PyPI - bernstein - Versions diffs - 1.2.2__tar.gz → 1.3.2__tar.gz - Mend

bernstein 1.2.2tar.gz → 1.3.2tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (475) hide show

{bernstein-1.2.2 → bernstein-1.3.2}/PKG-INFO RENAMED Viewed

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bernstein
-Version: 1.2.2
+Version: 1.3.2
 Summary: Declarative agent orchestration for engineering teams
 Author-email: Alex Chernysh <alex@alexchernysh.com>
 License-Expression: Apache-2.0
@@ -48,11 +48,14 @@ Description-Content-Type: text/markdown
 ### One YAML. Multiple coding agents. Ship while you sleep.
 <picture>
-  <source media="(prefers-color-scheme: dark)" srcset="docs/assets/dashboard.svg">
-  <source media="(prefers-color-scheme: light)" srcset="docs/assets/dashboard.svg">
-  <img alt="Bernstein Dashboard" src="docs/assets/dashboard.svg" width="700">
+  <source media="(prefers-color-scheme: dark)" srcset="docs/assets/tui.svg">
+  <source media="(prefers-color-scheme: light)" srcset="docs/assets/tui.svg">
+  <img alt="Bernstein TUI — live task dashboard" src="docs/assets/tui.svg" width="700">
 </picture>
+<p align="center"><strong>Web dashboard</strong> — real-time task monitoring, cost tracking, agent status</p>
+<p align="center"><img alt="Bernstein Web Dashboard" src="docs/assets/web-dashboard.png" width="700" style="border-radius:8px"></p>
 [![Python 3.12+](https://img.shields.io/badge/python-3.12+-3776ab?logo=python&logoColor=white)](https://python.org)
 [![Tests](https://img.shields.io/badge/tests-4250+-2ea44f)]()
 [![License](https://img.shields.io/badge/license-Apache_2.0-blue)](LICENSE)
@@ -77,7 +80,7 @@ bernstein -g "Add JWT auth with refresh tokens, tests, and API docs"
 Bernstein takes a goal, breaks it into tasks, assigns them to AI coding agents running in parallel, verifies the output, and commits the results. You come back to working code, passing tests, and a clean git history.
-**No framework to learn. No vendor lock-in.** If you have [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex CLI](https://github.com/openai/codex), [Gemini CLI](https://github.com/google-gemini/gemini-cli), or [Qwen](https://github.com/QwenLM/Qwen-Agent) installed, Bernstein uses them. Mix models in the same run — cheap free-tier agents for boilerplate, heavy models for architecture. Switch providers without rewriting anything. Agents spawn, work, exit. No context drift. No babysitting.
+**No framework to learn. No vendor lock-in.** If you have [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex CLI](https://github.com/openai/codex), [Cursor](https://www.cursor.com), [Gemini CLI](https://github.com/google-gemini/gemini-cli), or [Qwen](https://github.com/QwenLM/Qwen-Agent) installed, Bernstein uses them. Mix models in the same run — cheap free-tier agents for boilerplate, heavy models for architecture. Switch providers without rewriting anything. Agents spawn, work, exit. No context drift. No babysitting.
