Evidence the run may read
- Employee question
- Approved policy and process library
- Jurisdiction and employment-context metadata
production blueprint · people operations
People policy answer draft agent blueprint.
Draft a response from current approved people policies and route exceptions or individual determinations to a specialist.
Search intent: AI agent for HR policy questions
Published by a2a cloudProduct-source methodology →
01 · the contract
Start from a bounded job, not a blank chat box.
trigger
An employee asks a policy or process question through an approved channel.
owner
people operations owner
agent stops at
A people advisor approves the answer, especially any individual or jurisdiction-specific interpretation.
KPI · increase
policy answers accepted first review (percent). Measure drafts approved without a material policy correction.
outputs
Artifacts the run must produce
- Policy-cited answer draft
- Applicability and uncertainty notes
- Specialist escalation recommendation
02 · topology
Small specialists. Named handoffs. One accountable decision.
Each stage produces an artifact another stage can inspect. The final node is a person, not an autonomous write to an external system.
- 01
intake
Validate and normalize the people policy answer draft inputs.
- 02
question classifier
Produce the policy-cited answer draft.
- 03
policy retriever
Produce the applicability and uncertainty notes.
- 04
people advisor reviewer
Challenge the people policy answer draft result and prepare an approval packet.
- 05
people operations owner
A people advisor approves the answer, especially any individual or jurisdiction-specific interpretation.
intake→ validated input packet →question-classifier
question-classifier→ Policy-cited answer draft →policy-retriever
policy-retriever→ Applicability and uncertainty notes →people-advisor-reviewer
people-advisor-reviewer→ approval packet with evidence references →human-approver
03 · authority
Grant the run only what this case needs.
Source material is read-only. Drafts land in a case-specific output path. Tools may read or propose; the human gate owns the external write.
case inputs
workspace/people-ops/people-policy-answer-draft/inputs/**Read only the evidence attached to this workflow instance.
case outputs
workspace/people-ops/people-policy-answer-draft/outputs/**Write drafts and evidence artifacts without modifying source records.
approved tools
people-ops:people-policy-answer-draft:read-or-proposeInvoke only tools explicitly granted for this run; external writes remain gated.
04 · implementation
A private, bounded starting manifest.
The blueprint starts private, caps its DAG, disables replanning, and exposes no public endpoint. Add only the tools and data adapters this workflow has approved.
a2a.yamlsafe starting point
name: people-ops-people-policy-answer-draft
version: 0.1.0
entrypoint: agent:BlueprintAgent
expose:
public: false
composition:
planning: deterministic_dag
max_nodes: 6
max_parallel: 1
max_replans: 005 · acceptance test
Pass only with evidence.
- 01
Evidence traceability
Every material conclusion cites an input artifact or a scoped tool result from this run.
- 02
The draft cites current policy, protects confidential data, and distinguishes general process from an individual determination.
The draft cites current policy, protects confidential data, and distinguishes general process from an individual determination.
- 03
Approval boundary
The run stops at a proposal and records the human decision before any external side effect.
failure containment
Stop small. Preserve the evidence.
The applicable policy, jurisdiction, employee context, or confidentiality boundary cannot be verified.
Containment: Return a partial result with unresolved items; do not broaden scope or perform an external write.
Operator: Attach the missing evidence, narrow the brief, or explicitly approve a new scoped run.
A required input or tool grant is unavailable.
Containment: Stop the affected branch and preserve completed artifacts in the case output workspace.
Operator: Grant only the missing resource or continue with that branch marked out of scope.
06 · proof
Sign the run facts. Keep money in the billing ledger.
Current platform receipts sign caller identity or classification, skill, bounded input evidence, verified grant IDs when present, outcome or result preview, and timing. Optional file, tool, artifact, handoff, evaluation, and review fields require separate instrumentation and are not populated by default. Price, fees, payouts, and later human approvals remain separate platform records.
The example uses only fields populated by the current platform sealing paths. It is illustrative, not a record of a real customer run.
ExecutionReceipt · selected fieldsEd25519 token
{
"receipt_id": "rcpt_01J...",
"schema_version": 1,
"agent_name": "people-ops-people-policy-answer-draft",
"caller": "user:workflow-owner",
"task_id": "case_people_policy_answer_draft",
"skill_name": "people_policy_answer_draft",
"input_hash": "4d7c...9a2f",
"grant_ids": [
"grt_case_inputs",
"grt_tool_propose"
],
"status": "ok",
"result_preview": "Output prepared: Policy-cited answer draft. Human decision remains separate.",
"elapsed_ms": 4218
}related blueprints
Keep the operating model consistent.
People Operations
Role description review
Check a proposed role description for clarity, internal consistency, approved requirements, and accessible language before publication.
People Operations
Candidate intake completeness
Validate application completeness and scheduling prerequisites without ranking candidates or inferring suitability.
People Operations
Interview loop packet
Assemble an approved competency plan, interviewer assignments, and structured evidence forms for a recruiting coordinator.
put the pattern to work
Start private. Scope the authority. Require the decision.
Deploy the workflow as a bounded internal agent, verify its outputs and the receipt fields actually emitted, then expand only the scopes your acceptance test proves it needs.