菜单
Generate documentation by traversing the SeleneDB graph: milestones, design decisions, security concerns, and deferred work. Assembles structured docs from graph annotations and relationships.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Multi-agent coordination via SeleneDB context bridge. Check active peers, read shared context, declare work intents, and avoid conflicts when multiple agents work simultaneously.
Commit discipline with graph linking. MERGE GitCommit nodes and link to active milestones, findings, and decisions. Verify before committing, never push unless asked.
Background annotation discipline. All skills auto-annotate graph nodes with rationale, observations, TODOs, and bookmarks as they work. Captures context that feeds into graph-docs documentation generation.
Track and recall context across sessions. Auto-captures rolling summaries as skills complete. Invoke to review session history, search past work, or see what happened recently.
Language-specific standards plus graph-sourced conventions. Reads Convention nodes from SeleneDB, enforces project rules alongside built-in standards. Promotes recurring violations to new conventions.
Structured multi-perspective debate for evaluating decisions. 3-phase methodology: independent generation, adversarial exchange, synthesis. Use when a decision has multiple valid approaches or high stakes.
| name | graph-docs |
| description | Generate documentation by traversing the SeleneDB graph: milestones, design decisions, security concerns, and deferred work. Assembles structured docs from graph annotations and relationships. |
| argument-hint | [document type: milestone-report | architecture | security-posture | decision-log | deferred-status | full-project] |
| disable-model-invocation | true |
Generate documentation by traversing the property graph rather than scanning files. The graph contains design decisions with rationale notes, security concerns with mitigation trails, milestones with commit histories, and deferred work with gate conditions. This skill assembles those into structured documents.
When NOT to use: Generating API docs from code (no dedicated skill for this — use standard tooling). Writing new docs from scratch (just write them).
Create a session per selene-integration.md.
Query graph stats for the project to understand available data:
MATCH (n)
WHERE n.project = $project
RETURN labels(n) AS type, count(*) AS total
ORDER BY total DESC
Determine document type from $ARGUMENTS. If not specified, ask the user:
| Type | What it generates | Key traversals |
|---|---|---|
milestone-report | Status report for a milestone | milestone→commits→decisions→notes |
architecture | Architecture overview | crates→modules→dependencies→conventions |
security-posture | Security audit summary | security_concerns→affected deps→mitigations |
decision-log | Chronological decision record | design_decisions by date with rationale notes |
deferred-status | Outstanding deferred work | deferred_work with gates and blocking relationships |
full-project | All of the above combined | Full graph traversal for the project |
Execute the traversal for the selected document type:
milestone-report:
MATCH (m:Milestone)
WHERE m.project = $project AND m.name = $milestone_name
OPTIONAL MATCH (c:GitCommit)-[:part_of]->(m)
OPTIONAL MATCH (doc:Document)-[:part_of]->(m)
OPTIONAL MATCH (d:Decision)-[:part_of]->(m)
OPTIONAL MATCH (n:Note)-[:annotates]->(m)
RETURN m, collect(DISTINCT c) AS commits,
collect(DISTINCT doc) AS documents,
collect(DISTINCT d) AS decisions,
collect(DISTINCT n) AS notes
architecture:
MATCH (cr:crate)-[:belongs_to_project]->(p:project {name: $project})
OPTIONAL MATCH (cr)-[:contains]->(mod:module)
OPTIONAL MATCH (cr)-[:depends_on]->(dep)
OPTIONAL MATCH (conv:Convention {active: true})
WHERE conv.project = $project
RETURN cr, collect(DISTINCT mod) AS modules,
collect(DISTINCT dep) AS dependencies,
collect(DISTINCT conv) AS conventions
security-posture:
MATCH (sc:SecurityConcern)
WHERE sc.project = $project
OPTIONAL MATCH (sc)-[:affects]->(target)
OPTIONAL MATCH (sc)-[:mitigated_by]->(fix)
RETURN sc.summary, sc.severity, sc.status, sc.category,
labels(target) AS affected_type, target.name AS affected_name,
labels(fix) AS fix_type
ORDER BY sc.severity, sc.status
decision-log:
MATCH (d:Decision)<-[:produced]-(s:Session)
WHERE s.project = $project
OPTIONAL MATCH (n:Note {kind: 'rationale'})-[:annotates]->(d)
RETURN d.summary, d.rationale, d.alternatives, d.confidence,
s.date, s.skill, n.content AS extra_rationale
ORDER BY s.date DESC
deferred-status:
MATCH (di:DeferredItem)
WHERE di.project = $project AND di.status = 'open'
OPTIONAL MATCH (di)-[:gated_by]->(g:Gate)
OPTIONAL MATCH (di)-[:blocks]->(blocked:DeferredItem)
RETURN di.item, di.description, di.priority, di.kind,
g.condition, g.met,
collect(blocked.item) AS blocks
ORDER BY di.priority
Render traversal results as structured markdown. Include:
Present the draft to the user for review before writing.
After the user approves the generated document:
INSERT (doc:Document {
project: $project,
title: $title,
doc_type: $doc_type,
content: $markdown_content
})
RETURN id(doc) AS doc_id
MATCH (s:Session) WHERE id(s) = $session_id
MATCH (doc:Document) WHERE id(doc) = $doc_id
INSERT (s)-[:produced]->(doc)
Optionally write to a file if the user requests it.
| Rationalization | Why It's Wrong |
|---|---|
| "Just query the graph manually" | Structured docs need assembly logic — traversal, grouping, and cross-referencing. Raw queries dump data, not documentation. |
| "Docs are better written from scratch" | Graph-sourced docs capture what actually happened (decisions, findings, commits). Hand-written docs drift from reality. |