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="docs/assets/architecture.svg">
@@ -131,50 +134,122 @@ See [`examples/quickstart/`](examples/quickstart/) for a ready-to-run example wi
 <summary><strong>All CLI commands</strong></summary>
 ```bash
-bernstein stop                     # graceful shutdown
+# Core
+bernstein -g "goal"                # start orchestration with inline goal
+bernstein                          # start from bernstein.yaml
+bernstein init                     # initialize .sdd/ workspace
+bernstein stop [--timeout N]       # graceful shutdown (--force for hard kill)
 bernstein ps                       # show running agent processes
-bernstein cancel <task_id>         # cancel a task
-bernstein cost                     # show cost summary
-bernstein live                     # open live TUI dashboard
+bernstein status                   # task summary, active agents, cost estimate
+bernstein cost [--json] [--share]  # show cost, tokens, duration per model
+bernstein demo [--real]            # zero-config demo (--real uses live agents)
+bernstein quickstart               # zero-config demo: 3 tasks on Flask TODO API
+# Monitoring
+bernstein live [--classic]         # live TUI dashboard
 bernstein dashboard                # open web dashboard in browser
-bernstein doctor                   # health check: deps, keys, ports
-bernstein plugins                  # list active plugins
+bernstein doctor [--fix]           # health check: deps, keys, ports (--fix auto-repairs)
+bernstein logs [-f] [-a AGENT]     # tail agent log output
 bernstein trace <task_id>          # step-by-step agent decision trace
-bernstein replay <trace_id>        # re-run a task from its trace
-bernstein init                     # initialize project
-bernstein demo                     # zero-to-running demo (no API key needed)
+bernstein replay <run_id>          # re-run from a recorded trace
+bernstein recap [--as-json]        # post-run summary: tasks, pass/fail, cost
+bernstein diff <task_id>           # git diff for what an agent changed
+# Task management
+bernstein plan [--export]          # show task backlog (--export to JSON)
+bernstein add-task TITLE           # inject a task into the running server
+bernstein cancel <task_id>         # cancel a running or queued task
+bernstein list-tasks [--status-filter] # list tasks with optional filters
+bernstein sync                     # sync .sdd/backlog/open/ with task server
+bernstein review                   # trigger immediate manager queue review
+bernstein approve <task_id>        # approve a pending task review
+bernstein reject <task_id>         # reject a pending task review
+bernstein pending                  # list tasks waiting for approval
+# Agent management
+bernstein agents sync              # pull latest agent catalog
+bernstein agents list              # list available agents
+bernstein agents validate          # check catalog health
+bernstein agents showcase          # show agent capabilities showcase
+bernstein agents match QUERY       # find best agent for a task description
+bernstein agents discover          # discover installed CLI agents
 # CI autofix
 bernstein ci fix <run-url>         # parse failing CI run, create fix task
 bernstein ci watch <repo>          # continuous monitoring, auto-fix on failure
 # Governance & audit
+bernstein audit show               # recent audit log events
+bernstein audit seal               # create Merkle seal of audit log
+bernstein audit verify             # verify Merkle proof integrity
 bernstein audit verify-hmac        # validate HMAC chain integrity
 bernstein audit query              # search audit log (--event-type, --actor, --since)
 bernstein verify --wal-integrity   # verify WAL hash chain
 bernstein verify --determinism     # check execution fingerprint reproducibility
+bernstein verify --memory-audit    # detect memory leaks in agent processes
+bernstein verify --formal          # formal verification mode
+bernstein manifest list            # list all run manifests
 bernstein manifest show <run-id>   # display run manifest
-bernstein manifest diff <a> <b>    # compare two run configurations
+bernstein manifest diff <a> <b>   # compare two run configurations
 # Benchmarks & eval
 bernstein benchmark run            # run golden benchmark suite
