@miller-tech/uap 2.11.0 → 2.11.1

package/README.md

@@ -9,7 +9,7 @@
 [![Tests](https://img.shields.io/badge/tests-450%2B_suites-brightgreen)](#testing)
 [![License](https://img.shields.io/badge/license-MIT-black)](LICENSE)
 
-`v1.224.0` · 367 TypeScript modules across 26 subsystems · 459 vitest suites (+ a ~1,200-test Python enforcer/proxy suite) · 9 agent harnesses
+`v2.11.0` · 367 TypeScript modules across 26 subsystems · 459 vitest suites (+ a ~1,200-test Python enforcer/proxy suite) · 9 agent harnesses
 
 [Quickstart](#quickstart) · [Why UAP?](#why-uap) · [The delivery pipeline](docs/guides/DELIVERY_PIPELINE.md) · [`uap deliver`](#the-deliver-harness) · [Docs](docs/INDEX.md)
 
@@ -110,17 +110,18 @@ UAP installs hooks into your agent harness, then mediates every tool call throug
                             ▼
 ┌─────────────────────────────────────────────────────────────┐
 │                       UAP CLI (uap)                         │
-│  setup · memory · deliver · verify · worktree · policy      │
-│  quality · plan · fidelity · interaction · task · coord ·   │
-│  droids · model · mcp-router · proxy · orchestrator ·       │
-│  self-harness · tune · design · principles · bench …        │
+│  setup · memory · deliver · verify · review · worktree ·    │
+│  policy · quality · plan · fidelity · interaction · task ·  │
+│  coord · droids · model · mcp-router · proxy · supervise ·  │
+│  orchestrator · self-harness · tune · design · principles · │
+│  bench · doctor · classify …                                │
 └──┬─────────┬──────────┬──────────┬──────────┬───────────────┘
    ▼         ▼          ▼          ▼          ▼
  Memory   Policy    MCP Router   Delivery   Coordination
  4 tiers  32 gates  compression  + verify   + deploy batch
 ```
 
-- **60 CLI commands** across 26 source subsystems (367 TypeScript modules).
+- **60+ CLI commands** across 26 source subsystems (367 TypeScript modules).
 - Deep dive: **[docs/architecture/OVERVIEW.md](docs/architecture/OVERVIEW.md)** · protocol spec: **[docs/architecture/PROTOCOL.md](docs/architecture/PROTOCOL.md)**.
 
 ---

package/docs/INDEX.md

@@ -1,6 +1,6 @@
 # UAP Documentation
 
-The complete documentation for the **Universal Agent Protocol** (`@miller-tech/uap` v1.224.0) — the discipline layer that turns a talented-but-unreliable AI coding agent into a dependable member of your software delivery line. 367 TypeScript modules across 26 subsystems, 459 vitest suites plus a ~1,200-test Python enforcer/proxy suite, 60 CLI command registrations, 32 executable policy enforcers, 9 supported agent harnesses.
+The complete documentation for the **Universal Agent Protocol** (`@miller-tech/uap` v2.11.0) — the discipline layer that turns a talented-but-unreliable AI coding agent into a dependable member of your software delivery line. 367 TypeScript modules across 26 subsystems, 459 vitest suites plus a ~1,200-test Python enforcer/proxy suite, 60+ CLI command registrations, 32 executable policy enforcers, 9 supported agent harnesses.
 
 > **Reviewing the system?** The reverse-engineered, code-verified reference set lives in
 > [`documentation/`](../documentation/architecture.md): architecture, flows, permissions
@@ -78,6 +78,7 @@ Full map: **[The UAP Delivery Pipeline](guides/DELIVERY_PIPELINE.md)**.
 | [Self-Harness](design/SELF_HARNESS.md) | Self-improving harness (shipped as `uap self-harness`): autonomous mine→propose→validate loop over a bounded Mod DSL; cross-model transfer, online mining |
 | [LLM Self-Tuning (analysis)](design/LLM_SELF_TUNING_ANALYSIS.md) | The design behind `uap tune`: quality signal, LLM-guided + Gaussian-process flag search, closed-loop validation, model profiles, real-time adaptation |
 | [Product Naming (analysis)](design/PRODUCT_NAMING.md) | UAP rename candidates: availability sweeps (npm/PyPI/domains), AI-space collision checks, railway-theme shortlist, recommendation. No decision committed; availability data is point-in-time |
+| [ADRs](architecture/adr/) | Architecture decision records: [0001 AutoMode risk floor](architecture/adr/0001-automode-risk-floor.md) (Python heuristic scorer ahead of the TS SYS1 head), [0002 Evidence-bound ship](architecture/adr/0002-evidence-bound-ship.md) (gate evidence bound to the candidate SHA) |
 
 ## Reference
 
@@ -92,6 +93,10 @@ Full map: **[The UAP Delivery Pipeline](guides/DELIVERY_PIPELINE.md)**.
 | [Platforms](reference/PLATFORMS.md) | The 9 supported harnesses + support matrix |
 | [Review Pre-Pass](reference/review-prepass.md) | Deterministic ruleset scanner (`uap review prepass`) that feeds line-anchored findings to the parallel review protocol |
 | [Visual Captures](reference/visual-captures.md) | Before/after capture binding for UI diffs (`uap review captures`) — ship-time gate 7 enforcement |
