Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.9.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein is a deterministic orchestrator that automates CLI coding tasks using various adapters. It provides reproducible runs by scheduling them in plain Python and ensures checkable results through an always-on lineage spine and replay journal. The action supports multiple agents, including Claude Code, Codex, Gemini CLI, and 40+ more, with options for local and air-gap installations and auditing receipts offline.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.8.4. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein is a deterministic orchestrator for CLI coding agents that schedules tasks in plain Python and ensures reproducibility and auditability. It uses an always-on lineage spine to track changes and an HMAC-chained audit log to verify results offline, making it suitable for environments with strict requirements on transparency and repeatability.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.8.2. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein orchestrates CLI coding agents deterministically using plain Python, ensuring reproducibility of tasks across runs. It maintains a lineage spine to track every artifact write and replay journal to preserve the execution history. The tool supports various providers like Claude Code, Codex, Gemini CLI, and more, with features for auditing and verifying run results offline.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.8.1. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein orchestrates a crew of CLI coding agents to automate tasks deterministically, ensuring reproducible results end-to-end. It uses plain Python scheduling and record-keeping mechanisms without relying on large language models (LLMs). The action provides features such as deterministic multi-agent orchestration, per-artefact lineage tracking, always-on replay journals, and audit logs for traceability and verification.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.7.1. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein is a deterministic orchestrator for CLI coding agents, scheduling tasks using plain Python to ensure reproducibility. It records every artifact write in an always-on lineage spine and replay journal, providing audit logs with receipts for offline verification.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.7.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein orchestrates a crew of CLI coding agents to execute tasks in parallel against a single goal using deterministic scheduling. It provides an HMAC-signed audit chain with bearer-token authentication, signed agent cards, and per-artefact lineage recording.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.5.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Bernstein automates multi-agent deterministic scheduling of CLIs such as Claude Code, Codex, and Gemini CLI in parallel Git worktrees. It uses an HMAC-signed audit chain to track each step of the process and provides signed agent cards for secure delegation. The lineage feature records every adapter file write, ensuring artifact provenance.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.4.4. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.4.4 Released 2026-07-12. A capability release that adds the experimental MCP Tasks protocol surface to the Bernstein MCP server, plus release-pipeline hardening. MCP Tasks extension Long-running runs are now exposed through the experimental MCP Tasks protocol. A task-capable MCP client (Claude Code, Cursor, Cline, and others) can start a run with bernstein_run, receive a CreateTaskResult, and then poll status and retrieve the result asynchronously through the get_task, get_task_result, list_tasks, and cancel_task handlers without holding a blocking session open. The task handle embeds the run’s audit-chain head hash ({task_id}:{head_hash}) so a stateless client can still tie a progress claim back to the signed chain.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.4.2. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.4.2 Released 2026-07-12. A patch release on top of v3.4.1. It wires the cost-aware batch and cache policies into the live run loop, runs the adapter conformance suite on a real Windows CI runner, and makes the packaged distribution image verifiable. There are no breaking changes and no configuration migration is required.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.3.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.3.0 Released 2026-07-11. An operator-capability release. The dashboard gains authentication with scoped tokens whose every decision lands a governance receipt, a new agy adapter joins the matrix with a nightly conformance canary sealing its probe results, and the run review board projects the run journal into a web view backed by sealed evidence bundles. Groundwork lands for Windows parity and for installing bernstein as an agent skill or plugin.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.2.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.2.0 Released 2026-07-10. A resumability and availability release. Interrupted runs continue from a verified ledger instead of starting over, provider outages reroute along declared fallback chains with a receipt for every decision, and certified local endpoints join the worker pool behind a certification gate.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.1.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.1.0 Released 2026-07-07. A verifiability feature release. Task completion now produces a sealed, content-addressed proof of what passed, not just a status line, and that proof verifies with the same audit chain that seals everything else.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v3.0.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v3.0.0 Released 2026-07-06. A verifiability release. Every major surface in this line ships its primary artifact as a proof: a signed lineage receipt, an HMAC-chained journal entry, a content-addressed record, or a deterministic projection that two operators can recompute to the same bytes. The features are not “capability plus an audit log” bolted together; the audit substrate is the shape of the capability. Strip the chain and the feature loses its meaning, not just its logging.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.16.1. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.16.1 Released 2026-07-06. Fixes Windows: adapter binaries installed as .cmd/.bat shims (for example Codex via nvm-windows) now spawn correctly. The worker resolves the adapter binary to its absolute path before spawning and routes resolved shims through the command interpreter on Windows, and PATHEXT is passed through the worker environment. Real executables and all POSIX spawns are unchanged. (#2287, #2290; thanks @ViteaVlaikov for the report and diagnostics)

