptions; virtual environments required.
-- **H5** — No hidden service logic.
-- **H6** — If the task grows beyond the proposal, stop and re-propose.
-- **H7** — Every state change must be traceable and recorded.
-- **H8** — Architecture changes MUST update docs in the same work cycle.
-- **H9** — Every agent command must have a timeout.
-- **H10** — No hardcoded version strings outside `pyproject.toml`.
-- **H11** — Every loop must have a deadline; no unbounded blocking I/O.
-- **H12** — Platform-aware automation: sh/bash on Unix, `.cmd`/`.ps1` on Windows.
-- **H13** — Every proposal must declare its epistemic boundaries and assumptions.
-- **H14** — Documentation must be updated in the same work cycle as code changes.
-
-**H15–H22 — Anti-hallucination and epistemic stability (OEA framework):**
-
-Rules H15–H22 are derived from the *"Ontology-Epistemic-Agentic (OEA) Recursive
-Generative Stability"* study (Layer1Labs Research, 2026), which empirically validated
-the primary control mechanisms for preventing hallucination and semantic drift in
-production LLM systems:
-
-- **H15** — Epistemic scope bounding: no claims outside verified knowledge; say "unknown" rather than fabricate.
-- **H16** — Anti-drift recursion guard: max 5 autonomous generation steps before a human checkpoint.
-- **H17** — Calibration direction: express uncertainty, not false confidence.
-- **H18** — RAG retrieval filtering: validate context relevance (similarity ≥ 0.6) before injection.
-- **H19** — Synthetic contamination prevention: never mix synthetic and real data silently.
-- **H20** — Falsifiability required: cite sources or flag claims as `[HYPOTHESIS]`.
-- **H21** — Disclose all model-specific assumptions (context window, format, temperature).
-- **H22** — Cross-platform CI: green on one OS ≠ cross-platform coverage.
+## Governance Rules
+
+22 hard rules enforced by `specsmith validate` and `specsmith audit`. Rules 1–14 cover
+core engineering and traceability (ledger required, proposals required, platform-awareness,
+documentation currency, etc.). Rules 15–22 address anti-hallucination and epistemic
+stability, derived from the OEA (Ontological Epistemic Anchoring) research framework
+(Layer1Labs, 2026).
+
+Full rule reference: [`docs/governance/RULES.md`](https://github.com/layer1labs/specsmith/blob/develop/docs/governance/RULES.md)
 
 ---