v0.4.15

Fixed

• commands/project/execution/complete-ticket.md: corrected YAML frontmatter that failed strict parsing in oh-my-pi and the GitHub renderer. Two issues: unquoted double quotes around "In Review" were interpreted as a quoted-string opener with trailing junk, and argument-hint: [ticket-key] (optional...) led with [ which opens a flow-style sequence and then failed on the prose tail. Wrapped both values in matching outer quotes (#193)

Added

• CommandFrontmatterChecker in scripts/validate-skills.py: strict-parses YAML frontmatter on every commands/**/*.md with PyYAML (no fallback to the lenient simple parser, which had been masking these bugs). Runs as part of the workflow validation pass.

Contributors

• @rlex — Reported broken YAML frontmatter in complete-ticket.md after the 0.4.14 release surfaced the parse error in oh-my-pi (#193)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.14

Fixed

• postgres-pro reference docs: corrected invalid column references in pg_stat_user_tables and pg_stat_user_indexes query examples (tablename → relname, indexname → indexrelname) so the snippets actually run. Hardened identifier assembly with format('%I.%I', schemaname, relname) for safety against quoted/case-sensitive identifiers. Clarified the VACUUM FREEZE comment to note it operates on the current database, not the entire cluster (#191)

Contributors

• @sobstel — Fix invalid pg_stat_* column references and VACUUM FREEZE scope comment in postgres-pro references (#191)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.13

Added

• Canonical Documentation backlink at the bottom of every SKILL.md linking to the docs site ([Documentation](https://jeffallan.github.io/claude-skills/skills/{domain}/{skill-name}/)), so skill aggregators (skills.sh, etc.) that consume raw source markdown render a real <a href> backlink to the docs site for SEO. The docs build pipeline (syncSkillPages in site/scripts/sync-content.mjs) strips this line at build time, so the docs site, public markdown mirrors, llms.txt, and llms-full.txt render cleanly without self-references. Documented the convention in CLAUDE.md.

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.12

Added

• Curated cross-references for 7 hub skills: 22 back-references added for agent routing quality (closes #68)
  • devops-engineer: terraform-engineer, kubernetes-specialist, sre-engineer, monitoring-expert, security-reviewer
  • fullstack-guardian: secure-code-guardian, architecture-designer, react-expert, typescript-pro
  • test-master: debugging-wizard, code-reviewer, feature-forge
  • database-optimizer: postgres-pro, graphql-architect
  • security-reviewer: api-designer, mcp-developer
  • architecture-designer: microservices-architect, code-reviewer
  • kubernetes-specialist: terraform-engineer, security-reviewer, chaos-engineer

Changed

• Bumped 7 hub skills from v1.1.0 → v1.1.1 (cross-reference updates)
• Added Prettier markdown formatting to pre-commit hooks, CI, and Makefile — covers skills/, docs/, and root .md files; excludes commands/ (prompt templates) (#158)
• Documented linting and formatting conventions in CONTRIBUTING.md with Make targets (#158)
• Added explicit name: frontmatter to 9 workflow commands across retrospectives/, discovery/, planning/, and execution/ directories — documents intent and protects against filename renames (filename-derived fallback was already working correctly) (#182, #183, #184, #185)

Fixed

• README: restored the broken link to the Workflow Commands Reference docs. The paragraph began with <!-- WORKFLOW_COUNT -->, which CommonMark parses as an HTML block and disables inline markdown, causing the embedded link to render as literal text (#180)
• commands/common-ground/COMMAND.md: added explicit name: common-ground so the slash command registers as /common-ground. Without it, Claude Code derived the name from the filename (COMMAND.md) and registered it as /COMMAND (#186)

Contributors

• @modir — Fix broken Workflow Commands docs link in README (#180)
• @xiaolai — Add missing name: frontmatter fields across workflow commands (#182, #183, #184, #185, #186)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.11

Added

• Context management reference (prompt-engineer/references/context-management.md) covering attention budgets, lost-in-the-middle mitigation, four-bucket context tiering, KV-cache optimization, observation masking, and degradation metrics (#168)