+| [System-1 Classifier](reference/system1-classifier.md) | `uap classify` — the local, airgap-pure assessment backend (noul/score/choice questions) behind the Wave 1 consumers; thresholds ship as policy config |
+| [Semantic Supervisor](reference/semantic-supervisor.md) | `uap supervise` — watchdog loop for `uap deliver` missions; nine assessment dimensions, deterministic safety-first policy, cooperative STOP only |
+| [AutoMode Risk](reference/automode-risk.md) | Pre-execution tool-call risk classification (`tools/agents/scripts/tool_risk.py`) — advisory only, ahead of the deterministic enforcer floor |
+| [Evidence-Bound Ship](reference/evidence-bound-ship.md) | Ship actions blocked unless `.uap/evidence/<head-sha>.json` proves the gates passed for the exact commit being shipped |
 | [Configuration](reference/CONFIGURATION.md) | All config files and env vars |
 | [**Configuration Reference**](reference/CONFIGURATION_REFERENCE.md) | Every setting `uap config` exposes — what it does, default, and a recommendation (generated from the registry); `uap config list/explain/set/doctor/wizard` ⭐ |
 | [Database Schema](reference/DATABASE_SCHEMA.md) | SQLite databases + Qdrant collections |

package/docs/guides/CAPACITY_POLICY.md

@@ -9,7 +9,8 @@ This generalizes the 2026-09-18 llama.cpp OOM fix: four SIGABRT crashes
 happened because a KV-cache budget was resolved against a GPU that also runs
 a desktop, and nothing declared that relationship. The fix held, but the
 *discipline* (declare budget + headroom + restart baseline, watch for
-violations) lived in a commit message. Now it is data.
+violations) lived only in a commit message; the capacity policy moves it
+into data.
 
 ## Policy file
 
@@ -73,9 +74,9 @@ elsewhere should ship their own `config/capacity-policy.json` (project) or
 `~/.config/uap/capacity-policy.json` (operator); the resolution order above
 makes the committed file the fallback only inside this repo.
 
-`uap doctor --strict` is the natural health source for the monitor service
-to consult before restarting anything, and for CI to refuse a deploy into a
-degraded host.
+`uap doctor --strict` is the health source the monitor service should
+consult before restarting anything, and CI can use it to refuse a deploy
+into a degraded host.
 
 ## Decisions
 

package/docs/plans/system1-uplift-2026-09-18.md

@@ -145,8 +145,7 @@ policy code.
 - **Training data:** UAP's own telemetry — `project_telemetry` events, gate
   decisions and outcomes, escalation results, stuck/attractor events,
   deliver-mission histories, and incident records (e.g., the 2026-09-18 OOM
-  timeline). This dataset is the moat: nobody else has UAP's decision-outcome
-  pairs.
+  timeline). Nobody else has UAP's decision-outcome pairs.
 - **Model form (v1):** small, CPU-fast, calibrated. Start with an embedding +
   calibrated head (nomic-embeddings already deployed; logistic/isotonic on
   top) for noul/score/choice questions. Evaluate against a distilled small

package/docs/reference/automode-risk.md

@@ -21,7 +21,7 @@ This is the Python-side calibrated heuristic floor for AutoMode. Its
 vocabulary is *inspired by* the unmerged TypeScript SYS1 baseline
 (`src/classify/baseline.ts` on `feature/301-sys1-classifier`) — not ported:
 the TS baseline scores prose state descriptions, this scorer parses shell
-structure, so the domains and machinery genuinely differ and "cross-stack
+structure, so the domains and machinery differ and "cross-stack
 parity" would overclaim. The promotion path is a **Python-resident head (or
 sidecar) behind the same `classify_tool_call` contract**, gated by the
 calibration harness at held-out destructive recall 1.0 — when it lands, the

package/docs/reference/evidence-bound-ship.md

@@ -75,7 +75,8 @@ local record, never committed payload.
 
 Per-gate `at` is **reconstructed**, not measured: rungs carry durations but no
 wall-clock finish time, so a gate's stamp is the recording time minus the
-durations of the gates that ran after it. Honest ordering, approximate clock.
+durations of the gates that ran after it. Ordering is preserved; timestamps
+are approximate.
 
 ## Forgery resistance
 
@@ -88,7 +89,7 @@ refused. The deliver CLI writes artifacts; the agent cannot. The trust root is
 
 ## Scope & limitations
 
-This gate is a **local** perimeter, and it is honest about what that means:
+This gate is a **local** perimeter, with the following limits:
 
 - **Enforcement fires only on ship actions routed through the local policy
   chain.** A merge pressed in the GitHub web UI, or a push from a machine

package/docs/reference/semantic-supervisor.md

@@ -109,9 +109,9 @@ Thresholds ship reviewed in `config/supervise-policy.json`:
   "classifierConfidenceMin": 0.7, "classifierTau": 0.6, "maxVerify": 3 }
 ```
 