 bernstein benchmark compare        # orchestrated vs. single-agent comparison
+bernstein benchmark swe-bench      # run SWE-bench harness
+bernstein eval run                 # run evaluation suite with scoring
+bernstein eval report              # generate evaluation report
+bernstein eval failures            # show evaluation failures
 # Evolution
+bernstein evolve run               # run autoresearch evolution loop
 bernstein evolve review            # list evolution proposals
 bernstein evolve approve <id>      # approve a proposal
-bernstein ideate                   # run creative evolution pipeline
-# Agent management
-bernstein agents sync              # pull latest agent catalog
-bernstein agents list              # list available agents
-bernstein agents validate          # check catalog health
-# Workspace
+bernstein evolve status            # show evolution pipeline status
+bernstein evolve export            # export evolution proposals
+bernstein ideate                   # creative evolution pipeline
+# Advanced
+bernstein chaos agent-kill         # kill a random agent (fault injection)
+bernstein chaos rate-limit         # simulate API rate limiting
+bernstein chaos file-remove        # remove files an agent depends on
+bernstein chaos status             # show chaos experiment status
+bernstein chaos slo                # check SLO compliance under chaos
+bernstein gateway start            # start MCP gateway proxy
+bernstein gateway replay <run-id>  # replay recorded MCP tool calls
+bernstein workflow validate FILE   # validate a workflow YAML
+bernstein workflow list            # list workflow DSL files
+bernstein workflow show NAME       # show workflow details
+bernstein mcp                      # run Bernstein as an MCP server
+bernstein watch [DIR]              # monitor directory, re-run on changes
+bernstein listen                   # voice command session (offline STT)
+bernstein checkpoint [--goal]      # snapshot session progress
+bernstein wrap-up [--stop]         # end session with structured brief
+# Configuration & workspace
 bernstein workspace                # show multi-repo workspace status
-bernstein plan                     # show task backlog
-bernstein logs                     # tail agent log output
-bernstein retro                    # generate retrospective report
+bernstein workspace clone          # clone missing repos
+bernstein workspace validate       # check workspace health
+bernstein config set KEY VALUE     # set a global config value
+bernstein config get KEY           # show effective config value
+bernstein config list              # list all config keys
+bernstein config validate          # validate project configuration
+# Utilities
+bernstein retro [--since H]        # generate retrospective report
+bernstein plugins                  # list active plugins
+bernstein install-hooks            # install git hooks
+bernstein completions              # generate shell completion scripts
+bernstein self-update [--check]    # upgrade from PyPI (--rollback to revert)
+bernstein worker --server URL      # join a cluster as a worker node
+bernstein help-all                 # comprehensive help for all commands
+# GitHub integration
+bernstein github setup             # configure GitHub App integration
+bernstein github test-webhook      # verify webhook configuration
+# Quarantine
+bernstein quarantine list          # list quarantined tasks
+bernstein quarantine clear         # clear quarantine
 ```
 </details>
@@ -214,6 +289,7 @@ Analyzes metrics, proposes changes to prompts and routing rules, sandboxes them,
 | [Amp](https://ampcode.com) | Sourcegraph | opus 4.6, gpt-5.4 | `--cli amp` | `brew install amp` |
 | [Claude Code](https://docs.anthropic.com/en/docs/claude-code) | Anthropic | opus 4.6, sonnet 4.6, haiku 4.5 | `--cli claude` | `npm install -g @anthropic-ai/claude-code` |
 | [Codex CLI](https://github.com/openai/codex) | OpenAI | gpt-5.4, o3, o4-mini | `--cli codex` | `npm install -g @openai/codex` |
+| [Cursor](https://www.cursor.com) | Cursor AI | sonnet 4.6, opus 4.6, gpt-5.4 | `--cli cursor` | Cursor app (sign in via app) |
 | [Gemini CLI](https://github.com/google-gemini/gemini-cli) | Google | gemini-3-pro, 3-flash | `--cli gemini` | `npm install -g @google/gemini-cli` |
