Skip to content

Release v0.11.0 #270

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
vicgupta merged 1 commit into from
Oct 31, 2026
Merged

Release v0.11.0 #270

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,10 +1,22 @@
# Changelog

## Unreleased
## [v0.11.0] - 2026-10-31

### Features

- Experimental GitHub Action eval mode (#260): `eval`, `eval-out`, `eval-min-pass-rate`, `cost-limit` and `time-limit` inputs run suites with the existing verified installer, expose report paths, gate CI via `fail-on-error`, and optionally append `report.md` to the job summary. Artifact uploads remain workflow-managed.
- Project rules from agent failures (#254, closes #249): `tinyagent ratchet add|list|remove|path` records rules in `.tinyagent/ratchet.toml` with the incident that caused them and an expiry date or review condition; active rules join the system prompt and can be enforced as permission deny rules.
- Computational checks for verify-on-stop (#255, part of #250): `verify_on_stop.checks` / repeatable `--verify-check <cmd>` run fast deterministic checks (lint, type check, tests), cheapest first, before a run that changed files may finish.
- Inferential review for verify-on-stop (#256, closes #250): `verify_on_stop.review` / `--verify-review` runs a model review last, once every check passes, and reports its findings.
- Session handoff (#257, closes #251): `--handoff` / `handoff.enabled` leaves `.tinyagent/handoff.md` with what a run did and what's next; the next fresh run reads it as data, and an optional startup command runs before the first turn.
- `--update` progress (#258): every step is reported on stderr with a `#` download bar, and failures name the step that failed.
- Brief help (#259): bare `tinyagent` and `tinyagent --help` show two examples and a Common options block; `--help --verbose` lists every flag.
- Experimental GitHub Action eval mode (#261, closes #260): `eval`, `eval-out`, `eval-min-pass-rate`, `cost-limit` and `time-limit` inputs run suites with the existing verified installer, expose report paths, gate CI via `fail-on-error`, and optionally append `report.md` to the job summary. Artifact uploads remain workflow-managed.
- Model routing (#262-#266): `models:` roles `main`, `auxiliary` (compaction, verify review) and `delegated` (`delegate`, `worker`) with `--aux-model`, `--delegate-model` and repeatable `--route task=spec`; per-role `fallback_models`; task-based `models.delegated.routes` (`match`, `type`) with a `model.route` event; `--no-routing`; and `usage_by_role` in trajectories, the `result` event, the summary line, the event log (`run_usage` table) and eval reports. `subagents.model` and `subagents.workers.model` remain as aliases.
- Bubblewrap sandbox engine (#269): `--sandbox-backend auto|landlock|bwrap` / `sandbox_backend` / `tinyagent sandbox --backend`; `auto` uses bwrap with Landlock stacked inside when bwrap works, else Landlock alone.

### Other

- Docs synced for verify-on-stop, token budgets and help changes (#253); README documentation table grouped by topic (#267); new top-level Tasks and Models pages (#268).

## [v0.10.3] - 2026-10-30

Expand Down
Loading