AlexGavrilov939/spec-dna
Catch the changes that break your Markdown specs (edited accepted decisions, broken links and ids, stale code citations, untested requirements) and give AI coding agents only the sections a file needs. Change lifecycle with requirement delt
stack: Rust
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
Copy this rig
# review before running: this installs third-party code
$ npx degit AlexGavrilov939/spec-dna/.claude ./rig-spec-dna # inspect, then merge into .claude/ MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ curl -fsSL --create-dirs -o .claude/agents/Explore.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/Explore.md $ curl -fsSL --create-dirs -o .claude/agents/architect.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/architect.md $ curl -fsSL --create-dirs -o .claude/agents/checker.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/checker.md $ curl -fsSL --create-dirs -o .claude/agents/context-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/context-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/core-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/core-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/diagnostics-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/diagnostics-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/docs-editor.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/docs-editor.md $ curl -fsSL --create-dirs -o .claude/agents/failure-analyst.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/failure-analyst.md $ curl -fsSL --create-dirs -o .claude/agents/fallback-advisor.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/fallback-advisor.md $ curl -fsSL --create-dirs -o .claude/agents/hygiene-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/hygiene-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/init-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/init-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/ownership.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/ownership.md $ curl -fsSL --create-dirs -o .claude/agents/release-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/release-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/test-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/test-engineer.md $ curl -fsSL --create-dirs -o .claude/agents/transcript-engineer.md https://raw.githubusercontent.com/AlexGavrilov939/spec-dna/master/.claude/agents/transcript-engineer.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 16.2k rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(./.env.*)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(git push --force:*)",
"Read(**/*.pem)",
"Read(.env)",
"Read(~/.aws/**)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Bash(rm -rf *)",
"Bash(sudo *)",
"Bash(sudo:*)",
"Bash(rm -rf /)",
"Bash(git push -f:*)",
"Read(.env.*)",
"Bash(git reset --hard*)",
"Bash(rm -rf /*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(gh pr merge *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(gh pr merge:*)",
"Bash(gh pr create *)",
"Bash(git reset --hard:*)",
"Bash(npm publish:*)",
"Bash(git rebase *)"
]
}
} Subagents (15)
| Explore model: haiku | Read-only search - finds files, searches code and reads docs in the repo or on the web, then returns short structured results, not file dumps. Use it proactively for a wide search or a docs lookup. Sa |
| architect model: opus | Decides how spec-dna works - the information model, ADRs, the contracts between modules, the CLI and JSON surface, and new dependencies. Writes no feature code. Use it to decide a design question, or |
| checker model: opus | Independent check of finished work - runs the tests, reads the diff and tries edge cases to break the change, then reports each real problem. Never edits code. Use it proactively before you say a task |
| context-engineer model: sonnet | Builds spec-dna context and show - the section plan, budgets, listed sections, inactive documents, the BM25 rank and the retrieval benchmark. Use for a change to what an agent receives. |
| core-engineer model: sonnet | Builds the core of spec-dna - parse, front matter, model, config and its schema, hashes, globs, git access, the cache, JSON output, templates, the new command and the CLI. Use for a change in these mo |
| diagnostics-engineer model: sonnet | Builds spec-dna check and its diagnostics - controls, base-relative gating, drift, code citations, evidence, the requirement lint and the SD codes. Use for a new or changed diagnostic, or a change to |
| docs-editor model: sonnet | Edits prose to ASD-STE100 Simplified Technical English in docs, README, CHANGELOG, ADRs, code comments and agent text, with no change of technical meaning. Use to review or fix the language of a chang |
| failure-analyst model: opus | Proves the cause of a failure with a small repro and command output, before a fix starts. Never fixes. Use before you fix a failure with no obvious cause. |
| fallback-advisor model: opus | Opus advisor for a decision point while the advisor tool is off or fails - an approach to pick, an error that comes back, a claim that a task is done. Advises only and never writes code. |
| hygiene-engineer model: sonnet | Builds the hygiene commands - mv, the split of requirements and sections, history, index and --stub files - and their preservation check. Use for a change that moves, splits or generates document text |
| init-engineer model: sonnet | Builds spec-dna init - detection, the questions, the preview, the generated adapters and the skill text - and keeps the quickstart and the demo true. Use for a change to init or to adapter text. |
| ownership | — |
| release-engineer model: sonnet | Prepares a release and keeps CI correct - the version, CHANGELOG, cross-builds, GitHub workflows, pre-commit hooks and Dependabot. Never pushes, tags a remote or merges. Use for a release or a CI chan |
| test-engineer model: sonnet | Owns the test strategy - tests, golden files, fixtures and the tool benchmark. Use to design the tests of a change, to fix a slow or unstable test, or to measure speed. |
| transcript-engineer model: sonnet | Builds spec-dna transcript - transcript formats, token and tool counts, the documents that an agent read - and guards its privacy rule. Use for a change to the transcript analysis or for a new transcr |
Similar rigs
davidpoblador/dotfiles
Dotfiles for macs and Linux hosts, converged by mise bootstrap: symlinked configs, brew and mas packages, macOS defaults, systemd/launchd services, fnox-encrypted secrets
YOLO Cowboy 6.4k tok · · Q8
henryzhangpku/agent-starter
Run a coding agent like a new hire: a written working agreement, guardrails enforced by hooks, a fresh-context reviewer, plan and decision logs. One-line bootstrap.
Orchestrator 1.4k tok · · Q20
helmut-hoffer-von-ankershoffen/heinzel
Heinzelmännchen: helpers accelerating product teams with a CLI, web UI, and MCP tools — like the little elves from Cologne.
Fort Knox 5.8k tok · · Q20
roadhero/claude-code-setup
My actual Claude Code setup: a stack-agnostic engineering spine, auto-activating rule packs, 50 scoped subagents across web/Android/iOS/compute, safety hooks, and a repo scaffolder. From the book "42: The AI Builder's Stack."
Orchestrator 7.3k tok · · Q32