 | [Qwen](https://github.com/QwenLM/Qwen-Agent) | Alibaba / OpenRouter | qwen3-coder, qwen-max | `--cli qwen` | `npm install -g qwen-code` |
 | [Roo Code](https://github.com/RooVetGit/Roo-Code) | Anthropic / OpenAI / any | opus 4.6, sonnet 4.6, gpt-4o | `--cli roo-code` | VS Code extension (headless CLI) |
@@ -281,6 +357,27 @@ CrewAI, AutoGen, and LangGraph work with any model via API wrappers — but they
 </details>
+## Noteworthy capabilities
+Beyond core orchestration, Bernstein ships several features that are useful but easy to miss in `--help` output:
+| Capability | Command | What it does |
+|------------|---------|--------------|
+| **Chaos engineering** | `bernstein chaos` | Fault injection (kill agents, rate-limit APIs, remove files) to test resilience. SLO compliance checks. |
+| **Cryptographic audit** | `bernstein audit seal/verify` | Tamper-evident execution logs with Merkle proofs. HMAC chain integrity verification. |
+| **Gateway proxy** | `bernstein gateway start` | Transparent MCP gateway that records and replays tool calls between agents and providers. |
+| **Workflow DSL** | `bernstein workflow` | Declarative YAML-based task pipelines with validation, listing, and inspection. |
+| **MCP server mode** | `bernstein mcp` | Runs Bernstein itself as an MCP tool server, exposable to other agents or editors. |
+| **Trace & replay** | `bernstein trace/replay` | Record agent decisions step-by-step, then deterministically replay any run. |
+| **Formal verification** | `bernstein verify --formal` | Experimental formal verification of execution invariants. |
+| **File watching** | `bernstein watch` | Monitors directories and re-triggers task execution on file changes. |
+| **Voice commands** | `bernstein listen` | Offline speech-to-text mapped to CLI commands. Experimental. |
+| **Cluster worker** | `bernstein worker` | Join a remote Bernstein cluster as a worker node for distributed orchestration. |
+| **Task delegation** | (not yet wired) | Agents and scripts can submit tasks programmatically via `delegate` command. |
+| **Event triggers** | (not yet wired) | Define event-driven triggers (file changes, webhooks, schedules) that create tasks automatically. |
+These features vary in maturity. Chaos engineering, audit, and trace/replay are production-tested. Voice commands and formal verification are experimental. See `docs/FEATURE_MATRIX.md` for full documentation coverage status.
 ## Observability
 ```bash

{bernstein-1.2.2 → bernstein-1.3.2}/README.md RENAMED Viewed

@@ -12,11 +12,14 @@
 ### One YAML. Multiple coding agents. Ship while you sleep.
 <picture>
-  <source media="(prefers-color-scheme: dark)" srcset="docs/assets/dashboard.svg">
-  <source media="(prefers-color-scheme: light)" srcset="docs/assets/dashboard.svg">
-  <img alt="Bernstein Dashboard" src="docs/assets/dashboard.svg" width="700">
+  <source media="(prefers-color-scheme: dark)" srcset="docs/assets/tui.svg">
+  <source media="(prefers-color-scheme: light)" srcset="docs/assets/tui.svg">
+  <img alt="Bernstein TUI — live task dashboard" src="docs/assets/tui.svg" width="700">
 </picture>
+<p align="center"><strong>Web dashboard</strong> — real-time task monitoring, cost tracking, agent status</p>
+<p align="center"><img alt="Bernstein Web Dashboard" src="docs/assets/web-dashboard.png" width="700" style="border-radius:8px"></p>
 [![Python 3.12+](https://img.shields.io/badge/python-3.12+-3776ab?logo=python&logoColor=white)](https://python.org)
 [![Tests](https://img.shields.io/badge/tests-4250+-2ea44f)]()
 [![License](https://img.shields.io/badge/license-Apache_2.0-blue)](LICENSE)
@@ -41,7 +44,7 @@ bernstein -g "Add JWT auth with refresh tokens, tests, and API docs"
 Bernstein takes a goal, breaks it into tasks, assigns them to AI coding agents running in parallel, verifies the output, and commits the results. You come back to working code, passing tests, and a clean git history.
-**No framework to learn. No vendor lock-in.** If you have [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex CLI](https://github.com/openai/codex), [Gemini CLI](https://github.com/google-gemini/gemini-cli), or [Qwen](https://github.com/QwenLM/Qwen-Agent) installed, Bernstein uses them. Mix models in the same run — cheap free-tier agents for boilerplate, heavy models for architecture. Switch providers without rewriting anything. Agents spawn, work, exit. No context drift. No babysitting.
