Your agents don't need
more power.
They need a process.
HiveSpec is a spec-driven delivery lifecycle for AI agent swarms. Each agent follows the same disciplined phases — from claiming an issue to shipping a verified PR.
claude plugin add EntityProcess/hivespec Spec. Build. Verify. Ship.
Every agent follows the same disciplined path from issue to merged PR
Spec
Claim the issue, explore the codebase, and write a design spec before touching code.
Build
Break the spec into tasks. Test-driven implementation. Red, green, refactor, commit.
Verify
Run real commands. Check blast radius. Evidence over assertions.
Ship
Classify risk. Auto-merge safe changes. Flag breaking changes for review.
Agents ship fast. They don't ship right.
No coding without specs
Agents jump straight to code. Skip the design. Break three files to fix one.
No guessing at impact
They modify a shared interface without checking who depends on it.
No "tests pass" without proof
They claim everything works. No command output. No evidence.
No blast radius check
They ship a PR without verifying consumers of modified code still work.
Every phase has a gate
Spec before code
hs-design blocks implementation until a design is written and approved.
Explore before designing
hs-explore finds all consumers of shared interfaces before proposing changes.
Red before green
hs-implement watches tests fail before writing implementation code.
Evidence before claims
hs-verify requires actual command output, not assertions.
Risk before merge
hs-ship classifies risk and requires confirmation for elevated-risk PRs.
Swarm-ready
Ten agents running HiveSpec concurrently is a swarm that delivers like a team.
Works with any agent
SKILL.md files. No lock-in. Any agent that reads them can run HiveSpec.
The spec can live anywhere
Give your agents discipline.
Install HiveSpec in one command and start shipping with confidence.
claude plugin add EntityProcess/hivespec