feat(skills): SOTA skill infrastructure — frontmatter, required_tools, eval datasets, 3 new dist targets

[Delqhi]

Summary

SOTA best-practices upgrade for all 36 bundled SIN-Code skills.

Changes

Skill frontmatter standardization (all 36 skills)

• Filled compatibility: (sin-code, opencode, claude-code, codex) on all skills
• Filled metadata: (author: SIN-Code, version: 3.20.0) on all native skills
• External skills: kept existing metadata, added compatibility

required_tools frontmatter field (8 key skills)

• New YAML field declares which SIN tools a skill requires (e.g. sin_edit, sin_test, sin_sckg, sin_oracle, sin_poc)
• Parsed at runtime by cmd/sin-code/internal/skillmgr/required_tools.go
• Merged additively (deduped, sorted) into agentloop.Loop.CoverageRequiredTools via loopbuilder.Config.ActiveSkills
• ToolCoverageEnforcer (issue Runtime tool coverage enforcer: must_use_tools / forbidden_tools in live loops #248) rejects task completion if required tools not invoked
• 17 race-clean tests

3 new skilldist targets (8 → 11, non-breaking)

• aider → .aider/conventions/<skill>.md (rule format)
• continue → .continue/rules/<skill>.md (rule format)
• zed → .zed/rules/<skill>.md (rule format)
• Also added to profile distribution (internal/profile/target.go)

3 skill eval datasets (four-arm comparator)

• evals/skill-code.json — 3 cases (build / refactor / plan)
• evals/skill-debug.json — 2 cases (race / nil-pointer RCA)
• evals/skill-github.json — 2 cases (actions / readme)
• Wired into eval-n8n.yml CI (M1: n8n delegation only)

Bug fixes

• skill-github-governance: lifecycle: external moved from metadata: to top-level (validator rejected it in strict mode)
• skill-github-readme: same lifecycle fix + 5 empty dirs filled (context/frameworks/tasks/templates)
• skill-code-create: v3.17.0 → v3.20.0, skill count 34 → 36, frontmatter filled, external duplicate removed

Docs

• AGENTS.md §10: skill dist table 8 → 11 targets, profile dist table 8 → 11, all counts updated
• CHANGELOG.md: Unreleased section added
• evals/README.md: 3 new datasets in catalog

Test results

ok  github.com/OpenSIN-Code/SIN-Code/cmd/sin-code/internal/skillmgr     1.358s
ok  github.com/OpenSIN-Code/SIN-Code/cmd/sin-code/internal/skilldist    1.437s
ok  github.com/OpenSIN-Code/SIN-Code/cmd/sin-code/internal/profile      1.700s
ok  github.com/OpenSIN-Code/SIN-Code/cmd/sin-code/internal/loopbuilder  12.866s
ok  github.com/OpenSIN-Code/SIN-Code/cmd/sin-code/internal/agentloop    14.192s

Validator: 36/36 skills pass --strict, 0 failed.

Mandates

• M1 (n8n CI): ✅ eval-n8n.yml uses webhook delegation only
• M2 (single binary): ✅ no new runtime deps (yaml.v3 already a direct dep)
• M5 (module path): ✅ all imports use github.com/OpenSIN-Code/SIN-Code
• M7 (race-free): ✅ all new tests pass -race -count=1

Files changed

60 files changed, 1266 insertions(+), 93 deletions(-)

Note: Pre-existing TUI WIP files (cmd/sin-code/tui/) are excluded from this PR.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
sin-code	Ready	Preview, Comment, Open in v0	Jun 17, 2026 9:52pm

[vercel]

Deployment failed with the following error:

Resource is limited - try again in 24 hours (more than 100, code: "api-deployments-free-per-day").

Learn More: https://vercel.com/manheiler-8784s-projects?upgradeToPro=build-rate-limit

[github-actions]

🏆 CEO Audit — A+ (100.0/100)

Metric	Value
Grade	A+
Score	100.0/100
Critical findings	0
High findings	0
Profile	QUICK
Min grade gate	B

📥 Download full report (Markdown)
📊 Download SARIF (for Code Scanning)

Run ~/.config/opencode/skills/ceo-audit/scripts/audit.sh . --profile=QUICK locally to reproduce.

[github-actions]

🏆 CEO Audit — A+ (100.0/100)

Metric	Value
Grade	A+
Score	100.0/100
Critical findings	0
High findings	0
Medium findings	0
Profile	QUICK
Min grade gate	B

📥 Download full report (Markdown)

Run ID: 27722105802 · Commit: ${github.sha}

Run ~/.config/opencode/skills/ceo-audit/scripts/audit.sh . --profile=QUICK locally to reproduce.