+**No framework to learn. No vendor lock-in.** If you have [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex CLI](https://github.com/openai/codex), [Cursor](https://www.cursor.com), [Gemini CLI](https://github.com/google-gemini/gemini-cli), or [Qwen](https://github.com/QwenLM/Qwen-Agent) installed, Bernstein uses them. Mix models in the same run — cheap free-tier agents for boilerplate, heavy models for architecture. Switch providers without rewriting anything. Agents spawn, work, exit. No context drift. No babysitting.
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="docs/assets/architecture.svg">
@@ -95,50 +98,122 @@ See [`examples/quickstart/`](examples/quickstart/) for a ready-to-run example wi
 <summary><strong>All CLI commands</strong></summary>
 ```bash
-bernstein stop                     # graceful shutdown
+# Core
+bernstein -g "goal"                # start orchestration with inline goal
+bernstein                          # start from bernstein.yaml
+bernstein init                     # initialize .sdd/ workspace
+bernstein stop [--timeout N]       # graceful shutdown (--force for hard kill)
 bernstein ps                       # show running agent processes
-bernstein cancel <task_id>         # cancel a task
-bernstein cost                     # show cost summary
-bernstein live                     # open live TUI dashboard
+bernstein status                   # task summary, active agents, cost estimate
+bernstein cost [--json] [--share]  # show cost, tokens, duration per model
+bernstein demo [--real]            # zero-config demo (--real uses live agents)
+bernstein quickstart               # zero-config demo: 3 tasks on Flask TODO API
+# Monitoring
+bernstein live [--classic]         # live TUI dashboard
 bernstein dashboard                # open web dashboard in browser
-bernstein doctor                   # health check: deps, keys, ports
-bernstein plugins                  # list active plugins
+bernstein doctor [--fix]           # health check: deps, keys, ports (--fix auto-repairs)
+bernstein logs [-f] [-a AGENT]     # tail agent log output
 bernstein trace <task_id>          # step-by-step agent decision trace
-bernstein replay <trace_id>        # re-run a task from its trace
-bernstein init                     # initialize project
-bernstein demo                     # zero-to-running demo (no API key needed)
+bernstein replay <run_id>          # re-run from a recorded trace
+bernstein recap [--as-json]        # post-run summary: tasks, pass/fail, cost
+bernstein diff <task_id>           # git diff for what an agent changed
+# Task management
+bernstein plan [--export]          # show task backlog (--export to JSON)
+bernstein add-task TITLE           # inject a task into the running server
+bernstein cancel <task_id>         # cancel a running or queued task
+bernstein list-tasks [--status-filter] # list tasks with optional filters
+bernstein sync                     # sync .sdd/backlog/open/ with task server
+bernstein review                   # trigger immediate manager queue review
+bernstein approve <task_id>        # approve a pending task review
+bernstein reject <task_id>         # reject a pending task review
+bernstein pending                  # list tasks waiting for approval
+# Agent management
+bernstein agents sync              # pull latest agent catalog
+bernstein agents list              # list available agents
+bernstein agents validate          # check catalog health
+bernstein agents showcase          # show agent capabilities showcase
+bernstein agents match QUERY       # find best agent for a task description
+bernstein agents discover          # discover installed CLI agents
 # CI autofix
 bernstein ci fix <run-url>         # parse failing CI run, create fix task
 bernstein ci watch <repo>          # continuous monitoring, auto-fix on failure
 # Governance & audit
+bernstein audit show               # recent audit log events
+bernstein audit seal               # create Merkle seal of audit log
+bernstein audit verify             # verify Merkle proof integrity
 bernstein audit verify-hmac        # validate HMAC chain integrity