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.16.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.16.0 Released 2026-07-05. An output-economy and worker-reliability release: response-style profiles wired end to end, ledger-attributed savings you can recompute, a three-arm cost-and-quality A/B harness, proactive context compaction with signed receipts, schema-enforced worker outcomes, and pinned team manifests. Plus a batch of contributor reliability fixes.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.15.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.15.0 Released 2026-07-05. Feature release built around five community contributions by Shane Mattner (@shanemmattner), plus scheduler and test-infrastructure fixes. Features Hold/release API (/orchestrator/holds): heartbeat-renewed leases that keep the spawner alive while an external workflow driver is active, replacing the fixed quiescence settle timer. TTL is a grace window; expired holds release on their own. Operator docs in docs/operations/HOLDS.md. (#2218, thanks @shanemmattner) Explicit max_turns on TaskCreate: per-task turn budgets flow end to end to the adapter spawn, bypassing the complexity heuristic when set; bounded 1..10000 at the schema boundary; retries carry the value forward. Docs in docs/operations/MAX_TURNS.md. (#2217, thanks @shanemmattner) Per-agent run instrumentation: JSONL records for every LLM call, tool call, and conversation message, captured in real time via SDK hooks and anchored to the project root (not the per-task worktree). Docs in docs/operations/INSTRUMENTATION.md. (#2219, thanks @shanemmattner) Council-of-agents redesign: per-member cost attribution, judge synthesis over candidate outputs, enable_thinking=false injected for Qwen models served from Alibaba Cloud endpoints (hostname-suffix detection), and BERNSTEIN_BUILTIN_ALLOW_RUN_COMMAND passthrough in env isolation. (#2220, thanks @shanemmattner) Inline role_model_policy.<role>.council seed blocks now forward into the runner manifest, behaving identically to the councils/*.yaml file convention. (#2231) Fixes Model selection no longer falls back to hardcoded model strings: every selection flows from configuration, and unconfigured paths raise ModelNotConfiguredError instead of silently guessing. (#2216, thanks @shanemmattner) Critical-path priority boost applies on the first spawn batch, so a low-priority dependency of a high-priority task is claimed first. (#2233) Worker exit codes: when the whole process group receives SIGINT, the worker no longer re-forwards the signal into the child’s interpreter shutdown window, so it reports the child handler’s exit code instead of 130. (#2238) Five integration tests repaired and re-enabled after root-causing failures that predated this cycle. (#2232) Internal Unit coverage for the council runner path. (#2230) Log-injection sanitization on hold fields; scanner findings resolved across instrumentation and council logging. (#2229)

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.14.1. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.14.1 Released 2026-07-04. Security and hygiene patch. Security Log injection: user-controlled values (task ids, roles, session ids, reasons, branches, and git output) are sanitized before they reach a log entry across the task-server routes, task store, spawner, orchestrator, and retrospective paths, so a crafted value can no longer forge log lines. Task-failure/reopen/cancel/block reasons are additionally CR/LF-stripped and length-capped at the request boundary. Sensitive-data logging: the openai_agents adapter and runner no longer log request headers, body, or the resolved API key value; only the env var name is recorded. Fixes The qwen adapter passes --approval-mode yolo, the current documented qwen-code auto-approve flag, clearing the adapter contract drift. (#2197) Internal Restored the integration-test harness (server auth disabled in the fixture, all role templates created, merge-preflight guards satisfied) so the end-to-end orchestration tests pass again. (#2205) Resolved refurb idiom findings across the routes, cost, agents, and orchestration modules.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.14.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.14.0 Released 2026-07-04. Multi-provider orchestration hardening. A large batch of correctness and run-safety fixes for real-codebase runs across mixed providers (Claude, OpenAI, OpenRouter, DeepSeek, MiniMax, Qwen, Gemini), contributed by @shanemmattner and finished to green. Fixes Stalled-manager watchdog no longer misdiagnoses a healthy manager as an auth failure: a manager doing real root-cause investigation before it POSTs its first child task is no longer killed at a hardcoded deadline, the config override is honored, and the failure record names the real cause. (#2179) Failure classifier stops false-positive-killing healthy workers: bare-substring patterns (413, 429, 401, max_tokens, context window) no longer match structured tool-call log data; detection is anchored to real error shapes. (#2183) Janitor acceptance checks tolerate idiomatic worker paths: path_exists honors explicit globs and an opt-in