-Loading is loud: a missing, corrupt, mis-versioned, or out-of-range file
-throws `SupervisorError` and the supervisor refuses to run — it never invents
-defaults. `UAP_SUPERVISE_POLICY` points at an alternate policy file (tests,
+Loading is strict: a missing, corrupt, mis-versioned, or out-of-range file
+throws `SupervisorError` and the supervisor refuses to run — it never
+invents defaults. `UAP_SUPERVISE_POLICY` points at an alternate policy file (tests,
 staged rollouts).
 
 ## Decisions

package/docs/reference/system1-classifier.md

@@ -20,7 +20,7 @@ fixtures work unchanged:
 | `action-class` | choice | `proceed` / `verify-first` / `escalate` / `stop` |
 
 v1 ships **one dimension end-to-end** (escalation risk) per the plan's scope
-discipline; the other two questions ride the same plumbing.
+discipline; the other two questions use the same plumbing.
 
 ## CLI
 
@@ -89,9 +89,9 @@ the thresholds that came from it.
 - **One GPU/CPU probe budget.** Assessments are synchronous and batched per
   call; Wave 1 consumers batch questions per state.
 - **The baseline is question-scoped, not question-blind.** It answers only
-  the built-in questions and refuses others loudly — a custom noul question
-  must not silently inherit the escalation vocabulary. The trained head is
-  the generalizer.
+  the built-in questions and rejects others with an error — a custom noul
+  question must not silently inherit the escalation vocabulary. The trained
+  head is the generalizer.
 - **Kinship:** `src/models/complexity.ts` runs the same
   heuristic-now-model-behind-the-same-interface-later play for task
   complexity. The two should converge deliberately when the trained head

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@miller-tech/uap",
-  "version": "2.11.0",
+  "version": "2.11.1",
   "description": "Autonomous AI agent memory system with CLAUDE.md protocol enforcement",
   "type": "module",
   "main": "dist/index.js",