Skip to content
CAI
Software that uses CAICheck a score

addyosmani/agent-skills

66.0

Adequate · 13 November 2026

1.9k

lines of production code

JavaScript

primary language

7

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

Added executable script to initialize the ideas directory

The idea-refine skill now includes an executable shell script (idea-refine.sh) that ensures the docs/ideas directory exists. When run, the script creates the directory if it is missing and outputs a JSON status object indicating whether the directory was just created or already existed, allowing users to reliably initialize the skill's workspace.

skills/idea-refine/scripts · high confidence

Expanded evaluation fixtures for behavioral and skill-based testing

The \evals/fixtures\ directory has been significantly expanded with new scenario files and code samples to support a broader range of evaluation capabilities. This update introduces fixtures for browser testing with devtools (including a signup reproduction app), CI/CD automation (slug generation), code review (user search diff), code simplification (config parser), context engineering (session audit), debugging (pagination and time-pressure scenarios), deprecation/migration (API inventory), documentation (ADR context), doubt-driven development (migration plan), frontend UI engineering (React button component), git workflow (working tree patch), incremental implementation (report filtering and export drafts), observability (payment retry), performance optimization (product rendering), planning (notifications spec), security (webhook preview), shipping/launch (checkout status), source-driven development (Express session task), spec-driven development (billing and portal briefs), test-driven development (split payment and ledger utilities), and agent skill usage (login regression). These additions provide concrete, executable, or descriptive contexts for evaluating agent behaviors across diverse software engineering tasks. Also in this area: 1 commit (0 fixes) modifying evals; 7 commits (3 fixes) modifying evals; Evals now require source inspection and flag unverified commands.

evals/fixtures · medium confidence · unverified

New CI validation scripts for skills, commands, and artifact paths

Added a suite of Node.js scripts to enforce consistency and correctness across the agent-skills repository. \validate-commands.js\ ensures command parity (existence and description sync) across Claude Code, Gemini CLI, and Antigravity CLI directories, and validates that Claude command frontmatter is valid YAML. \validate-artifact-paths.js\ prevents silent drift in the spec/plan/build pipeline by checking that all producer and consumer files reference only a canonical allowlist of artifact paths. \validate-reference-links.js\ checks that skills correctly link to shared checklists in the \references/\ directory, accounting for relative path resolution from both \SKILL.md\ and nested reference files. \run-evals.js\ and its test \run-evals-test.js\ implement a tiered evaluation framework, including deterministic Tier 2 routing/coverage checks and opt-in Tier 3 behavioral evaluations with strict grading validation (e.g., rejecting blank evidence). \floor-guard-reference-test.js\ provides contract tests for the floor-guard constraint logic.

scripts · high confidence

New session-start hook and opt-in citation cache for source-driven development

The hooks directory now includes a session-start hook that emits the standard SessionStart envelope with agent-skills context, replacing the previous auto-injection of the meta-skill. Additionally, an opt-in citation cache (sdd-cache) is introduced for the source-driven-development skill: it intercepts WebFetch calls to cache and revalidate documentation content via HTTP 304 responses, reducing redundant fetches while preserving freshness guarantees. A new simplify-ignore hook is also added to protect specific code blocks from being simplified by the model, using placeholders that are expanded back to original content on session stop.

hooks · high confidence

New skill validation library and test suite

Added \scripts/lib/skill-lint.js\ as the central library for validating SKILL.md files, enforcing rules such as required YAML frontmatter, name/description constraints, required section presence, and cross-skill reference checks. Included \scripts/lib/skill-lint-test.js\ with comprehensive tests for these validation rules, including edge cases like fenced code block stripping and prototype chain security fixes.

scripts/lib · high confidence

Behavioural changes

The .opencode/skills path has been converted from a regular directory to a symbolic link pointing to ../skills/. This change ensures that the opencode skills are located and loaded correctly by resolving the symlink to the actual skills directory.

.opencode · high confidence

Fixes

1 commit (1 fix) fixing skills/constraint-driven-development

A fix in skills/constraint-driven-development — 1 commit (1 fix), 2 files: “fix(constraints): enforce security and accessibility quality gates”.

skills/constraint-driven-development/scripts · medium confidence · unverified

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

How this codebase got here

Score

  • CAI 63 → 66 (+2.8)
  • Rubric changed (rubric-2026.10.5 → rubric-2026.11.3) — scores are not directly comparable.

