Agentic Coding Kit v6.3

A native, development-focused workflow kit for long-horizon coding work in Codex, Claude Code, OpenCode, Muse Code, and GitHub Copilot CLI. The active harness session is the orchestrator. Skills define reusable loops; small, bounded agents keep exploration, implementation, review, testing, and specialist judgment out of the main context.

The kit is intentionally model-neutral. It does not force one provider's model names into another harness.

How it works

• One host-native primary orchestrator owns the outcome; specialists never become nested orchestrators or dispatch successors.
• The primary chooses a direct INLINE route where a skill permits it, or owns a delegated LOOP; specialists never choose the route or inherit ownership.
• Implementation loops share only a compact GOAL, numbered ACCEPTANCE, and repository-grounded PLAN, then use one Coder, fresh primary proof, one combined Reviewer, and at most two unsuccessful repairs for the same failure.
• Agent returns use only Result, Evidence, and optional Next.
• Extra test hardening, browser QA, UI critique, and security review are conditional evidence gates rather than ceremonial stages.
• Reviewer findings name file and line, what is wrong, the evidence, and the minimum fix; repair receives the finding verbatim plus owned files and checks to re-run.

Start here

Requirements

• Windows or macOS.
• Node.js 20 or newer.
• At least one supported harness installed separately.
• Git for project-scope installation and source inspection.

Build the management CLI once from the repository root (which itself has no package.json):

npm ci --prefix cli
npm run bundle --prefix cli

Install

Warning: a user-scope install replaces the selected harness's complete global instructions, agents, skills, commands, and primary configuration. Back up custom agents, skills, model settings, MCP servers, hooks, or permissions first. Use --yes only for deliberate non-interactive installs.

powershell -NoProfile -ExecutionPolicy Bypass -File .\scripts\install-all.ps1 `
  --scope user --profile core --security preserve --memory preserve

bash scripts/install-all.sh \
  --scope user --profile core --security preserve --memory preserve

For one harness use install-codex, install-claude, install-opencode, install-muse, or install-copilot with the platform's .ps1/.sh suffix. core installs every coding loop and all eight core agents; full adds browser execution and visual UI critique. An all-host install dry-runs every host before changing any of them.

Skills

Nineteen lazy-loaded skills. The orchestrator loads only the one or two each task needs; bodies are never preloaded.

Skill	Purpose
build	Implement features, fixes, refactors, migrations, UI, API, data, and configuration.
design	Produce a feature, product, prototype, or UI design before implementation.
grill	Run an explicitly requested one-question-at-a-time decision interview.
review	Independent review of a diff, branch, contract, design, subsystem, or test delta.
pr-ready	Repair and package a diff for efficient human PR review.
threat-model	Read-only trust-boundary, attack-path, control, and mitigation analysis.
wiki	Initialize, reinitialize, or audit curated repository engineering knowledge.
async	Cancellation, retries, concurrent state, cleanup, and idempotency.
backend	Backend boundaries, persistence, failure behavior, and API contracts.
components	Find and compose existing UI primitives against their real APIs.
debug	Discriminate an unexplained failure or a repair that is not making progress.
deslop	Simplify an authorized diff with evidence while preserving safeguards.
frontend	Intentional UI: hierarchy, tokens, icons, states, and evidence.
migrate	Persisted-data, schema, or mixed-version rollout changes.
perf	A specified latency, throughput, or resource bottleneck.
python	Python behavior, typing, resources, and tests.
security	A named trust boundary, authorization rule, or cross-system contract.
test	Changed behavior through public seams.
typescript	TypeScript contracts, narrowing, and async behavior.

Retired names (analyze, architecture, experiment) have no installed skill. No agent is spawned for ceremony: a Repo Scout, Test Engineer, Architect, Diagnostician, Sage, Security Reviewer, Browser QA, or UI Critic joins only for a concrete discovery need, risk, or proof gap.

Agents

Agent	Responsibility	Writes
architect	Architecture and change-boundary decisions.	No
repo-scout	Bounded repository discovery and evidence mapping.	No
coder	Production implementation and durable behavior evidence.	Production and tests
reviewer	Independent code, design, and test-delta judgment.	No
test-engineer	Independent high-value test hardening.	Tests and fixtures only
diagnostician	Repeated or ambiguous failures.	No
sage	Principal-engineering challenge for difficult decisions.	No
security-reviewer	Material trust-boundary changes.	No

full adds browser-qa and ui-critic. Every agent returns to the primary; completed specialists are never reactivated and transcripts are not forwarded.

Harness locations and invocation

Harness	User agents	User skills	Typical invocation
Codex	~/.codex/agents/*.toml	~/.agents/skills/<name>/SKILL.md	$build ..., $review ...
Claude Code	~/.claude/agents/*.md	~/.claude/skills/<name>/SKILL.md	/build ..., /review ...
OpenCode	~/.config/opencode/agents/*.md	~/.config/opencode/skills/<name>/SKILL.md	/build ... thin command or native skill
Muse Code	~/.config/muse/agents/*.md	~/.config/muse/skills/<name>/SKILL.md	Natural language; skills discovered natively
Copilot CLI	~/.copilot/agents/*.agent.md	~/.copilot/skills/<name>/SKILL.md	Natural language; /skills and /agent to inspect

Verify the installed state with node cli/dist/kit.cjs doctor --host all --scope user. Uninstall removes only manifest-owned files, keys, and managed blocks: node cli/dist/kit.cjs uninstall --host all --scope user.

Repository wiki

wiki init / reinit / audit build a source-backed map of how a repository actually works (entry points, flows, boundaries, conventions, tests). Normal work never edits .wiki; only explicit wiki init or wiki reinit may.

Verification

npm run typecheck --prefix cli
npm test --prefix cli
npm run validate --prefix cli
npm run check:drift --prefix cli

Source layout

Path	Purpose
core/	Canonical manifest, schemas, orchestrator, agents, and skills.
packs/	Canonical optional specialist sources.
adapters/	Generated host-native artifacts; edit canonical sources instead.
cli/	Renderer, installer, doctor, migration, wiki, and tests.
scripts/	Thin Windows and macOS launchers.

Security

Do not run the installer elevated against a directory writable by another user. permissive removes ordinary approval and sandbox protections; use it only with repositories, credentials, machines, and networks you trust.

License

MIT. See LICENSE.