Release 0.11.1

[JacobStephens2]

Summary

0.11.1 adds session Model attribution to Run notifications and stores Session logs beside their owning Command log (#620, #602).

• Run notifications name each session’s Harness, observed Models and configured Effort, including delegated Claude answers, Resumes and child Runs. Missing evidence is labelled; unreadable records warn without discarding readable attribution. Stream Model evidence survives an absent completion report (Include session Models in Run notifications #620, Deepen Session log recording and Model evidence ownership #633).
• Security notifications distinguish failed reproductions from completed audits, identifying the reproduction as refused or failed (Report Security reproduction failures separately from completed audits #604). Deliberately incomplete Security reviews carry their public reason into the Run’s Cause, PR outcome and Hold; private findings remain recorded privately (Surface reasons for deliberate incomplete Security reviews #612).
• Architect notification emails include the Architecture review issue’s title when available, including through dispatch failures and Spec runs (Describe Architect run work in notification emails #636).
• New Session logs go into commands/<command>/ alongside their owning Command log. Dispatched Runs, Tickets and Base fixes inherit that folder; existing logs stay in place. Tools reading sessions/ need the new location (Store Session logs beside their owning Command log #602).
• thirdshift help shrinks from 353 lines to 45, retaining commands, shared options and configuration precedence, with a README link for details (Slim down thirdshift help #623).
• thirdshift setup keeps Security after all Harness settings, including upgrades and repairs, preserving configured values and comments (Keep Setup Security settings outside the Harness area #600).
• Codex code-review sessions request fresh sub-agents given only their task and necessary evidence; the request also applies on Resume (Request fresh Codex reviewers across sessions and Resumes #619).
• Release summaries use the configured default Harness, Model and Effort, support complete inputs above the argument-size limit, and retain GitHub’s generated-notes fallback. Summary launches use restricted tool settings (Use the configured Harness for release notes #605).
• Security finding recording and description operations share a record interface, preserving stored formats, Day shift edits and existing policies (Deepen Security finding recording behind one interface #614, Deepen Security finding record description operations #624, Deepen Security finding record descriptions #625).

Also: Security Harness fallback documentation, README and generated prompt updates (#601, #605, #612, #623, #633, #636).

Version diff

diff --git a/Cargo.lock b/Cargo.lock
index 52cc0c2..956869c 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1469,3 +1469,3 @@ dependencies = [
 name = "thirdshift"
-version = "0.11.0"
+version = "0.11.1"
 dependencies = [
diff --git a/Cargo.toml b/Cargo.toml
index 393e23d..d2968f2 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -2,3 +2,3 @@
 name = "thirdshift"
-version = "0.11.0"
+version = "0.11.1"
 edition = "2024"