gate-management
Manage Gate entities — validation checkpoints that WorkItems, scopes, or policy applications must pass (code review, security scan, tests). Use when defining or updating a checkpoint for v2 workflow surfaces — code review, tests green, security scan, stakeholder approval.
- License 许可
- MIT
- 最近更新
- 2026-11-11
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- gate-management
- metadata
- { "processkit": { "id": "SKILL-gate-management", "uses": [ { "skill": "event-log", "purpose": "Log events to keep the audit trail accurate after every write." }, { "skill": "index-management", "purpose": "Query existing entities and keep the SQLite index fresh after writes." }, { "skill": "id-management", "purpose": "Allocate unique entity identifiers via central ID generation." } ], "layer": 3, "created": 1775433600000, "version": "1.0.0", "category": "processkit", "provides": { "mcp_tools": [ "create_gate", "get_gate", "list_gates", "evaluate_gate" ], "templates": [ "gate" ], "primitives": [ "Gate" ] }, "apiVersion": "processkit.projectious.work/v2" } }
- description
- Manage Gate entities — validation checkpoints that WorkItems, scopes, or policy applications must pass (code review, security scan, tests). Use when defining or updating a checkpoint for v2 workflow surfaces — code review, tests green, security scan, stakeholder approval.
02 资源文件 6
- 资产文件assets/gate.yaml267 B
- 参考文档mcp/mcp-config.json213 B
- 参考文档mcp/SERVER.md1.6 KB
- 脚本mcp/server.py15.7 KB
- 脚本scripts/.gitkeep—
- 脚本scripts/test_gate_management.py2.8 KB
03 SKILL.md 正文
Gate Management
Intro
A Gate is a validation checkpoint — "code review passed", "tests green", "security scan clean", "stakeholder signed off". In v2, Bindings attach gates to concrete workflow surfaces such as WorkItems and Scopes.
MCP server. This skill ships a self-contained MCP server at
mcp/server.py(PEP 723 script — requiresuvand Python ≥ 3.10 on PATH). Agent harnesses reach its tools by reading a single MCP config file at startup, so the contents ofmcp/mcp-config.jsonmust be merged into the harness's MCP config and placed at the harness-specific path before this skill is usable. If processkit was installed by an installer, that wiring is the installer's responsibility; if processkit was installed manually, the project owner must do it by hand.
Overview
Shape
---
apiVersion: processkit.projectious.work/v2
kind: Gate
metadata:
id: GATE-code-review
created: 2026-04-06T00:00:00Z
spec:
name: code-review
description: "At least one reviewer has approved the change."
kind: manual # manual | automated | hybrid
validator: "A second actor reviews and approves."
required_roles: [reviewer]
blocking: true # if false, gate is advisory only
---
Workflow
- Pick
GATE-<name>where name is a short noun phrase. - Set
kind:manual(human does it),automated(check runs automatically), orhybrid(both). - Describe the
validator— the concrete check. One or two sentences. - List
required_rolesif specific actors must do the validation. - Set
blocking: truefor hard gates (cannot proceed),falsefor advisory.
Gate outcomes
Gates emit LogEntries when they are evaluated:
gate.passed— validation succeededgate.failed— validation failed (include failure details)gate.waived— validation was bypassed (includedetails.waived_byand reason)
Gotchas
Agent-specific failure modes — provider-neutral pause-and-self-check items:
- Defining a Gate without an evaluator. A Gate that says "code review passed" but doesn't say HOW that's checked is decoration, not a checkpoint. Every Gate must have either an automated check (script, MCP tool, CI job) or a clearly named manual approver.
- Auto-approving manual gates. If a Gate is
kind: manual, the agent does NOT have authority to mark it passed. Surface the gate to the user, ask for explicit approval, and only then callevaluate_gatewith the approver's ACTOR-ID. - Missing the rollback path on failure. Every Gate that can fail must specify what happens when it does — block the process, page oncall, escalate, defer to a different gate. A Gate that fails silently is worse than no gate.
- Writing gates that are too coarse. "Tests pass" is too vague — which suite? Which subset? With what coverage threshold? Granularity matters because the agent needs to know what specifically is being checked when the gate is invoked.
- Forgetting to log
gate.passed/gate.failed/gate.waived. These events are how the audit trail tracks process health. If you evaluate a gate without logging the outcome, the index can't answer "how often does this gate fail" or "who waived it last". - Hand-waving "automated" gates that aren't actually wired. If
kind: automated, the evaluator must exist and be runnable from the agent's environment. Marking a gate automated when it actually requires a human is a process failure dressed as a technical one. - Treating gate failures as warnings. Gates are blockers by
design. If a Gate fails, the dependent process step does NOT
proceed, even under pressure. If the user wants to bypass, that's
a
gate.waivedevent with the waiver's name and reason — never an unrecorded continuation.
Full reference
Fields
| Field | Type | Notes |
|---|---|---|
name |
string | Short kebab-case identifier |
description |
string | What this gate checks |
kind |
enum | manual / automated / hybrid |
validator |
string | Prose description of the check |
validator_command |
string | Optional. CLI command that automates the check |
required_roles |
list[string] | Roles that can sign off |
blocking |
bool | true = cannot proceed if failed |
evidence_required |
bool | Require a link/artifact in the pass event |
Binding gates to v2 workflow surfaces
Use workitem-gate for a concrete run or task, and scope-gate for a
gate that applies across a Scope. Legacy process-gate Bindings are
migration-only.
kind: Binding
spec:
type: workitem-gate
subject: BACK-release-run
target: GATE-security-scan
scope: SCOPE-project-x
This way the same Gate can apply to multiple runs or scopes, and the mapping changes without editing the Gate or governed entity.
Waiving a gate
If a blocking gate is bypassed, log gate.waived with details.waived_by
(actor ID), details.reason, and details.expires_at. Do not edit the
Gate definition — the waiver is an event, not a property.
04 同仓库更多 Skills
standard-ui-design
暂无简介
standard-agent-use
Company standard for working with AI agents. Use when deciding what to delegate to agents, capturing open work, keeping agent memory or knowledge notes, improving an agent workflow, or introducing agent work in a team or for a client.
standard-agent-building
暂无简介
skill-management
Manage Skill package manifests as v1 ontology entities.
schema-management
Manage generated processkit schema contracts through deterministic MCP rebuild and inspection workflows.
proposition-management
Record claims, risks, beliefs, world facts, WSJF estimates, and assumptions through the Proposition interface.