Lenses

  • Code Health 52 → 62 (+9.5)
  • Architecture 95 → 95 (+0.2)
  • Maturity 85 → 87 (+1.7)
  • Readiness 53 → 53 (+0.0)
  • Security 100 → 100 (+0.0)
  • Performance 100 → 100 (+0.0)

Resolved (21)

  • Documentation: no contributor guidance (README.md)
  • Documentation: no project overview (README.md)
  • FunctionTooLong: run-evals.runDeterministic (scripts/run-evals.js)
  • Hotspot: scripts/validate-reference-links.js (scripts/validate-reference-links.js)
  • Scanner failed to run — not a clean result
  • Scanner failed to run — not a clean result
  • run-evals.parseGrading (cognitive 28) (scripts/run-evals.js)
  • run-evals.parseGrading (cyclomatic 27) (scripts/run-evals.js)
  • run-evals.runBehavioral (cognitive 30) (scripts/run-evals.js)
  • run-evals.runBehavioral (cyclomatic 21) (scripts/run-evals.js)
  • run-evals.runDeterministic (cognitive 111) (scripts/run-evals.js)
  • run-evals.runDeterministic (cyclomatic 69) (scripts/run-evals.js)
  • skill-lint.frontmatterYamlErrors (cognitive 16) (scripts/lib/skill-lint.js)
  • skill-lint.frontmatterYamlErrors (cyclomatic 17) (scripts/lib/skill-lint.js)
  • skill-lint.lintSkillContent (cognitive 38) (scripts/lib/skill-lint.js)
  • skill-lint.lintSkillContent (cyclomatic 27) (scripts/lib/skill-lint.js)
  • validate-commands.main (cognitive 37) (scripts/validate-commands.js)
  • validate-commands.main (cyclomatic 23) (scripts/validate-commands.js)
  • validate-reference-links.main (cognitive 30) (scripts/validate-reference-links.js)
  • validate-skills.main (cognitive 32) (scripts/validate-skills.js)
  • …and 1 more

New (3)

  • Duplicated block (7 lines × 3) (evals/fixtures/review-evidence/after.py)
  • Repeated repair: scripts/validate-reference-links.js (scripts/validate-reference-links.js)
  • Skipped (documented): documented staged-secret scans name the file, never the secret (scripts/staged-secret-scan-test.js)

Changes since last survey

  • 73 commits — 64 feature/other, 9 fixes

By area

  • (repo) — 49 commits
  • (root) — 3 commits
  • evals/README.md — 3 commits
  • evals/fixtures — 3 commits
  • evals/plugin — 3 commits
  • scripts/run-evals-test.js — 2 commits
  • .github/workflows — 1 commit
  • docs/antigravity-setup.md — 1 commit
  • docs/copilot-cli-setup.md — 1 commit
  • evals/cases — 1 commit
  • hooks/SDD-CACHE.md — 1 commit
  • hooks/simplify-ignore-test.sh — 1 commit
  • references/accessibility-checklist.md — 1 commit
  • skills/constraint-driven-development — 1 commit
  • skills/doubt-driven-development — 1 commit
  • skills/security-and-hardening — 1 commit

Notable commits

  • fix: fix(constraints): enforce security and accessibility quality gates
  • fix: fix(evals): drop command names from the skill-fired graders after #714; record the re-run
  • fix: fix(evals): drop grader files that belong to a separate change
  • fix: fix(evals): reject incomplete executor streams before grading
  • fix: fix(git-workflow-and-versioning): say the skill applies when the diff is pasted inline (#620)
  • fix: fix(hooks): preserve re-indented placeholders and harden Stop cleanup in simplify-ignore.sh
  • fix: fix(security-and-hardening): keep staged-secret scan from echoing credentials
  • fix: fix(security-and-hardening): name session cookies in the description
  • fix: fix(ui): correct contrast thresholds and settled resize checks
  • change: Merge #463: conform root plugin.json to Agent Plugins 1.0.0
  • change: Merge #668: route clipped prompt forms to their long-form skill
  • change: Merge #683: type the error boundary's children props
  • change: Merge #688: filter list items by text in the Playwright example
  • change: Merge #690: pass --fix through npm run to the linter
  • change: Merge #691: pass the rollback version through env, not straight into run
  • change: Merge #692: declare the empty-state action prop in frontend-ui-engineering
  • change: Merge #694: enforce security and accessibility quality gates
  • change: Merge #696: say code-simplification applies to pasted code
  • change: Merge #698: make the deprecation adapter example type-check
  • change: Merge #699: alias the attribution-build onINP import
  • …and 53 more

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

Survey your own repository

addyosmani/agent-skills was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.

About this page

  • The score is its most recent published measurement, taken on 13 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
  • Measured at commit 301b02b12bff98e4688e6984af6556802322bbde — the exact code this score is about.
  • Scored under rubric-2026.11.3 — the same rubric and the same method as every other entry in this index.
  • Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.61+preprod-f3449e6eb431.