Ezra144israel
UserSix skills tell coding agents how to work. Two guards enforce the parts that should not depend on judgment.
Categories
Indexed Skills (16)
test-verification
Requires behavioral, failure-path, and durable-seam evidence for tests and review. Use when writing tests, reviewing test coverage, assessing behavioral test quality, or accepting high-risk behavior on test evidence.
governed-operator
Full governance constitution for multi-agent or solo governed workflows. Load for any work that touches shared state, requires independent review, spans multiple seats, or involves planning, architecture, code changes, reviews, dispatches, status reports, or canonical records. Not needed for quick factual questions or trivial one-off answers.
reasoning-doctrine
The working method — how to think, stage, self-check, and hold alignment across long tasks without needing reminders. Use this skill on EVERY nontrivial task: analysis, planning, implementation, debugging, document work, reviews, or anything spanning more than a few tool calls or more than one response. Especially mandatory on long-running or multi-stage work, where drift is the primary failure mode. If a task has stages, this skill governs how you move through them.
write-maintainable-code
Evaluate minimum-sufficient implementation routes and enforce the selected route for a fixed, authorized outcome. Use after outcome, scope, authority, and acceptance evidence are fixed. Do not choose outcomes, set acceptance, adjudicate governance, grade finished work, design tests alone, conduct security audits, or perform unrelated cleanup.
portable-adaptive-planning
Use for planning, roadmaps, design, architecture, sequencing, or before consequential work when no current FINAL plan plus GO covers it. Fire before changes to data or schemas, credentials or auth, production or releases, repository or source-home boundaries, irreversible work, or materially costly mistakes. Re-enter when that scope appears mid-task or when restoring prior plan state. Skip factual answers, trivial reversible one-step work, and authorized Builder or Reviewer execution under a current FINAL blueprint plus GO unless a reopen condition fires.
ship-it-or-fix-it
Oracle-frozen Builder and independent-Judge convergence cycle. Load ONLY when the operator explicitly sets Governance Dial G2 for the task, or explicitly names this skill or an active work unit already running it. Never auto-activate on task class, such as security, auth, or payments. If a task seems to warrant G2 and the operator has not said so, ask first. Not for ordinary governed implementation, analysis, review-only, or documentation work.
cil-grounding
Session-resident CIL grounding at session start for Ezra's governed environment, portable to every surface. Ported from the operator-locked cil-auto-grounding--convergence-v3 (Claude Code local, blinded-scored PASS 2026-08-15). Read once at session start alongside governed-operator and reasoning-doctrine; reread only after compaction/summarization, a source-identity change, or an explicit operator refresh.
cil-mode-switch
Portable CIL Mode Switch source candidate. Enter, hold, and automatically restore a CIL-conditioned operating posture across a session, without storing or restoring authority. Pre-wiring source only — it installs nothing and grants nothing.
grilling
Grill the user relentlessly about a plan, decision, or idea, working a design tree in rounds. Use when the user wants to stress-test their thinking, before a convergence packet, or on any 'grill' trigger phrase.
better-coding
Evaluate minimum-sufficient implementation routes and enforce the selected route for a fixed, authorized outcome. Use after outcome, scope, authority, and acceptance evidence are fixed. Do not choose outcomes, set acceptance, adjudicate governance, grade finished work, design tests alone, conduct security audits, or perform unrelated cleanup.
plans
Use for planning, roadmaps, design, architecture, sequencing, or before consequential work when no current FINAL plan plus GO covers it. Fire before changes to data or schemas, credentials or auth, production or releases, repository or source-home boundaries, irreversible work, or materially costly mistakes. Re-enter when that scope appears mid-task or when restoring prior plan state. Skip factual answers, trivial reversible one-step work, and authorized Builder or Reviewer execution under a current FINAL blueprint plus GO unless a reopen condition fires.
ship-or-fix
Oracle-frozen Builder and independent-Judge convergence cycle. Load ONLY when the operator explicitly sets Governance Dial G2 for the task, or explicitly names this skill or an active work unit already running it. Never auto-activate on task class, such as security, auth, or payments. If a task seems to warrant G2 and the operator has not said so, ask first. Not for ordinary governed implementation, analysis, review-only, or documentation work.
teamwork
Full governance constitution for work that touches shared state, requires independent review, spans seats, or involves planning, architecture, code, reviews, dispatches, status reports, or canonical records. Not needed for quick factual questions or trivial one-off answers.
test-the-test
Requires behavioral, failure-path, and durable-seam evidence for tests and review. Use when writing tests, reviewing test coverage, assessing behavioral test quality, or accepting high-risk behavior on test evidence.
thinking
The working method for every nontrivial task: frame the objective, ground load-bearing facts, converge on a route, execute without drift, and verify before delivery. Use for analysis, planning, implementation, debugging, document work, reviews, and multi-stage tasks.
run-review-repair-loop
Conduct a bounded, repository-grounded code self-review and repair loop with objective 1-5 scoring, exact scope preservation, targeted fixes, and focused and full verification. In repositories with owned doctrine, dynamically apply current engineering, security, testing, maintainability, build-gate, and return-contract standards through progressive activation. Use when the user asks you to review your own recent code, grade the work, repeatedly fix actionable findings until it earns an evidence-backed 5/5 or reaches an iteration limit, or prepare a change for independent review without publishing or deploying it.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.