fuzzy basename fallback, so a run where workers placed correct output at repo-idiomatic paths is not cascaded to a false sev1. (#2186) Injected .claude/skills/bernstein-*.md files are excluded from work-branch commits, so they stop causing a merge conflict on every worker merge. (#2187) Per-call token usage is priced and surfaced on the openai_agents provider path, so budget guards are no longer inert on non-Claude runs. Model pricing matches the most specific key first, so mini and flash variants price at their own rate instead of the parent model rate. strict_json_schema is disabled for non-OpenAI models that reject it, with diagnostic logging. Plus the rest of the 22-fix batch across adapters, tasks, cost, routing, quality, and observability. Features Tunable agent run-length limits: max_turns, an error-budget floor, and max_agent_runtime_s, configurable per run. Internal Pricing table extracted into a dependency-free cost.model_prices leaf so adapters can price a call without reaching scheduler internals; public create_pr API preserved; a diagnostic pre-call log that dumped request headers/body verbatim is now redacted.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.13.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.13.0 Released 2026-07-02. Run-safety guardrails and per-role endpoint configuration. Fixes (run safety) GitHub backlog auto-sync is now opt-in and off by default. Previously a run in a repository with open GitHub issues would pull every open issue into the backlog before task scoping, which could silently discard a seeded goal and spawn work against the entire issue list. Enable it explicitly with the github.sync_backlog seed config key (or the BERNSTEIN_SYNC_GITHUB_BACKLOG env override). (#2178) A seeded goal is no longer silently dropped when the backlog is non-empty: the run now prints a loud warning naming the precedence and how to force the goal, instead of quietly planning from the backlog. (#2178) Agent worktree merges refuse to land on the repository default branch. The merge and push path resolves the protected default (origin/HEAD, then init.defaultBranch, then the conventional names, treating both main and master as protected when the remote head is ambiguous) and refuses to merge or push agent work onto it, recording the refusal, so a run started from a default-branch checkout can no longer push unreviewed commits straight to the trunk. (#2178) Features (per-role model configuration) role_model_policy entries gain optional base_url and api_key_env next to model/provider, so different roles can target different OpenAI-compatible endpoints in one workflow (for example a fast manager endpoint and cheaper worker endpoints). api_key_env names an environment variable and is validated against the same fail-closed provider allowlist as the runner. YAML anchors give reuse across roles with no new file format. Absent fields keep today’s behavior. (#2159) ModeProfile gains top_p, top_k, and max_tokens beside its existing temperature, and the previously-deferred apply_mode_to_spawn wiring is completed so a mode profile’s sampling parameters actually reach the spawn and the runner manifest. (#2159) Opt-in builtin tools for the openai_agents runner, for runs without an MCP gateway, selected by tool_source: builtin (the gateway remains the default). read_file, write_file, and list_dir are workdir-confined (absolute and parent-escape paths are rejected). run_command is a restricted process-exec primitive: bare-name commands only, shell interpreters blocked, resolved against PATH, available only under a configured OS sandbox provider or an explicit opt-in; its filesystem confinement is the OS sandbox, not the builtin. Every builtin call is recorded to the run event log so a gateway-free run stays auditable. (#2159) Quality Resolved refurb FURB123 findings in the OWASP control-map builders.

Version updated for https://github.com/sipyourdrink-ltd/bernstein to version v2.11.0. This action is used across all versions by 5 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. What’s Changed v2.11.0 Released 2026-07-02. Features The openai_agents runner accepts optional sampling and endpoint parameters: temperature, top_p, top_k, base_url, and api_key_env on the runner manifest, flowing into the SDK client and model settings. Absent fields keep the previous behavior byte-identical. When base_url is set the runner switches to the chat-completions API and excludes the custom client from tracing, so a third-party key is never sent to the default tracing endpoint. Every effective parameter is logged in the runner start event, so runs stay self-describing. Design validated in daily runs by @shanemmattner (#2159). (#2173) api_key_env is fail-closed: it must name a known LLM provider key from the built-in allowlist; anything else requires the operator to allow it via BERNSTEIN_ALLOWED_API_KEY_ENVS on the host, which a repository cannot set. Requesting sampling parameters on an adapter without the new SUPPORTS_SAMPLING_PARAMS capability fails loudly instead of silently dropping them. (#2173) SDK runners now write heartbeats, so they are visible to the stall watchdog between spawn and exit. (#2173) Fixes The Docker sandbox path from v2.10.0 is hardened: each spawned agent gets its own sandbox session (one exec timeout no longer tears down every agent’s container), committed work is bundled out of the container and fetched into the host repo under sandbox/<session_id> refs, sandbox lifecycle events land in the HMAC-chained audit log with emissions serialized so concurrent lifecycles cannot fork the chain, and provisioning probes task-server reachability and warns on daemons without host networking. (#2162, #2172) The bernstein worker loop can spawn agents again: it constructed the spawner with arguments that never existed and raised TypeError on the first claimed task. The server URL now also reaches spawned agents through the environment allowlist. (#2163, #2171) Dependencies Routine CI action digest updates (github/codeql-action, docker/setup-buildx-action).