hr-human-collaboration
Skill1 starsChanged 6 months agoApache-2.0View on GitHub
Collaboration plane (Harness HITL) — seven-phase human-agent loop: proposal, options, human tradeoffs, execute, revalidate, specialized tools, output review. Use at T1+ before specialized skills; load routing portion automatically at orchestration.
In this repository37 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/agent-ai-governance.mdcCursor rule
- .cursor/rules/agent-architecture.mdcCursor rule
- .cursor/rules/agent-code-health.mdcCursor rule
- .cursor/rules/agent-db-migration-state.mdcCursor rule
- .cursor/rules/agent-developer-advocate.mdcCursor rule
- .cursor/rules/agent-devops-lifecycle.mdcCursor rule
- .cursor/rules/agent-finops.mdcCursor rule
- .cursor/rules/agent-integrations.mdcCursor rule
- .cursor/rules/agent-legal-hr-compliance.mdcCursor rule
- .cursor/rules/agent-mlops.mdcCursor rule
- .cursor/rules/agent-nextjs.mdcCursor rule
- .cursor/rules/agent-packaging-supply-chain.mdcCursor rule
- .cursor/rules/agent-qa.mdcCursor rule
- .cursor/rules/agent-security.mdcCursor rule
- .cursor/rules/orchestrator-hr-erp.mdcCursor rule
- .cursor/rules/repo-boundaries.mdcCursor rule
- .cursor/rules/site-contract.mdcCursor rule
- .cursor/rules/vercel-jwt-smoke.mdcCursor rule
- llms.txtllms.txt
- hr-contributor-handoffSkill
- hr-data-custodySkill
- hr-devops-lifecycleSkill
- hr-domain-boundariesSkill
- hr-foundation-governanceSkill
- hr-governance-learningSkill
- hr-orchestration-lanesSkill
- hr-product-gateSkill
- hr-product-mcp-governanceSkill
- hr-quality-labSkill
- hr-regulated-domainSkill
- hr-swarm-governanceSkill
- site-deliverySkill
- code-reviewSkill
- hr-data-custodySkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 42
Timeout: never auto-approve revalidation.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
HR human collaboration (Collaboration plane)
Use when
- Orchestrating T1+ work (Orchestrator step 0)
- Presenting strategy options with risk/reward tradeoffs
- Recording human decisions before specialized skills or Task delegation
- Phase 7 output review against approved strategy
Do not use when
- T0 docs-only chore (
step 1 chore N/A) - Product employment-AI runtime actions — use Product HITL (
lib/governance/hitl.ts)
Seven-phase checklist
| Phase | Mode | Action |
| 1 Proposal | HOTL | Restate intent, tier, bounded contexts |
| 2 Options | HOTL | ≥2 strategies with risk, reward, rollback, lanes |
| 3 Human input | HITL | Record principal + timestamp; routing skills only |
| 4 Execute | HOTL | Implement approved plan |
| 5 Revalidate | HOTL | Drift check; human confirms before phase 6 |
| 6 Specialized | HITL | Tasks, MCP, @hr-regulated-domain, etc. |
| 7 Output review | HOTL | Verifier confirms deliverables match chosenStrategy |
Protocols
- Reject at 3 or 5: return to phase 2; append
rejectionHistory[] - Timeout: never auto-approve revalidation; deny specialized unlock or escalate to Human PO
- Challenge: human edits plan sections or alternate
chosenStrategy
Anti-patterns
- Prompt-only “ask user first” without hook/manifest gate
- Loading specialized skills before phase 6 / revalidation
- Timeout-as-approval
- Rubber-stamping options without reading rollback column
Resources
- references/collaboration-loop.md
- specs/templates/collaboration-plan.md
- ADR 0020
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

