EphemeraAdmin: Update: task_type_override, custom_pages, create_stub/stub_expansion, routing table

2026-04-09T19:17:57Z

Update: task_type_override, custom_pages, create_stub/stub_expansion, routing table

EphemeraAdmin: Import: Help:Ephemera Agent/Context Assembly

2026-04-01T04:29:34Z

Import: Help:Ephemera Agent/Context Assembly

New page

{{DISPLAYTITLE:Ephemera Agent — Context Assembly}}
__TOC__

== Overview ==

Before any generative LLM call, <code>agent.php</code> assembles a structured context package from the wiki. This ensures the generator model has access to relevant canon, instructions, and evidence without needing to make additional API calls during generation.

The pipeline is active when <code>agent.php</code> is deployed on your server. For simple query tasks, it is bypassed entirely via a fast pre-check.

== When Context Assembly Is Bypassed ==

The system may skip context assembly and call the LLM directly when:

* The task is classified as a simple query
* No page creation or editing is required
* The user is performing a quick lookup, browse action, or lightweight maintenance task

== Pipeline Steps ==

# '''Load Instructions:Config''' from the wiki to establish runtime parameters
# '''Fast pre-check''' (<code>is_obviously_query()</code>) — if the task is clearly read-only (starts with "show", "list", "search", etc. and contains no generative keywords), skip the entire pipeline and route directly to the generator
# '''Planner LLM call''' — classify task type and extract named entities (cheap/fast model)
# '''Load World Bible''' — always first; anchors all generation to the universe
# '''Load core instructions''' — Canon Policy, Continuity Rules
# '''Resolve instruction dependency graph''' — BFS traversal of <code>requires</code> links, up to <code>max_depth</code> hops
# '''Fetch encyclopedia pages''' — one per named entity
# '''1-hop link expansion''' — fetch top-k related pages ranked by co-occurrence frequency across seed pages
# '''Search Sources: namespace''' — top results per entity
# '''Fetch filtered talk sections''' — Reliability Assessment, Bias Analysis, Editorial Notes only (never full talk pages)
# '''Wikipedia lookup''' — if enabled, fetch the introductory paragraph for each entity from the Wikipedia API
# '''Deduplicate''' — remove pages that appear in multiple sections
# '''Rank by keyword overlap''' — score pages by overlap with the task description using TF-style scoring
# '''Trim to token budget''' — drop lowest-priority content first (Wikipedia → talk → sources → encyclopedia). Instructions are never trimmed.
# '''Serialise''' — assemble into a structured prompt string with section headers

== Context Prompt Format ==

The assembled context is prepended to the generator's system prompt in this order:

<pre>
[INSTRUCTIONS]
--- Instructions:World Bible ---
{content}
--- Instructions:Core/Canon Policy ---
{content}
...

[ENCYCLOPEDIA CONTEXT]
--- PageTitle ---
{content}
...

[SOURCES]
--- Sources:DocumentTitle ---
{content}
...

[EDITORIAL NOTES]
--- Talk: Sources:DocumentTitle ---
= Reliability Assessment =
{content}
= Bias Analysis =
{content}
...

[REAL-WORLD REFERENCE — Wikipedia]
Factual real-world information for grounding only.
Do NOT treat as in-universe canon for Encyclopedia Ephemera.
--- EntityName ---
Wikipedia: ArticleTitle
{intro paragraph}
</pre>

== Task Types ==

The planner model classifies each task into one of these types:

{| class="wikitable" style="width:100%"
! Type !! Description
|-
| <code>create_encyclopedia_article</code> || Create a new encyclopedia article about an entity, event, or concept
|-
| <code>create_source</code> || Create an in-universe source document
|-
| <code>edit_page</code> || Edit or update an existing wiki page
|-
| <code>maintenance_report</code> || Produce a maintenance or audit report
|-
| <code>query</code> || Read-only information retrieval
|-
| <code>batch_operation</code> || Perform the same operation across multiple pages
|-
| <code>stub_expansion</code> || Expand a stub or incomplete article
|-
| <code>red_link_generation</code> || Create a page for a detected red link
|-
| <code>unknown</code> || Could not be determined — defaults to generative path
|}

Additional task types can be added via <code>[[Instructions:Config]]</code> without PHP changes.

== Instruction Dependency Graph ==

Each instruction page declares its dependencies in a <code>requires</code> metadata field. The pipeline resolves this graph using breadth-first traversal:

* '''max_depth''' (default: 2) — stop after this many hops
* '''fully_resolve''' (default: false) — if true, ignore max_depth and resolve the complete graph

Example: selecting <code>Instructions:Create/Source/Interview</code> automatically also fetches <code>Instructions:Create/Source (Base Workflow)</code> (depth 1) and <code>Instructions:Core/Continuity Rules</code> (depth 2).

== Debugging ==

The '''▶ CONTEXT''' panel in the agent log shows what was assembled after each context-aware call. For deeper debugging, the raw <code>context_meta</code> object is included in every <code>agent.php</code> response and visible in browser developer tools.

[[Category:Help]]
[[Category:EphemeraAgent]]

Help:Ephemera Agent/Context Assembly - Revision history

EphemeraAdmin: Update: task_type_override, custom_pages, create_stub/stub_expansion, routing table

EphemeraAdmin: Import: Help:Ephemera Agent/Context Assembly