 bernstein audit query              # search audit log (--event-type, --actor, --since)
 bernstein verify --wal-integrity   # verify WAL hash chain
 bernstein verify --determinism     # check execution fingerprint reproducibility
+bernstein verify --memory-audit    # detect memory leaks in agent processes
+bernstein verify --formal          # formal verification mode
+bernstein manifest list            # list all run manifests
 bernstein manifest show <run-id>   # display run manifest
-bernstein manifest diff <a> <b>    # compare two run configurations
+bernstein manifest diff <a> <b>   # compare two run configurations
 # Benchmarks & eval
 bernstein benchmark run            # run golden benchmark suite
 bernstein benchmark compare        # orchestrated vs. single-agent comparison
+bernstein benchmark swe-bench      # run SWE-bench harness
+bernstein eval run                 # run evaluation suite with scoring
+bernstein eval report              # generate evaluation report
+bernstein eval failures            # show evaluation failures
 # Evolution
+bernstein evolve run               # run autoresearch evolution loop
 bernstein evolve review            # list evolution proposals
 bernstein evolve approve <id>      # approve a proposal
-bernstein ideate                   # run creative evolution pipeline
-# Agent management
-bernstein agents sync              # pull latest agent catalog
-bernstein agents list              # list available agents
-bernstein agents validate          # check catalog health
-# Workspace
+bernstein evolve status            # show evolution pipeline status
+bernstein evolve export            # export evolution proposals
+bernstein ideate                   # creative evolution pipeline
+# Advanced
+bernstein chaos agent-kill         # kill a random agent (fault injection)
+bernstein chaos rate-limit         # simulate API rate limiting
+bernstein chaos file-remove        # remove files an agent depends on
+bernstein chaos status             # show chaos experiment status
+bernstein chaos slo                # check SLO compliance under chaos
+bernstein gateway start            # start MCP gateway proxy
+bernstein gateway replay <run-id>  # replay recorded MCP tool calls
+bernstein workflow validate FILE   # validate a workflow YAML
+bernstein workflow list            # list workflow DSL files
+bernstein workflow show NAME       # show workflow details
+bernstein mcp                      # run Bernstein as an MCP server
+bernstein watch [DIR]              # monitor directory, re-run on changes
+bernstein listen                   # voice command session (offline STT)
+bernstein checkpoint [--goal]      # snapshot session progress
+bernstein wrap-up [--stop]         # end session with structured brief
+# Configuration & workspace
 bernstein workspace                # show multi-repo workspace status
-bernstein plan                     # show task backlog
-bernstein logs                     # tail agent log output
-bernstein retro                    # generate retrospective report
+bernstein workspace clone          # clone missing repos
+bernstein workspace validate       # check workspace health
+bernstein config set KEY VALUE     # set a global config value
+bernstein config get KEY           # show effective config value
+bernstein config list              # list all config keys
+bernstein config validate          # validate project configuration
+# Utilities
+bernstein retro [--since H]        # generate retrospective report
+bernstein plugins                  # list active plugins
+bernstein install-hooks            # install git hooks
+bernstein completions              # generate shell completion scripts
+bernstein self-update [--check]    # upgrade from PyPI (--rollback to revert)
+bernstein worker --server URL      # join a cluster as a worker node
+bernstein help-all                 # comprehensive help for all commands
+# GitHub integration
+bernstein github setup             # configure GitHub App integration
+bernstein github test-webhook      # verify webhook configuration
+# Quarantine
+bernstein quarantine list          # list quarantined tasks
