fix(#83): use correct variable in new issues confidence gate

[fullsend-ai-coder]

In the new-issues loop of post-scribe.sh, line 376 used ${TOPIC} (set only in the existing-issues loop) instead of ${TITLE} (the loop-local variable). In new_issues_only mode, TOPIC is never assigned, causing an unbound variable crash under set -euo pipefail when a new issue has an invalid confidence value.

Change gate_reject call on line 376 from ${TOPIC} to ${TITLE}, matching the correct usage on line 380 in the same loop. Add a test case that exercises new_issues_only mode with a non-numeric confidence value, which crashed before this fix and now produces a clean gate rejection.

Note: pre-commit could not run shellcheck hook (sandbox network restriction). The low-confidence-rejected test has a pre-existing failure due to missing bc in sandbox.

---

Closes #83

Post-script verification

• Branch is not main/master (agent/83-fix-scribe-confidence-gate)
• Secret scan passed (gitleaks — e017eaa080f40ab8cc347b589b92c0847d571b28..HEAD)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[github-actions]

Functional tests did not run

Functional tests run automatically for org/repo members and collaborators on pull requests.

For other contributors, a maintainer must add the ok-to-test label after the latest push.

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 6:36 PM UTC · Completed 6:43 PM UTC
Commit: d2abdbd · View workflow run →

[fullsend-ai-retro]

PR #84 fixed a copy-paste variable bug in post-scribe.sh (${TOPIC} → ${TITLE}) that was identified by the review agent on PR #10 but merged under a COMMENTED verdict. The retro agent on PR #10 filed issue #83, the code agent created PR #84 with a correct 1-line fix and 1 new test case, and human ralphbean approved and merged — completing the full retro→issue→code→merge loop in ~5 hours. The code change is minimal, correct, and well-tested. CI (script-test, test, commit-lint, DCO) all passed. One improvement proposal: a latent security-gate bypass in the same file (echo flag misinterpretation) was identified by the review agent on PR #10 but never tracked as an issue.

Proposals filed

• Fix echo-flag misinterpretation in post-scribe.sh security gate functions (in fullsend-ai/agents)