0merUfuk
UserAutonomous agent ecosystem — CLI tools that provision, scaffold, and maintain Claude Code agent ecosystems
Categories
Indexed Skills (17)
fix
Applies safe mechanical fixes from the last /audit report. Corrects context file drift, closes resolved issues, updates stale paths and versions. Never touches Go source code, templates, or git history.
audit
Audits the-matrix context files against ground truth (git, code, filesystem). Verifies versions, code claims, docs accuracy, and monorepo structure. Use when context files may have drifted from reality, after implementations, or before releases.
commit
Smart commit for the-matrix monorepo. Analyzes changes, groups by logical intent, creates conventional commits with proper types and scopes. Handles both standalone multi-commit grouping and post-session single-commit flows.
dep-audit
Audit project dependencies for known vulnerabilities using govulncheck, go mod verify, trivy, and npm audit. Merges and deduplicates findings across scanners for a unified vulnerability report.
doublecheck
Adversarial verification of implementation correctness. Runs 4 subagents that attack from different angles: gap finding, assumption attacking, ground truth verification, and devil's advocacy. Use after completing work, before releases, or to attack a plan before implementation.
issue
Create GitHub Issues with enforced conventions for the-matrix monorepo. Applies consistent title format, label taxonomy, and body template. Use for filing bugs, feature requests, tech debt, test gaps, and deferred findings.
owasp-review
On-demand OWASP Top 10:2025 and ASI01-ASI10 security audit for any feature or PR. Invokes the security-reviewer agent against the current branch diff.
provision
Creates or updates a .claude/ artifact (agent, skill, or rule) following the-matrix conventions. Generates files with correct YAML frontmatter, documentation structure, and convention compliance. Use when the ecosystem needs a new agent role, a new skill workflow, or a new constraint rule.
release
Full release workflow for the-matrix monorepo. Analyzes changes, determines version bump, creates a version-bump PR, tags after merge via `make release`, monitors the CI pipeline (GoReleaser + Homebrew tap), and bumps to the next dev version. Always a unified release of all 4 tools at the same version.
secret-scan
Scan the codebase and git history for leaked secrets using gitleaks. Detects API keys, tokens, passwords, and platform-specific patterns (Stripe sk_live_/sk_test_, AWS AKIA, RabbitMQ amqp://).
security-scan
Static application security testing using gosec and semgrep. Scans Go source code for common vulnerability patterns and ranks findings by severity using the impact x exploitability x blast_radius formula.
self-improve
L1 (propose-to-file) self-improvement cycle entry point. Reads the latest /self-state snapshot, ranks dispatch_signals inline, drafts structured improvement proposals to files, and appends a cycle entry to the durable ledger. Proposes only — never spawns any agent, creates branches, PRs, or merges; the founder executes selected proposals manually via the manager.
self-state
Produces a structured state.v1 JSON snapshot of the-matrix ecosystem state for autonomous self-improvement decisions. Composes existing /audit, trinity health, /dep-audit, /security-scan via 4 parallel subagents over a deterministic Go collector baseline.
session-learn
Captures development session findings and proposes targeted improvements to the .claude/ ecosystem. Reads session artifacts, briefs the architect agent, and applies human-approved additions to rules, agent prompts, and checklists.
strategy-monthly
Monthly strategy review. Orchestrates product-lead, tech-lead, and growth-lead in parallel, synthesizes findings, and produces an updated strategy assessment. Refreshes relevant strategy-output/ files if strategic direction has changed. Use at month boundaries or when major strategic questions arise.
strategy-weekly
Weekly strategy sync. Spawns growth-lead for channel/content review, cross-references with sprint plan, and produces a weekly tactical brief. Use every Monday or on demand when you need a quick project health pulse.
taste-calibration
Monthly founder ritual that re-anchors the Goodhart drift defense: reviews autonomous loop activity since last calibration, confirms or updates the canonical taste SHA in .claude/baselines/taste-anchor.md, appends a calibration-log row, and optionally refreshes golden exemplars. Run once per month or after any significant loop activity.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.