+bernstein quarantine clear         # clear quarantine
 ```
 </details>
@@ -178,6 +253,7 @@ Analyzes metrics, proposes changes to prompts and routing rules, sandboxes them,
 | [Amp](https://ampcode.com) | Sourcegraph | opus 4.6, gpt-5.4 | `--cli amp` | `brew install amp` |
 | [Claude Code](https://docs.anthropic.com/en/docs/claude-code) | Anthropic | opus 4.6, sonnet 4.6, haiku 4.5 | `--cli claude` | `npm install -g @anthropic-ai/claude-code` |
 | [Codex CLI](https://github.com/openai/codex) | OpenAI | gpt-5.4, o3, o4-mini | `--cli codex` | `npm install -g @openai/codex` |
+| [Cursor](https://www.cursor.com) | Cursor AI | sonnet 4.6, opus 4.6, gpt-5.4 | `--cli cursor` | Cursor app (sign in via app) |
 | [Gemini CLI](https://github.com/google-gemini/gemini-cli) | Google | gemini-3-pro, 3-flash | `--cli gemini` | `npm install -g @google/gemini-cli` |
 | [Qwen](https://github.com/QwenLM/Qwen-Agent) | Alibaba / OpenRouter | qwen3-coder, qwen-max | `--cli qwen` | `npm install -g qwen-code` |
 | [Roo Code](https://github.com/RooVetGit/Roo-Code) | Anthropic / OpenAI / any | opus 4.6, sonnet 4.6, gpt-4o | `--cli roo-code` | VS Code extension (headless CLI) |
@@ -245,6 +321,27 @@ CrewAI, AutoGen, and LangGraph work with any model via API wrappers — but they
 </details>
+## Noteworthy capabilities
+Beyond core orchestration, Bernstein ships several features that are useful but easy to miss in `--help` output:
+| Capability | Command | What it does |
+|------------|---------|--------------|
+| **Chaos engineering** | `bernstein chaos` | Fault injection (kill agents, rate-limit APIs, remove files) to test resilience. SLO compliance checks. |
+| **Cryptographic audit** | `bernstein audit seal/verify` | Tamper-evident execution logs with Merkle proofs. HMAC chain integrity verification. |
+| **Gateway proxy** | `bernstein gateway start` | Transparent MCP gateway that records and replays tool calls between agents and providers. |
+| **Workflow DSL** | `bernstein workflow` | Declarative YAML-based task pipelines with validation, listing, and inspection. |
+| **MCP server mode** | `bernstein mcp` | Runs Bernstein itself as an MCP tool server, exposable to other agents or editors. |
+| **Trace & replay** | `bernstein trace/replay` | Record agent decisions step-by-step, then deterministically replay any run. |
+| **Formal verification** | `bernstein verify --formal` | Experimental formal verification of execution invariants. |
+| **File watching** | `bernstein watch` | Monitors directories and re-triggers task execution on file changes. |
+| **Voice commands** | `bernstein listen` | Offline speech-to-text mapped to CLI commands. Experimental. |
+| **Cluster worker** | `bernstein worker` | Join a remote Bernstein cluster as a worker node for distributed orchestration. |
+| **Task delegation** | (not yet wired) | Agents and scripts can submit tasks programmatically via `delegate` command. |
+| **Event triggers** | (not yet wired) | Define event-driven triggers (file changes, webhooks, schedules) that create tasks automatically. |
+These features vary in maturity. Chaos engineering, audit, and trace/replay are production-tested. Voice commands and formal verification are experimental. See `docs/FEATURE_MATRIX.md` for full documentation coverage status.
 ## Observability
 ```bash

{bernstein-1.2.2 → bernstein-1.3.2}/pyproject.toml RENAMED Viewed

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 [project]
 name = "bernstein"
-version = "1.2.2"
+version = "1.3.2"
 description = "Declarative agent orchestration for engineering teams"
 readme = "README.md"
 requires-python = ">=3.12"
@@ -145,6 +145,11 @@ exclude = [
     "src/bernstein/core/workflow_dsl.py",
     "src/bernstein/cli/self_update_cmd.py",
     "src/bernstein/tui/widgets.py",
+    "src/bernstein/cli/diff_cmd.py",
+    "src/bernstein/cli/voice_cmd.py",
+    "src/bernstein/cli/worker_cmd.py",
+    "src/bernstein/core/formal_verification.py",
+    "src/bernstein/core/seed.py",
 ]
 [tool.pytest.ini_options]

bernstein-1.3.2/src/bernstein/adapters/cursor.py ADDED Viewed

@@ -0,0 +1,111 @@
+"""Cursor Agent CLI adapter."""
