Publish configurable 4:3 TUI replay bundles with GIF fallbacks

[konard]

Summary

• forwards aspectRatio and renderer artifactOptions through captureAgentTui;
• gives all six supported agent TUIs an 80-column, 4:3 default capture geometry (80×30 with the renderer's 2:1 cell ratio);
• emits SVG and GIF replay artifacts by default and allows callers to select svg, gif, cast, frames, and transcript outputs;
• documents renderer sizing, padding, theme colors, and explicit borderRadius: 0;
• patches command-stream@0.17.0's CommonJS/ESM gifenc normalization until the upstream correction tracked in GIF artifact rendering fails on Node 20: GIFEncoder is not a function link-foundation/command-stream#183 is released.

Reproduction

Before this change, capture geometry was forwarded as undefined unless callers hardcoded both dimensions, artifactOptions was discarded, and the replay bundle did not provide a GIF fallback.

The first implementation exposed a second reproducible problem on Node 20 across Ubuntu, macOS, and Windows:

TypeError: GIFEncoder is not a function
    at renderGif (.../command-stream/src/terminal-artifacts.mjs:429:15)

gifenc is CommonJS and Node exposes its API under the dynamic import's default namespace, while Bun exposes named exports. The included dependency patch prefers named exports and falls back to default; clean npm ci installs apply it automatically.

Tests

js/test/tui.test.mjs verifies:

• every supported agent defaults to an 80×30 initial PTY;
• explicit dimensions remain unchanged;
• renderer options reach generated SVG (borderRadius: 0, colors, font, and cell sizing);
• the default replay bundle contains a non-empty GIF;
• selective bundles remove formats the caller did not request.

Local validation:

npm ci                    command-stream@0.17.0 patch applied
npm test                  239 passed
bun test                  239 passed
npm run lint              passed (pre-existing warnings only)
npm run format:check      passed
npm run check:duplication passed
npm pack --dry-run        patch included in package

Fixes #46

[konard]

Implementation and regression coverage are complete locally, but final reproducible Node CI is blocked by an upstream command-stream@0.17.0 GIF interop defect: link-foundation/command-stream#183. The exact fix is a one-line CommonJS/ESM namespace normalization; with that applied only to the local installed dependency, all 239 JavaScript tests pass. Waiting for an upstream release rather than checking a patch into this package.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.6 Sol
• Provider: OpenAI
• Public pricing estimate: $9.129109

📊 Context and tokens usage:

• 184.7K / 200K (92%) input tokens, 17.7K / 128K (14%) output tokens

Total: (184.7K + 6.5M cached) input tokens, 17.7K output tokens, $9.129109 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.6-sol
• Thinking level: low (~7999 tokens)
• Model: GPT 5.6 Sol (gpt-5.6-sol)

📎 Log file uploaded as Gist (2126KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🔄 Auto-restart triggered (iteration 1)

Reason: CI failures detected

Starting new session to address the issues.

---

Auto-restart-until-mergeable mode is active. This run will stop after 5 restart iterations.

[konard]

Working session summary

Fixed and pushed to PR 47.

• Patched command-stream@0.17.0 GIF import interoperability for Node and Bun.
• Verified clean npm ci applies the patch.
• Full Node and Bun suites pass: 239 tests each.
• Lint, formatting, duplication, and package-content checks pass.
• All fresh JavaScript and Rust CI runs pass across Ubuntu, macOS, and Windows.
• PR is ready, mergeable, and has a clean working tree.

Commit: c36c5c3

---

This summary was automatically extracted from the AI working session output.

[konard]

🔄 Auto-restart-until-mergeable Log (iteration 1)

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.6 Sol
• Provider: OpenAI
• Public pricing estimate: $7.309405
• Token usage: 147,603 input, 12,411 output, 2,646 reasoning, 5,274,880 cache read

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.6-sol
• Model: GPT 5.6 Sol (gpt-5.6-sol)

📎 Log file uploaded as Gist (3961KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

🎉 Auto-merged

This pull request has been automatically merged by hive-mind.

• All CI checks have passed

---

Auto-merged by hive-mind with --auto-merge flag