Changed

• Updated prompt-engineer skill (v1.1.0 → v1.2.0): added context management routing, expanded triggers, added rag-architect and debugging-wizard to related skills
• Total reference files: 365 → 366

Contributors

• @Genius-apple — Context management content adapted from the context-engineer skill submission (#168)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.10

Changed

• Improved skill quality across 65 skills via Tessl review optimization (#172)
  • Expanded descriptions with capability verbs (what it does + when to use it)
  • Removed redundant "Role Definition" and "When to Use This Skill" sections
  • Added structured workflow validation checkpoints with failure recovery loops
  • Added inline code examples demonstrating key patterns per technology
  • Tightened MUST DO / MUST NOT DO constraints with rationale
  • Removed "Knowledge Reference" keyword lists (absorbed into descriptions and triggers)
• Narrowed Description Trap from "trigger-only" to "no process steps" — capability verbs now permitted in descriptions per AgentSkills.io spec
• Updated description format to [Brief capability statement]. Use when [triggering conditions].
• Validator now checks "Use when" in description instead of startswith("Use when")
• Standardized README links from HTML to Markdown, improved Quick Start readability (#166)

Fixed

• Fixed stale counts on docs site landing page (65 → 66 skills, 357 → 365 references)
• Added Astro site files to update-docs.py release automation so counts stay in sync

Contributors

• @popey — Improve skill quality across 65 skills via Tessl review (#172)
• @hasan613 — Standardize README links and improve Quick Start section (#166)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.9

Added

• Issue linking documentation in atlassian-mcp/references/jira-queries.md with parameter semantics, code examples, and anti-patterns (#163)

Changed

• Updated php-pro Symfony console command example to use framework's built-in exception handling (#164)
• Added Framework Idiom Principle to CLAUDE.md reference file standards
• Added inline jira_create_issue_link parameter hints to approve-synthesis and create-implementation-plan commands (#163)

Fixed

• Fixed reversed Jira "Blocks" link parameters — inward_issue_key is the blocker, outward_issue_key is the blocked issue (#163)
• Fixed nested code block rendering in prompt-engineer skill's CoT example (#160)

Contributors

• @wiretail — Report and document reversed Jira issue link semantics (#163)
• @Big-Shark — Update console example in symfony-patterns (#164)
• @fiberproduct — Fix nested code block formatting in prompt-patterns.md (#160)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.8

Added

• 🎉 Milestone: Appeared on GitHub Weekly Trending repos (#8 overall)

Changed

• Added Agent Skills CLI as alternative installation method in QUICKSTART.md (#151)
• Upgraded GitHub Actions for Node 24 compatibility (#146)
  • actions/checkout v4 → v6
  • actions/setup-node v4 → v6
  • actions/setup-python v5 → v6
• Upgraded actions/upload-pages-artifact v3 → v4 (#147)

Fixed

• Removed incorrect Python/Pydantic V1 reference from php-pro skill MUST NOT DO section (#154)

Contributors

• @salmanmkc — Upgrade GitHub Actions for Node 24 compatibility (#146)
• @salmanmkc — Upgrade GitHub Actions to latest versions (#147)
• @Karanjot786 — Add Agent Skills CLI installation method (#151)
• @Aivanaso — Remove copy-paste error from php-pro skill (#154)

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.7

Added

• Bidirectional cross-reference validation in validate-skills.py (closes #100)
  • CrossRefChecker class: warns when skill A references B but B doesn't reference A
  • Orphan detection: warns about skills with no incoming or outgoing references
  • New --check crossrefs CLI option for isolated cross-reference validation
  • All issues are WARNING severity (advisory, won't fail CI)

Changed

• Upgraded Code of Conduct to Contributor Covenant v2.1 (closes #131)
• Refactored CrossRefChecker._build_graph() to reuse BaseChecker._extract_frontmatter()

---

Documentation: https://jeffallan.github.io/claude-skills/

v0.4.6

---

Documentation: https://jeffallan.github.io/claude-skills/