+from __future__ import annotations
+import json
+import subprocess
+from typing import TYPE_CHECKING, Any
+if TYPE_CHECKING:
+    from pathlib import Path
+from bernstein.adapters.base import CLIAdapter, SpawnResult, build_worker_cmd
+from bernstein.adapters.env_isolation import build_filtered_env
+from bernstein.core.models import ApiTier, ApiTierInfo, ModelConfig, ProviderType, RateLimit
+class CursorAdapter(CLIAdapter):
+    """Spawn and monitor Cursor Agent CLI sessions."""
+    def spawn(
+        self,
+        *,
+        prompt: str,
+        workdir: Path,
+        model_config: ModelConfig,
+        session_id: str,
+        mcp_config: dict[str, Any] | None = None,
+    ) -> SpawnResult:
+        log_path = workdir / ".sdd" / "runtime" / f"{session_id}.log"
+        log_path.parent.mkdir(parents=True, exist_ok=True)
+        cmd = ["cursor", "agent"]
+        # Session isolation via separate user data directory
+        data_dir = workdir / ".sdd" / "runtime" / "cursor" / session_id
+        data_dir.mkdir(parents=True, exist_ok=True)
+        cmd += ["--user-data-dir", str(data_dir)]
+        # MCP config injection
+        if mcp_config:
+            cmd += ["--add-mcp", json.dumps(mcp_config)]
+        # Prompt via positional argument (cursor agent reads it directly)
+        cmd.append(prompt)
+        # Wrap with bernstein-worker for process visibility
+        pid_dir = workdir / ".sdd" / "runtime" / "pids"
+        wrapped_cmd = build_worker_cmd(
+            cmd,
+            role=session_id.rsplit("-", 1)[0],
+            session_id=session_id,
+            pid_dir=pid_dir,
+            model=model_config.model,
+        )
+        # Cursor inherits OAuth session from ~/.cursor/ — no extra env keys needed
+        env = build_filtered_env([])
+        with log_path.open("w") as log_file:
+            try:
+                proc = subprocess.Popen(
+                    wrapped_cmd,
+                    cwd=workdir,
+                    env=env,
+                    stdout=log_file,
+                    stderr=subprocess.STDOUT,
+                    start_new_session=True,
+                )
+            except FileNotFoundError as exc:
+                raise RuntimeError("cursor not found in PATH. Install from https://cursor.com") from exc
+            except PermissionError as exc:
+                raise RuntimeError(f"Permission denied executing cursor: {exc}") from exc
+        return SpawnResult(pid=proc.pid, log_path=log_path)
+    def name(self) -> str:
+        return "Cursor"
+    def detect_tier(self) -> ApiTierInfo | None:
+        """Detect Cursor subscription tier from ~/.cursor/ config.
+        Cursor subscription tiers:
+        - Free: 50 slow requests/month
+        - Pro: $20/mo, 500 fast requests + unlimited slow
+        - Business: $40/mo, unlimited fast requests
+        Since Cursor does not expose subscription tier via CLI, we check
+        for the presence of the auth directory as a proxy for being logged in.
+        The tier is reported as PRO when logged in (most common paid tier).
+        Returns:
+            ApiTierInfo with detected tier and rate limits, or None if not logged in.
+        """
+        from pathlib import Path
+        cursor_dir = Path.home() / ".cursor"
+        if not cursor_dir.exists():
+            return None
+        # Cursor Pro is the most common paid tier — conservative estimate
+        tier = ApiTier.PRO
+        rate_limit = RateLimit(
+            requests_per_minute=50,  # 500 fast req/month ≈ ~50/min burst
+            tokens_per_minute=20_000,
+        )
+        return ApiTierInfo(
+            provider=ProviderType.CURSOR,
+            tier=tier,
+            rate_limit=rate_limit,
+            is_active=True,
+        )

bernstein 1.2.2__tar.gz → 1.3.2__tar.gz

bernstein 1.2.2tar.gz → 1.3.2tar.gz