Agent-Skills

Apply approved fixes for unresolved CodeRabbit review comments, Codex P1-P3 findings, PR feedback, and code review issues with validation evidence. Use when asked to address review comments, fix review findings, clear unresolved comments, or autofix PR feedback.

Generate and compare grounded product or engineering directions with tradeoffs. Use when users want possibilities, critique, or direction-setting before a spec.

Improve existing Harness Engineering implementations or workflows with evidence-backed changes. Use when users ask for targeted enhancement of shipped or drafted work.

Debug and repair validated Harness Engineering defects with bounded scope, reproduction evidence, root-cause notes, regression protection, and validation proof. Use when a bug is already evidenced and the fix should not expand into broad improvement work.

Create evidence-backed HE reframe migration programs. Use when structural drift, routing ambiguity, or source-prompt gaps need phased rollback-safe execution.

Review services, APIs, and multi-component systems for reliability risks including failure modes, cascading failures, resilience gaps, and SLO readiness. Use when the work involves new services, significant service changes, multiple external dependencies, or high blast-radius failure scenarios.

Review and prune stale branches safely. Use when branch cleanup needs evidence, protected-branch caution, PR awareness, and non-destructive recommendations.

Compress HE cognition artifacts into evidence-backed strategy. Use when intent, review, triage, ADR, core, or source-prompt comparison evidence needs durable direction.

Deepen an existing system or UI spec so boundaries, lifecycle rules, failure handling, and validation are strong enough for planning. Use when the user wants Harness Engineering spec hardening or a requirements review pass before planning.

Analyze broad, mixed, or unclear Plugin Factory follow-up requests and select the correct plugin lane. Use when plugin intent lacks a clear lane owner.

Implement approved Harness Engineering work. Use when a plan, todo list, or tiny spec needs traceable delivery and validation.

Creates release-ready skill packages from proven workflows by drafting SKILL.md frontmatter, trigger-rich instructions, references/contract.yaml, eval scenarios, task profile, and validation commands. Use when the user asks to create a skill, skillify a workflow, package a process, write SKILL.md, make a reusable skill template, or prepare a skill for release checks.

Route ambiguous Harness Engineering requests to one lifecycle stage when users ask where to start, resume, plan, implement, review, debug, schedule a heartbeat, or resolve domain terminology.

Review, create, and validate Bash scripts when shell work needs strict mode, quoting safety, portability, or interpreter-compatible behavior.

Create, validate, install, fold, or troubleshoot Codex subagent role TOML, agents-table config, discoverability wiring, and duplicate-role merges. Use when a user asks for a Codex agent role, reviewer agent, role config, TOML role file, subagent setup, or overlapping agents to merge.

Review PRs, branches, diffs, and workflow artifacts for package-level go/no-go readiness with severity-ranked synthesis. Use when users need readiness synthesis rather than detailed technical-risk critique.

Executes bounded implementation work from approved specs, plans, issues, or small fixes by editing files, running validation, preserving unrelated work, and recording rollback and handoff evidence. Use when asked to implement, execute a plan, apply changes, or build a scoped feature.

Review diffs, PRs, specs, plans, or review-feedback items and return severity-ranked engineering findings with exact locations. Use when technical risks or feedback correctness must be verified before implementation.

Scan Codex session history for skill failures, usage patterns, and coverage gaps. Use when the user wants daily skill-health monitoring or evidence-backed recommendations about installing, improving, merging, or pruning skills.

Capture a completed Codex workflow as a reusable SKILL.md package by analyzing session context plus optional session-collector evidence, interviewing the user with structured prompts, and writing a validated skill artifact. Use when the user asks to skillify or operationalize a repeatable process.

Create or refresh evidence-bound Harness Engineering learning artifacts from verified solved problems. Use when a fix worked, a repeated failure should become durable knowledge, or .harness/solutions and Project Brain need maintenance.

Create bounded, evidence-backed Harness Engineering specs from approved intent. Use when a selected issue, milestone, reframe phase, or execution slice needs acceptance criteria, traceability, risk gates, and validation boundaries before planning or implementation.

Route plugin-factory requests to the right lane. Use when plugin creation, building, installation, review, or routing is broad, mixed, or under-specified.

Build behavior-safe code changes with TDD and RED/GREEN evidence. Use when he-plan or he-work requires TDD for a concrete behavior target.

Improve existing Harness Engineering skills, references, contracts, and evals from concrete evidence such as failed evals, repeated review findings, usage traces, or documented regressions. Use when a bounded hardening pass is required; do not use for speculative redesign.