discussion-management
Manage Discussion entities — structured, multi-turn conversations that explore questions and produce decisions. Use when a question or design choice needs multi-turn exploration before a decision is reached — RFCs, design debates, trade-off analyses.
- License 许可
- MIT
- 最近更新
- 2026-11-11
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- discussion-management
- metadata
- { "processkit": { "id": "SKILL-discussion-management", "uses": [ { "skill": "event-log", "purpose": "Log events to keep the audit trail accurate after every write." }, { "skill": "decision-record", "purpose": "Record consequential decisions made during this skill's workflow." }, { "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": 4, "created": 1775433600000, "version": "1.0.0", "category": "processkit", "commands": [ { "args": "topic or question to discuss", "name": "pk-discuss", "description": "Start a structured discussion on a topic: research, challenge assumptions, present alternatives, work toward a conclusion" } ], "provides": { "mcp_tools": [ "open_discussion", "get_discussion", "transition_discussion", "add_outcome", "list_discussions" ], "templates": [ "discussion" ], "primitives": [ "Discussion" ] }, "apiVersion": "processkit.projectious.work/v2" } }
- description
- Manage Discussion entities — structured, multi-turn conversations that explore questions and produce decisions. Use when a question or design choice needs multi-turn exploration before a decision is reached — RFCs, design debates, trade-off analyses.
02 资源文件 6
- 资产文件assets/discussion.yaml365 B
- 参考文档commands/pk-discuss.md305 B
- 参考文档mcp/mcp-config.json225 B
- 参考文档mcp/SERVER.md1.5 KB
- 脚本mcp/server.py9.0 KB
- 脚本scripts/.gitkeep—
03 SKILL.md 正文
Discussion Management
Intro
A Discussion is a structured, multi-turn conversation exploring a question. It captures the back-and-forth of research, proposals, objections, and convergence — producing (or failing to produce) a DecisionRecord as its outcome. Think of it as the audit trail behind a decision.
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: Discussion
metadata:
id: DISC-002
title: "aibox Refocus — Core Principles and Scope"
created: 2026-04-05
participants: [owner, claude]
spec:
state: active # active | resolved | archived
question: "What is aibox's core scope? What belongs in processkit? What is out of scope entirely?"
related: [DISC-001]
---
# Discussion body
## 1. Problem statement
...
## 2. What aibox IS
...
## Decisions (DEC-NNN records)
- DEC-017: Scope refocus
- DEC-018: Two-repo split
...
Workflow
- Start a Discussion when a non-trivial question arises.
- Write the
questionas one crisp sentence. - Use numbered sections in the body: problem statement, options considered, analysis, decisions.
- As decisions crystallize, write DecisionRecords and list them at the bottom of the Discussion.
- When converged, set
state: resolved. Logdiscussion.resolved.
Discussions vs DecisionRecords
| Discussion | DecisionRecord |
|---|---|
| Multi-turn, exploratory | Single outcome, crisp |
| Captures reasoning journey | Captures the destination |
| May produce 0–many DEC records | Produced by at most one Discussion |
| Often long-form | Tightly structured |
Use a Discussion when you don't yet know the answer. Use a DecisionRecord when you do.
Gotchas
Agent-specific failure modes — provider-neutral pause-and-self-check items:
- Opening a Discussion when the choice is already obvious. If the user already knows what they want, write a DecisionRecord directly. Discussions are for genuinely open questions where the reasoning trail matters; using one for a settled call wastes ceremony.
- Letting Discussions go stale. A Discussion with no activity for weeks should either resolve (with an outcome) or close (state: archived). Open-forever discussions clutter the index and signal pretend-thinking.
- Not recording the outcome as a DecisionRecord on convergence.
When a Discussion reaches an answer, the answer must become a
DecisionRecord referenced in the discussion's
outcomesfield. Without that promotion, the value of the discussion evaporates the next time someone asks "what did we decide". - Putting decisive content in the body but leaving
outcomesempty. Theoutcomesfield is what downstream queries read. If the body says "we decided X" butoutcomes: [], the index doesn't know a decision was made and can't link the discussion to the DecisionRecord. - Confusing Discussion (pre-decisional) with DecisionRecord (post-decisional). Discussions capture reasoning in flight; DecisionRecords capture settled choices. Don't write reasoning into a DecisionRecord, and don't write conclusions into a Discussion body without also creating the DecisionRecord.
- Forgetting to add participants. The
participantsfield is how queries like "what is Alice currently thinking about" work. An unattributed Discussion is invisible to participant queries. - Treating one user message as a "discussion". A Discussion is multi-turn by definition. If there's only one back-and-forth, it doesn't need the structure — answer directly or write a DecisionRecord.
Full reference
Fields
| Field | Type | Notes |
|---|---|---|
state |
enum | active / resolved / archived |
question |
string | The driving question |
participants |
list[string] | Actor IDs |
related |
list[string] | Other discussion IDs |
outcomes |
list[string] | DEC IDs produced |
opened_at |
datetime | When the discussion started |
closed_at |
datetime | When it was marked resolved |
Long-running discussions
Some discussions span weeks or months. Keep them in one file — use section dividers with dates rather than creating multiple files. The index MCP server can show discussion activity over time by reading LogEntries that reference it.
Archiving
When a discussion is old and no longer actively referenced, move it to
context/archive/discussions/ and set state: archived. Do not delete —
the reasoning is often the most valuable context for future work.
Example: DISC-002
The DISC-002 document in the aibox repo is a full example of a processkit Discussion — multi-section, multiple open questions resolved over time, producing DEC-017..DEC-024. It predates this skill (was written in plain Markdown) and can be retroactively brought into the format by adding the frontmatter block.
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.