fix(#2632): use SKILL.md frontmatter name for remote skill display

[fullsend-ai-coder]

Remote skills loaded via URL were logged as "tree" during sandbox bootstrap because filepath.Base() on the cache path returns the last segment of GitHub's /tree/ URL convention. Add resolveSkillDisplayName() that reads SKILL.md frontmatter via skill.ParseFrontmatter() and uses the name field when available, falling back to filepath.Base() for local skills or when no frontmatter is present.

Note: pre-commit could not run in sandbox (shellcheck network error). go vet and go test passed.

---

Closes #2632

Post-script verification

• Branch is not main/master (agent/2632-skill-display-name)
• Secret scan passed (gitleaks — 10b62b5510e1c8a22ed08ad0b1061aa346dd1373..HEAD)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[github-actions]

E2E tests did not run

E2E 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.

See E2E testing guide for details.

[github-actions]

Site preview

Preview: https://644b2de9-site.fullsend-ai.workers.dev

Commit: 92529462528bcd93fa33e6b93ec4aea9502adc32

[codecov]

Codecov Report

❌ Patch coverage is 90.00000% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
internal/runtime/claude.go	90.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[ralphbean]

LGTM.

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 9:09 PM UTC · Completed 9:17 PM UTC
Commit: 9252946 · View workflow run →

[fullsend-ai-retro]

Retro: PR #2633 — fix remote skill display name

Workflow quality: Excellent. This was a clean, efficient end-to-end workflow with zero rework cycles.

Timeline

Time (UTC)	Event
19:59	Issue #2632 filed by ascerra — detailed bug report with root cause, suggested fix, and production evidence
20:03–20:06	Triage agent ran (3 min) — confirmed low-severity bug, provided clear fix recommendation with test cases
20:10–20:17	Code agent ran (7 min) — produced PR #2633 with 1 commit (+73/−1, 2 files)
20:17	PR opened — CI bots posted: E2E gate notice, site preview, Codecov (90% patch coverage)
20:55	ralphbean approved with "LGTM."
21:04	PR merged

Total time: ~65 minutes (issue to merge). Code agent time: 7 minutes. Human review wait: ~38 minutes.

What went well

• Issue quality drove agent success. The issue included root cause analysis, a code-level suggested fix, proposed test cases, and production evidence. The code agent followed the suggestion closely and produced a clean implementation on the first try.
• Good test coverage. 5 table-driven test cases covering the happy path and all 3 fallback paths (missing file, missing frontmatter, missing name field).
• Zero rework. No fix cycles, no review change requests. Single commit, single review, merged.

Observation: no visible review agent activity

No review agent comment or review verdict was posted on this PR. The dispatch repo shows review.yml runs around the right timeframe (runs 28128704829 at 20:49 and 28128907845 at 20:53, both successful), but no corresponding comment appeared on the PR. It is unclear whether the review agent ran for this PR and chose not to post (by design for clean PRs), or whether it ran for a different PR. This is not necessarily a problem — the human reviewer approved it quickly — but it means there is no auditable trace of automated review for this change.

Existing issues #1373 (post review on merged PRs) and #2619 (review status comment timing) are tangentially related but do not cover the specific gap of ensuring a visible review trace exists on every agent-authored PR.

Proposals

One proposal below addresses the review traceability gap. No other improvements are warranted — this workflow performed well across all optimization dimensions.

Proposals filed

• Review agent should always leave a visible trace on agent-authored PRs (in fullsend-ai/fullsend)