A light-weight and straightforward workflow for spec-driven development with Claude Code and Codex. Written in Rust 🦀
This repository also configures its own agents. See what speq-skill tells them →
Skill Claude Code
Audit a speq project's health — spec-library structure, feature/decision-log/plan validation, ADR noise, mission-to-spec sync, unrecorded plans, and gitignore hygiene — then guide fixes. Use when the user asks to audit, health-check, doctor, lint, or sanity-check the specs or repo, or after cloning or inheriting a…
Skill Claude Code
Query specs via the speq CLI: feature listing and reading, structure validation, and the absence check. Every speq orchestrator and sub-agent invokes this first for spec discovery or validation.
Skill Claude Code
TDD cycle and code quality guardrails — failing-test-first, evidence, dependency, error-handling, and test-quality rules. Used by speq implementer and reviewer agents during a speq implement run. Not for general coding.
Skill Claude Code
Code review tag taxonomy and findings output format — guardrail violations, dead code, test quality, bad comments, optimizations, YAGNI/over-engineering, error handling, and design depth. Used by the speq code-reviewer agent during a speq implement run. Not for general code review.
Skill Claude Code
Semantic code navigation and editing via Serena MCP — symbol-level tools required over grep/find/manual edits. Used by speq agents during a speq plan, implement, or record run. Not for general coding.
Skill Claude Code
Complexity-management design principles — deep modules, information hiding and leakage, module boundaries, dependency direction, strategic vs. tactical tradeoffs. Used by speq planner, reviewer, and implementer agents during a speq run. Not for general design work.
Skill Claude Code
External documentation and research via Context7 and WebSearch — library APIs and design patterns. Use when current external sources are needed.
Skill Claude Code
Read-only git guardrails. No history writes, verify changes before completing. Used by speq agents during a speq run. Not for general git work. The headless orchestrators speq-plan-pr and speq-implement-pr use /speq-git-operations instead.
Skill Claude Code
The git/gh operation-to-command mapping, safety rules, and return formats — invoked directly by the headless orchestrators speq-plan-pr and speq-implement-pr, the only components permitted to write git history or touch a remote.
Skill Claude Code
Headless follow-up to /speq-plan-pr. Continues a plan on its feat/plan-name branch and runs it end-to-end: implements via /speq-implement, bumps the version, commits and pushes, runs the real test suites, records only if green, then opens/updates the PR and marks it ready. Resumes from the PR Lifecycle checkpoint in…
Skill Claude Code
Orchestrate implementation of a reviewed plan: task breakdown, TDD sub-agents, code review, and verification report. Use when the user asks to implement, build, or execute a plan under specs/plans/ — after /speq-plan, before /speq-record. Arg: .
Skill Claude Code
Create or update specs/mission.md and the first specs/architecture.md through a Socratic interview; detects brownfield vs greenfield. Use when the user asks to bootstrap or initialize a speq project, write or revise the project mission, or when /speq-audit reports mission drift and seeds this skill with its findings.
Skill Claude Code
Headless, non-interactive version of /speq-plan for CI or agent-driven runs. Plans a feature without a live interview, commits the result to a feat/plan-name branch, and opens a draft PR. If a decision genuinely needs a human, it persists the partial plan and open questions and asks in a PR comment instead of…
Skill Claude Code
Adversarial plan-review method and challenge taxonomy — premortem framing, multi-axis challenge taxonomy, severity discipline, and findings output format. Use when reviewing a plan.
Skill Claude Code
Plan a feature or change through a clarifying interview, producing spec deltas, plan.md, and decision-log.md via planner-agent with adversarial review. Use when the user asks to plan, spec, design, or scope a new feature, a change or removal of existing behavior, a refactor, or a fix — before any implementation.
Skill Claude Code
Plan-authoring workflow — spec delta authoring, test mapping, plan.md/decision-log.md generation, expert-task tagging, headless escalation, and revision mode. Use when authoring a plan.
Skill Claude Code
Merge implemented spec deltas into the permanent specs library and archive the plan. Use when implementation is verified — after /speq-implement produces verification-report.md — or when the user asks to record, merge, or archive a finished plan. Arg: .
Skill Claude Code
Delta-merge procedure — DELTA marker semantics, library-organization thresholds, and ADR-promotion field mapping. Use when recording a plan's deltas.
Skill Claude Code
Nine prose rules covering conclusion-first order, one idea per sentence, named actors, consistent names, plain explanations, no filler, neutral tone, current results only, and no em dashes. Used by speq orchestrators and agents for speq artifacts and PR text. Not for general writing.
Skill Claude Code
The rules for when a decision becomes an ADR. The default is no, only durable decisions qualify, search first, one decision per ADR, and a human accepts an ADR when they record the plan. Use only inside a speq plan, record, or audit workflow, when deciding whether a decision becomes an ADR. Not for general decision…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: