Skip to content

Self-improving AI Employees for everyone

docs: fix stale labels and counts in the diagrams and agent cards

trohitg4cbc388596fe
diff --git a/README.md b/README.md
index e8945a26..ca4c54d4 100644
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@ Download the app for your computer and install it (these links always point at t
- **macOS, Intel**: [OpenCompany-macos-x64.dmg](https://github.com/zeenie-ai/OpenCompany/releases/latest/download/OpenCompany-macos-x64.dmg)
- **Linux**: [OpenCompany-linux-x86_64.AppImage](https://github.com/zeenie-ai/OpenCompany/releases/latest/download/OpenCompany-linux-x86_64.AppImage), or [OpenCompany-linux-amd64.deb](https://github.com/zeenie-ai/OpenCompany/releases/latest/download/OpenCompany-linux-amd64.deb) on Debian and Ubuntu
-Nothing else to install. On first launch the app sets itself up (a one-time download, a minute or two) and opens Home. Describe a job to hire your first employee; if no AI provider is connected yet, Home asks you to connect one, or you can add it yourself in Settings > Connectors. Switch to Dev mode for the workflow editor, where three example employees are already on the canvas; open one to see how it is put together.
+Nothing else to install. On first launch the app sets itself up (a one-time download, a minute or two) and opens Home. Describe a job to hire your first employee; if no AI provider is connected yet, Home asks you to connect one, or you can add it yourself in Settings > Connectors. Switch to Dev mode for the workflow editor, where three example employees are listed in the workflow sidebar; open one to see how it is put together.
The builds are not code-signed yet, so macOS asks you to allow the app under System Settings > Privacy & Security, and Windows SmartScreen needs "More info > Run anyway".
diff --git a/docs-internal/node-logic-flows/ai_agents/aiAgent.md b/docs-internal/node-logic-flows/ai_agents/aiAgent.md
index 79bf0e00..95892b9b 100644
--- a/docs-internal/node-logic-flows/ai_agents/aiAgent.md
+++ b/docs-internal/node-logic-flows/ai_agents/aiAgent.md
@@ -11,14 +11,14 @@
## Purpose
`aiAgent` is the general-purpose tool-calling agent node. It reads a prompt
-and system message, optionally merges conversation history from a connected
-`simpleMemory`, loads instructions from connected skill nodes, binds tool nodes
-as provider-neutral `AgentToolSpec` values, and runs
-`run_native_agent_loop` until the LLM produces a final answer. All of the
-heavy lifting (native SDK invocation, tool execution, memory persistence)
-lives behind `AIService.execute_agent`; the handler's job is purely to gather
-the connected-node payloads via
-`edge_walker.collect_agent_connections` (a 5-tuple: memory, skill, tool,
+and system message, continues its stored conversation when a Context node is
+connected on `input-context` (RFC-0002), loads instructions from connected
+skill nodes, binds tool nodes as provider-neutral `AgentToolSpec` values, and
+runs `run_native_agent_loop` until the LLM produces a final answer. All of the
+heavy lifting (native SDK invocation, tool execution, conversation
+persistence) lives behind `AIService.execute_agent`; the handler's job is
+purely to gather the connected-node payloads via
+`edge_walker.collect_agent_connections` (a 5-tuple: context, skill, tool,
input, task) and forward them.
## Inputs (handles)
@@ -27,8 +27,8 @@ input, task) and forward them.
|--------|-----------------|----------|---------|
| `input-main` | main | no | Upstream data. Used as auto-prompt when `prompt` is empty. |
| `input-skill` | main | no | Skill nodes (`masterSkill` is expanded into individual skills). |
-| `input-memory` | main | no | A `simpleMemory` node whose `memory_content` markdown is parsed into native `Message` values. |
-| `input-tools` | main | no | Tool nodes (search, calculator, HTTP, Android toolkit, child agents, etc.). |
+| `input-context` | main | no | A Context node. The agent loads its stored conversation at run start and saves it after every turn, only in a started workflow. Legacy `simpleMemory` -> `input-memory` edges are rewritten on load into a Context node plus a Memory tool edge. |
+| `input-tools` | main | no | Tool nodes (search, calculator, HTTP, Android service nodes, the Memory tool, child agents, etc.). |
| `input-task` | main | no | `taskTrigger` output - completed delegated-task payload. |
## Parameters
@@ -82,7 +82,7 @@ flowchart TD
G -- yes --> H[Extract message/text/content<br/>from input_data<br/>-> parameters.prompt]
G -- no --> I
H --> I[Get status broadcaster]
- I --> J[await ai_service.execute_agent<br/>with memory/skill/tool/broadcaster]
+ I --> J[await ai_service.execute_agent<br/>with context/skill/tool/broadcaster]
J --> K[Return envelope]
```
@@ -90,9 +90,10 @@ flowchart TD
- **`edge_walker.collect_agent_connections`** scans `context.edges` for edges whose target
is this node, then routes each by `targetHandle`:
- - `input-memory` + source type `simpleMemory` -> build `memory_data` dict.
- `session_id` is auto-set to the agent's `node_id` unless the memory node
- explicitly sets `sessionId` to a non-empty, non-`default` value.
+ - `input-context` -> the agent-context builder registered by the Context
+ plugin returns the conversation descriptor (`context_data`). Legacy
+ `input-memory` + `simpleMemory` edges in old graphs still yield their
+ recorded Memory descriptor.
- `input-skill` -> append to `skill_data`. `masterSkill` is expanded into
one entry per **enabled** skill in `skillsConfig`; instructions come from
the DB-stored `instructions` first, with a skill-folder fallback.
@@ -125,9 +126,9 @@ flowchart TD
## Side Effects
- **Database writes**: none directly in `execute_op`. `AIService.execute_agent`
- writes `token_usage_metrics` rows (via `CompactionService.track`), and
- `memory_content` markdown changes persist back through the
- `simpleMemory` node's `save_node_parameters` path owned by the service.
+ writes `token_usage_metrics` rows (via `CompactionService.track`). With a
+ Context node connected, the conversation is saved to `agent_conversations`
+ after every turn of a started workflow; a manual node Run saves nothing.
- **Broadcasts**: `prepare_agent_call` resolves `StatusBroadcaster` and passes
it into `execute_agent`. The service then emits `update_node_status`
(`thinking`, `executing_tool`, `success`, ...) plus `token_usage_update`
@@ -159,9 +160,10 @@ flowchart TD
## Edge cases & known limits
-- **Memory session auto-derivation**: two agents sharing the same `simpleMemory`
- will each get a **different** session (each uses its own `node_id`) unless
- the memory node's `sessionId` is set to a non-empty, non-`default` value.
+- **One conversation per agent**: the stored conversation is keyed by
+ `(workflow_id, generation, agent_node_id)`, so every caller of a started
+ workflow continues the same conversation. Reset admits a new generation and
+ clears the stored conversations.
- **`masterSkill` expansion swallows missing skills**: if a skill key has no
`instructions` in DB and the skill loader raises, the skill entry is still
appended with an empty instruction body (logged as a warning, not an error).
@@ -187,7 +189,7 @@ flowchart TD
(android_agent, coding_agent, web_agent, ...). See
[`chatAgent.md`](./chatAgent.md) for the same contract with a different
system prompt.
-- **Memory node**: [`simpleMemory.md`](./simpleMemory.md)
+- **Context node**: [`context.md`](./context.md); **Memory tool**: [`simpleMemory.md`](./simpleMemory.md)
- **Architecture docs**: [Agent Architecture](../../agent_architecture.md),
[Agent Delegation](../../agent_delegation.md),
[Native LLM SDK](../../native_llm_sdk.md),
diff --git a/docs-internal/node-logic-flows/ai_agents/chatAgent.md b/docs-internal/node-logic-flows/ai_agents/chatAgent.md
index c1590124..c6e4ff0a 100644
--- a/docs-internal/node-logic-flows/ai_agents/chatAgent.md
+++ b/docs-internal/node-logic-flows/ai_agents/chatAgent.md
@@ -30,7 +30,7 @@ own plugin folder; there is no `functools.partial` wiring anymore.
|--------|-----------------|----------|---------|
| `input-main` | main | no | Upstream data. Auto-prompt fallback when `prompt` is empty. |
| `input-skill` | main | no | Skill nodes (including `masterSkill` aggregation). |
-| `input-memory` | main | no | `simpleMemory` node for conversation history. |
+| `input-context` | main | no | Context node: the stored conversation, loaded at run start and saved after every turn of a started workflow. Legacy `simpleMemory` -> `input-memory` edges are rewritten on load into a Context node plus a Memory tool edge. |
| `input-tools` | main | no | Tool nodes compiled into provider-neutral `AgentToolSpec` / `ToolDef` values for each native LLM step. |
| `input-task` | main | no | `taskTrigger` output - formatted and prepended to the prompt. |
| `input-teammates` | main | no | **Team-lead only** (`orchestrator_agent`, `ai_employee`). Agents on this handle become authorized Task Manager assignees. |
@@ -71,7 +71,7 @@ flowchart TD
H --> I{node_type in<br/>TEAM_LEAD_TYPES?}
I -- yes --> J[collect_teammate_connections<br/>-> append each teammate to tool_data]
I -- no --> K
- J --> K[await ai_service.execute_chat_agent<br/>memory/skill/tool/broadcaster/context/db]
+ J --> K[await ai_service.execute_chat_agent<br/>context/skill/tool/broadcaster/db]
K --> L[Return envelope]
```
@@ -79,7 +79,7 @@ flowchart TD
- **Connection collection** is delegated to
`edge_walker.collect_agent_connections`; same rules as `aiAgent` (see that doc
- for memory session derivation, `masterSkill` expansion, Android toolkit,
+ for the Context descriptor, `masterSkill` expansion, Android service tools,
child-agent tool discovery).
- **Native agent loop**: current executions call the shared
`run_native_agent_loop` / `run_native_llm_step` service through
@@ -100,10 +100,11 @@ flowchart TD
## Side Effects
-- **Database writes**: none directly. In-process execution can update a
- connected `simpleMemory` node and record compaction usage for that memory
- session; the Temporal workflow accumulates normalized usage in its result
- envelope. Standalone executions without connected memory do not persist a
+- **Database writes**: none directly. With a Context node connected, the
+ conversation is saved to `agent_conversations` after every turn of a started
+ workflow (a manual node Run saves nothing), and in-process execution records
+ compaction usage; the Temporal workflow accumulates normalized usage in its
+ result envelope. Executions without a connected Context do not persist a
token metric here.
- **Broadcasts**: `StatusBroadcaster` is fetched and passed down, enabling
`update_node_status` events (`thinking`, `executing_tool`, `success`), plus
@@ -146,7 +147,7 @@ flowchart TD
- **Twin node**: [`aiAgent`](./aiAgent.md) (same collection helper,
different service method).
-- **Memory node**: [`simpleMemory`](./simpleMemory.md)
+- **Context node**: [`context`](./context.md); **Memory tool**: [`simpleMemory`](./simpleMemory.md)
- **Architecture docs**: [Agent Architecture](../../agent_architecture.md),
[Agent Delegation](../../agent_delegation.md),
[Agent Teams](../../agent_teams.md),
diff --git a/docs-internal/node-logic-flows/ai_chat_models/ollamaChatModel.md b/docs-internal/node-logic-flows/ai_chat_models/ollamaChatModel.md
index 6515ca2e..82e51256 100644
--- a/docs-internal/node-logic-flows/ai_chat_models/ollamaChatModel.md
+++ b/docs-internal/node-logic-flows/ai_chat_models/ollamaChatModel.md
@@ -25,7 +25,7 @@ Run local LLMs (llama, mistral, qwen, deepseek-r1, ...) through a locally-runnin
|------|------|---------|----------|---------------------|-------------|
| `prompt` | string | `""` | yes | - | User message |
| `system_prompt` | string | `""` | no | - | System prompt |
-| `model` | string | `""` (injected) | no | - | Whatever the user has pulled, e.g. `qwen2.5`, `llama3.x`, `deepseek-r1`. Open-world: name not pattern-checked by `is_model_valid_for_provider` |
+| `model` | string | `""` (injected) | no | - | A model Ollama had loaded when the server was last fetched (`ps()` lists loaded models only), e.g. `qwen2.5`, `llama3.x`, `deepseek-r1`. Open-world: name not pattern-checked by `is_model_valid_for_provider` |
| `temperature` | number\|null | `null` | no | - | 0-2 |
| `max_tokens` | number\|null | `null` | no | - | 1-200000; default per-loaded-model ctx ÷ 4 (capped 4096) |
| `top_p` | number\|null | `1.0` | no | - | |
diff --git a/docs-internal/node-logic-flows/specialized_agents/_pattern.md b/docs-internal/node-logic-flows/specialized_agents/_pattern.md
index 0e24d80b..108e474f 100644
--- a/docs-internal/node-logic-flows/specialized_agents/_pattern.md
+++ b/docs-internal/node-logic-flows/specialized_agents/_pattern.md
@@ -52,7 +52,7 @@ All 13 agents share these input handles.
|--------|-----------------|----------|---------|
| `input-main` | input/main | no | Upstream data; auto-prompt fallback when `prompt` param is empty |
| `input-skill` | input/skill | no | Skill node(s) providing SKILL.md context |
-| `input-memory` | input/memory | no | `simpleMemory` node for conversation history |
+| `input-context` | input/context | no | Context node: the stored conversation, loaded at run start and saved after every turn of a started workflow |
| `input-tools` | input/tools | no | Tool nodes compiled into provider-neutral `AgentToolSpec` definitions |
| `input-task` | input/task | no | `taskTrigger` events from delegated child agents |
| `input-teammates` | input/teammates | no | **Only on `orchestrator_agent` and `ai_employee`** (via `team_lead_agent_handles()`) -- connected agents become Task Manager-authorized assignees |
@@ -106,7 +106,7 @@ Wrapped in the standard envelope:
flowchart TD
A[BaseNode.execute -> execute_op] --> P[prepare_agent_call]
P --> B[collect_agent_connections]
- B --> B1[memory_data from input-memory]
+ B --> B1[context_data from input-context]
B --> B2[skill_data from input-skill<br/>masterSkill expanded to N skills]
B --> B3[tool_data from directly connected input-tools]
B --> B4[input_data from input-main]
diff --git a/docs/diagrams/agent-context-memory.svg b/docs/diagrams/agent-context-memory.svg
index 6c92b38c..bf86a0f9 100644
--- a/docs/diagrams/agent-context-memory.svg
+++ b/docs/diagrams/agent-context-memory.svg
@@ -1,6 +1,6 @@
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1920 1080" width="100%" role="img" aria-labelledby="title desc">
-<title id="title">Agent Context vs. Simple Memory</title>
-<desc id="desc">Agent Context is the plain stored conversation: one agent_conversations row per (workflow, generation, agent node), loaded at run start and saved after every turn, so each firing continues the same conversation; the Context node on input-context is the opt-in switch and viewing panel. Simple Memory is a separate explicit tool on input-tools: the model or a human remembers, recalls, updates or forgets versioned facts in an isolated namespace, and those items are never injected automatically. Sources: server/nodes/context/__init__.py; server/nodes/context/_descriptor.py; server/services/agent_context/conversation.py; server/services/agent_context/listeners.py; server/nodes/context/_handlers.py; server/nodes/tool/simple_memory/__init__.py; server/services/memory/tool_store.py; client/src/components/parameterPanel/ContextPanel.tsx; client/src/components/parameterPanel/MemoryToolPanel.tsx</desc>
+<title id="title">Agent Context vs. Memory</title>
+<desc id="desc">Agent Context is the plain stored conversation: one agent_conversations row per (workflow, generation, agent node), loaded at run start and saved after every turn, so each firing continues the same conversation; the Context node on input-context is the opt-in switch and viewing panel. Memory is a separate explicit tool on input-tools: the model or a human remembers, recalls, updates or forgets versioned facts in an isolated namespace, and those items are never injected automatically. Sources: server/nodes/context/__init__.py; server/nodes/context/_descriptor.py; server/services/agent_context/conversation.py; server/services/agent_context/listeners.py; server/nodes/context/_handlers.py; server/nodes/tool/simple_memory/__init__.py; server/services/memory/tool_store.py; client/src/components/parameterPanel/ContextPanel.tsx; client/src/components/parameterPanel/MemoryToolPanel.tsx</desc>
<defs>
<marker id="arrow" viewBox="0 0 10 10" refX="9" refY="5" markerWidth="8" markerHeight="8" orient="auto"><path d="M0,0 L10,5 L0,10 z" fill="#e8eaed"/></marker>
<style>
@@ -18,8 +18,8 @@
</style>
</defs>
<rect width="1920" height="1080" fill="#0d0f13"/>
-<text x="72" y="58" class="title">Agent Context vs. Simple Memory</text>
+<text x="72" y="58" class="title">Agent Context vs. Memory</text>
<text x="72" y="91" class="subtitle">Stored-conversation continuity and explicit durable knowledge are separate systems</text>
-<g data-group-id="context-plane"><rect x="72" y="140" width="720" height="800" rx="18" fill="#15171c" stroke="#bd93f9" stroke-width="2"/><text x="92" y="172" class="group-label">AGENT CONTEXT · STORED CONVERSATION</text></g><g data-group-id="memory-plane"><rect x="1128" y="140" width="720" height="800" rx="18" fill="#15171c" stroke="#50fa7b" stroke-width="2"/><text x="1148" y="172" class="group-label">SIMPLE MEMORY · EXPLICIT TOOL</text></g><g data-relationship-id="context-binds-runtime"><title>opt in at input-context</title><path d="M 400 300.0 L 430 430 L 790 430 L 810 520.0" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="481.6" y="415.0" width="236.79999999999998" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="600" y="433" class="edge-label" text-anchor="middle">opt in at input-context</text></g><g data-relationship-id="runtime-appends-context"><title>save messages after each turn</title><path d="M 810 520.0 L 650 520 L 650 603 L 420 602.5" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="524.0" y="505.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="650" y="523" class="edge-label" text-anchor="middle">save messages after each turn</text></g><g data-relationship-id="journal-compacts"><title>compact at COMPACTION_RATIO</title><path d="M 270.0 695 L 270 725 L 590 725 L 590.0 760" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="315.4" y="710.0" width="229.2" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="430" y="728" class="edge-label" text-anchor="middle">compact at COMPACTION_RATIO</text></g><g data-relationship-id="checkpoint-replays"><title>summary + current request seed</title><path d="M 740 835.0 L 780 835 L 780 700 L 960 700 L 960.0 645" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="694.0" y="705.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="820" y="723" class="edge-label" text-anchor="middle">summary + current request seed</text></g><g data-relationship-id="memory-registers-tool"><title>register tool at input-tools</title><path d="M 1520 300.0 L 1490 430 L 1130 430 L 1110 520.0" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1201.6" y="415.0" width="236.79999999999998" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1320" y="433" class="edge-label" text-anchor="middle" textLength="218.8" lengthAdjust="spacingAndGlyphs">register tool at input-tools</text></g><g data-relationship-id="runtime-invokes-memory"><title>remember / recall / update / forget</title><path d="M 1110 520.0 L 1290 520 L 1290 603 L 1500 602.5" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1145.0" y="505.0" width="290.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1290" y="523" class="edge-label" text-anchor="middle" textLength="272.0" lengthAdjust="spacingAndGlyphs">remember / recall / update / forget</text></g><g data-relationship-id="memory-persists-items"><title>persist isolated versioned items</title><path d="M 1650.0 695 L 1650 725 L 1330 725 L 1330.0 760" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1356.4" y="710.0" width="267.2" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1490" y="728" class="edge-label" text-anchor="middle" textLength="249.2" lengthAdjust="spacingAndGlyphs">persist isolated versioned items</text></g><g data-relationship-id="memory-returns-items"><title>return selected recalled items</title><path d="M 1180 835.0 L 1140 835 L 1140 700 L 960 700 L 960.0 645" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="974.0" y="665.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1100" y="683" class="edge-label" text-anchor="middle" textLength="234.0" lengthAdjust="spacingAndGlyphs">return selected recalled items</text></g><g data-component-id="context-node"><title>Context Node</title><rect x="120" y="220" width="280" height="160" rx="12" fill="#1b1e25" stroke="#bd93f9" stroke-width="2"/><rect x="120" y="220" width="280" height="7" rx="4" fill="#bd93f9"/><text x="138" y="246" class="type-label">OPT-IN SWITCH + PANEL</text><text x="138" y="272" class="node-label">Context Node</text><text x="138" y="298" class="node-detail" textLength="244.0" lengthAdjust="spacingAndGlyphs">output-context → input-context</text><text x="138" y="318" class="node-detail">view, clear, reset; no params</text></g><g data-component-id="context-journal"><title>agent_conversations</title><rect x="120" y="510" width="300" height="185" rx="12" fill="#1b1e25" stroke="#ff79c6" stroke-width="2"/><rect x="120" y="510" width="300" height="7" rx="4" fill="#ff79c6"/><text x="138" y="536" class="type-label">CONVERSATION STORE</text><text x="138" y="562" class="node-label">agent_conversations</text><text x="138" y="588" class="node-detail">one row per generation + agent</text><text x="138" y="608" class="node-detail">messages JSON, ts per message</text><text x="138" y="628" class="node-detail">loud loads, best-effort saves</text></g><g data-component-id="context-checkpoint"><title>Compaction summary</title><rect x="440" y="760" width="300" height="150" rx="12" fill="#1b1e25" stroke="#ffb86c" stroke-width="2"/><rect x="440" y="760" width="300" height="7" rx="4" fill="#ffb86c"/><text x="458" y="786" class="type-label">CONTEXT STATE</text><text x="458" y="812" class="node-label">Compaction summary</text><text x="458" y="838" class="node-detail">system prompt kept verbatim</text><text x="458" y="858" class="node-detail">summary rides a user turn</text></g><g data-component-id="agent-runtime"><title>Agent Runtime</title><rect x="810" y="395" width="300" height="250" rx="12" fill="#1b1e25" stroke="#bd93f9" stroke-width="2"/><rect x="810" y="395" width="300" height="7" rx="4" fill="#bd93f9"/><text x="828" y="421" class="type-label">SHARED CONSUMER</text><text x="828" y="447" class="node-label">Agent Runtime</text><text x="828" y="473" class="node-detail">sends the stored conversation</text><text x="828" y="493" class="node-detail">invokes Memory explicitly</text><text x="828" y="513" class="node-detail">keeps both namespaces separate</text></g><g data-component-id="memory-node"><title>Simple Memory</title><rect x="1520" y="220" width="280" height="160" rx="12" fill="#1b1e25" stroke="#50fa7b" stroke-width="2"/><rect x="1520" y="220" width="280" height="7" rx="4" fill="#50fa7b"/><text x="1538" y="246" class="type-label">LOCKED TOOL</text><text x="1538" y="272" class="node-label">Simple Memory</text><text x="1538" y="298" class="node-detail">output-tool → input-tools</text><text x="1538" y="318" class="node-detail">remember · recall · list</text><text x="1538" y="338" class="node-detail">get · update · forget</text></g><g data-component-id="memory-store"><title>MemoryToolStore</title><rect x="1500" y="510" width="300" height="185" rx="12" fill="#1b1e25" stroke="#ff79c6" stroke-width="2"/><rect x="1500" y="510" width="300" height="7" rx="4" fill="#ff79c6"/><text x="1518" y="536" class="type-label">DURABLE DATA</text><text x="1518" y="562" class="node-label">MemoryToolStore</text><text x="1518" y="588" class="node-detail">owner / workflow / node scope</text><text x="1518" y="608" class="node-detail">versioned idempotent mutations</text><text x="1518" y="628" class="node-detail">lexical + optional embedding</text></g><g data-component-id="memory-items"><title>Facts &amp; Decisions</title><rect x="1180" y="760" width="300" height="150" rx="12" fill="#1b1e25" stroke="#50fa7b" stroke-width="2"/><rect x="1180" y="760" width="300" height="7" rx="4" fill="#50fa7b"/><text x="1198" y="786" class="type-label">MEMORY ITEMS</text><text x="1198" y="812" class="node-label">Facts &amp; Decisions</text><text x="1198" y="838" class="node-detail">returned only when recalled</text><text x="1198" y="858" class="node-detail">never automatic transcript</text></g><g data-callout-id="context-invariant"><rect x="440" y="220" width="300" height="160" rx="10" fill="#1b1e25" stroke="#f1fa8c" stroke-width="2" stroke-dasharray="7 6"/><text x="456" y="250" class="callout-title">Automatic continuity</text><text x="456" y="276" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">System managed; no model tool call</text><text x="456" y="298" class="callout-text">Same conversation across firings</text><text x="456" y="320" class="callout-text">Reset starts a new generation</text></g><g data-callout-id="memory-invariant"><rect x="1180" y="220" width="300" height="160" rx="10" fill="#1b1e25" stroke="#f1fa8c" stroke-width="2" stroke-dasharray="7 6"/><text x="1196" y="250" class="callout-title">Explicit knowledge</text><text x="1196" y="276" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">Model or human chooses mutations</text><text x="1196" y="298" class="callout-text">Stable facts and decisions only</text><text x="1196" y="320" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">Clear never clears Agent Context</text></g>
+<g data-group-id="context-plane"><rect x="72" y="140" width="720" height="800" rx="18" fill="#15171c" stroke="#bd93f9" stroke-width="2"/><text x="92" y="172" class="group-label">AGENT CONTEXT · STORED CONVERSATION</text></g><g data-group-id="memory-plane"><rect x="1128" y="140" width="720" height="800" rx="18" fill="#15171c" stroke="#50fa7b" stroke-width="2"/><text x="1148" y="172" class="group-label">MEMORY · EXPLICIT TOOL</text></g><g data-relationship-id="context-binds-runtime"><title>opt in at input-context</title><path d="M 400 300.0 L 430 430 L 790 430 L 810 520.0" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="481.6" y="415.0" width="236.79999999999998" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="600" y="433" class="edge-label" text-anchor="middle">opt in at input-context</text></g><g data-relationship-id="runtime-appends-context"><title>save messages after each turn</title><path d="M 810 520.0 L 650 520 L 650 603 L 420 602.5" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="524.0" y="505.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="650" y="523" class="edge-label" text-anchor="middle">save messages after each turn</text></g><g data-relationship-id="journal-compacts"><title>compact at COMPACTION_RATIO</title><path d="M 270.0 695 L 270 725 L 590 725 L 590.0 760" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="315.4" y="710.0" width="229.2" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="430" y="728" class="edge-label" text-anchor="middle">compact at COMPACTION_RATIO</text></g><g data-relationship-id="checkpoint-replays"><title>summary + current request seed</title><path d="M 740 835.0 L 780 835 L 780 700 L 960 700 L 960.0 645" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="694.0" y="705.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="820" y="723" class="edge-label" text-anchor="middle">summary + current request seed</text></g><g data-relationship-id="memory-registers-tool"><title>register tool at input-tools</title><path d="M 1520 300.0 L 1490 430 L 1130 430 L 1110 520.0" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1201.6" y="415.0" width="236.79999999999998" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1320" y="433" class="edge-label" text-anchor="middle" textLength="218.8" lengthAdjust="spacingAndGlyphs">register tool at input-tools</text></g><g data-relationship-id="runtime-invokes-memory"><title>remember / recall / update / forget</title><path d="M 1110 520.0 L 1290 520 L 1290 603 L 1500 602.5" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1145.0" y="505.0" width="290.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1290" y="523" class="edge-label" text-anchor="middle" textLength="272.0" lengthAdjust="spacingAndGlyphs">remember / recall / update / forget</text></g><g data-relationship-id="memory-persists-items"><title>persist isolated versioned items</title><path d="M 1650.0 695 L 1650 725 L 1330 725 L 1330.0 760" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="1356.4" y="710.0" width="267.2" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1490" y="728" class="edge-label" text-anchor="middle" textLength="249.2" lengthAdjust="spacingAndGlyphs">persist isolated versioned items</text></g><g data-relationship-id="memory-returns-items"><title>return selected recalled items</title><path d="M 1180 835.0 L 1140 835 L 1140 700 L 960 700 L 960.0 645" fill="none" stroke="#e8eaed" stroke-width="2" marker-end="url(#arrow)"/><rect x="974.0" y="665.0" width="252.0" height="30" rx="7" fill="#15171c" stroke="#2b2f37"/><text x="1100" y="683" class="edge-label" text-anchor="middle" textLength="234.0" lengthAdjust="spacingAndGlyphs">return selected recalled items</text></g><g data-component-id="context-node"><title>Context Node</title><rect x="120" y="220" width="280" height="160" rx="12" fill="#1b1e25" stroke="#bd93f9" stroke-width="2"/><rect x="120" y="220" width="280" height="7" rx="4" fill="#bd93f9"/><text x="138" y="246" class="type-label">OPT-IN SWITCH + PANEL</text><text x="138" y="272" class="node-label">Context Node</text><text x="138" y="298" class="node-detail" textLength="244.0" lengthAdjust="spacingAndGlyphs">output-context → input-context</text><text x="138" y="318" class="node-detail">view, clear, reset; no params</text></g><g data-component-id="context-journal"><title>agent_conversations</title><rect x="120" y="510" width="300" height="185" rx="12" fill="#1b1e25" stroke="#ff79c6" stroke-width="2"/><rect x="120" y="510" width="300" height="7" rx="4" fill="#ff79c6"/><text x="138" y="536" class="type-label">CONVERSATION STORE</text><text x="138" y="562" class="node-label">agent_conversations</text><text x="138" y="588" class="node-detail">one row per generation + agent</text><text x="138" y="608" class="node-detail">messages JSON, ts per message</text><text x="138" y="628" class="node-detail">loud loads, best-effort saves</text></g><g data-component-id="context-checkpoint"><title>Compaction summary</title><rect x="440" y="760" width="300" height="150" rx="12" fill="#1b1e25" stroke="#ffb86c" stroke-width="2"/><rect x="440" y="760" width="300" height="7" rx="4" fill="#ffb86c"/><text x="458" y="786" class="type-label">CONTEXT STATE</text><text x="458" y="812" class="node-label">Compaction summary</text><text x="458" y="838" class="node-detail">system prompt kept verbatim</text><text x="458" y="858" class="node-detail">summary rides a user turn</text></g><g data-component-id="agent-runtime"><title>Agent Runtime</title><rect x="810" y="395" width="300" height="250" rx="12" fill="#1b1e25" stroke="#bd93f9" stroke-width="2"/><rect x="810" y="395" width="300" height="7" rx="4" fill="#bd93f9"/><text x="828" y="421" class="type-label">SHARED CONSUMER</text><text x="828" y="447" class="node-label">Agent Runtime</text><text x="828" y="473" class="node-detail">sends the stored conversation</text><text x="828" y="493" class="node-detail">invokes Memory explicitly</text><text x="828" y="513" class="node-detail">keeps both namespaces separate</text></g><g data-component-id="memory-node"><title>Memory</title><rect x="1520" y="220" width="280" height="160" rx="12" fill="#1b1e25" stroke="#50fa7b" stroke-width="2"/><rect x="1520" y="220" width="280" height="7" rx="4" fill="#50fa7b"/><text x="1538" y="246" class="type-label">LOCKED TOOL</text><text x="1538" y="272" class="node-label">Memory</text><text x="1538" y="298" class="node-detail">output-tool → input-tools</text><text x="1538" y="318" class="node-detail">remember · recall · list</text><text x="1538" y="338" class="node-detail">get · update · forget</text></g><g data-component-id="memory-store"><title>MemoryToolStore</title><rect x="1500" y="510" width="300" height="185" rx="12" fill="#1b1e25" stroke="#ff79c6" stroke-width="2"/><rect x="1500" y="510" width="300" height="7" rx="4" fill="#ff79c6"/><text x="1518" y="536" class="type-label">DURABLE DATA</text><text x="1518" y="562" class="node-label">MemoryToolStore</text><text x="1518" y="588" class="node-detail">owner / workflow / node scope</text><text x="1518" y="608" class="node-detail">versioned idempotent mutations</text><text x="1518" y="628" class="node-detail">lexical + optional embedding</text></g><g data-component-id="memory-items"><title>Facts &amp; Decisions</title><rect x="1180" y="760" width="300" height="150" rx="12" fill="#1b1e25" stroke="#50fa7b" stroke-width="2"/><rect x="1180" y="760" width="300" height="7" rx="4" fill="#50fa7b"/><text x="1198" y="786" class="type-label">MEMORY ITEMS</text><text x="1198" y="812" class="node-label">Facts &amp; Decisions</text><text x="1198" y="838" class="node-detail">returned only when recalled</text><text x="1198" y="858" class="node-detail">never automatic transcript</text></g><g data-callout-id="context-invariant"><rect x="440" y="220" width="300" height="160" rx="10" fill="#1b1e25" stroke="#f1fa8c" stroke-width="2" stroke-dasharray="7 6"/><text x="456" y="250" class="callout-title">Automatic continuity</text><text x="456" y="276" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">System managed; no model tool call</text><text x="456" y="298" class="callout-text">Same conversation across firings</text><text x="456" y="320" class="callout-text">Reset starts a new generation</text></g><g data-callout-id="memory-invariant"><rect x="1180" y="220" width="300" height="160" rx="10" fill="#1b1e25" stroke="#f1fa8c" stroke-width="2" stroke-dasharray="7 6"/><text x="1196" y="250" class="callout-title">Explicit knowledge</text><text x="1196" y="276" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">Model or human chooses mutations</text><text x="1196" y="298" class="callout-text">Stable facts and decisions only</text><text x="1196" y="320" class="callout-text" textLength="268.0" lengthAdjust="spacingAndGlyphs">Clear never clears Agent Context</text></g>
<g aria-label="Legend"><circle cx="72" cy="1038" r="7" fill="#bd93f9"/><text x="86" y="1044" class="legend">Agents / system</text><circle cx="317" cy="1038" r="7" fill="#8be9fd"/><text x="331" y="1044" class="legend">Models / interfaces</text><circle cx="562" cy="1038" r="7" fill="#50fa7b"/><text x="576" y="1044" class="legend">Tools / results</text><circle cx="807" cy="1038" r="7" fill="#ff79c6"/><text x="821" y="1044" class="legend">Triggers / events</text><circle cx="1052" cy="1038" r="7" fill="#ffb86c"/><text x="1066" y="1044" class="legend">Workflow / runtime</text><circle cx="1297" cy="1038" r="7" fill="#f1fa8c"/><text x="1311" y="1044" class="legend">Security note</text></g>
</svg>
diff --git a/docs/diagrams/default-workflows.svg b/docs/diagrams/default-workflows.svg
index 9553bdd0..fde3f0e9 100644
--- a/docs/diagrams/default-workflows.svg
+++ b/docs/diagrams/default-workflows.svg
@@ -58,7 +58,7 @@
<!-- toolbox cluster -->
<path class="conn-dot" d="M 480 255 L 480 305"/>
<rect class="box p-purple" x="380" y="305" width="560" height="80" rx="8"/>
- <text x="660" y="329" text-anchor="middle" class="h2 accent-purple">Toolbox (11 tools)</text>
+ <text x="660" y="329" text-anchor="middle" class="h2 accent-purple">Toolbox (10+ tools)</text>
<text x="660" y="351" text-anchor="middle" class="small">web search &#183; Perplexity &#183; browser &#183; file read/write &#183; fs search</text>
<text x="660" y="369" text-anchor="middle" class="small">shell &#183; Python &#183; processes &#183; todos &#183; Vercel</text>
diff --git a/docs/diagrams/execution-flow.svg b/docs/diagrams/execution-flow.svg
index a1fe4c78..d684f602 100644
--- a/docs/diagrams/execution-flow.svg
+++ b/docs/diagrams/execution-flow.svg
@@ -43,7 +43,7 @@
<rect class="panel p-purple" x="780" y="100" width="420" height="96" rx="8"/>
<text x="990" y="130" class="h1" text-anchor="middle">Toolbar "Start"</text>
<text x="990" y="156" class="sub" text-anchor="middle">deploy_workflow WS msg -&gt; DeploymentManager.deploy</text>
- <text x="990" y="178" class="sub" text-anchor="middle">graph snapshot persisted &#183; button toggles to Stop</text>
+ <text x="990" y="178" class="sub" text-anchor="middle">graph snapshot persisted &#183; button becomes Pause</text>
<!-- Canvas lane: straight down to convergence -->
<path class="conn" d="M 250 196 L 250 1068" marker-end="url(#arrow)"/>
diff --git a/docs/diagrams/how-it-works.svg b/docs/diagrams/how-it-works.svg
index 721f44e6..866ca6db 100644
--- a/docs/diagrams/how-it-works.svg
+++ b/docs/diagrams/how-it-works.svg
@@ -74,7 +74,7 @@
<text class="h2" x="1262" y="172" text-anchor="middle">Test it, then start it</text>
<text class="sub" x="1262" y="196" text-anchor="middle">Run tests one node in place.</text>
<text class="sub" x="1262" y="214" text-anchor="middle">Start deploys the whole workflow</text>
- <text class="sub" x="1262" y="232" text-anchor="middle">(the button flips to Stop)</text>
+ <text class="sub" x="1262" y="232" text-anchor="middle">(the button becomes Pause)</text>
<text class="label" x="1262" y="272" text-anchor="middle">runs on a schedule, on an event, or in real time</text>
<!-- Arrow 4 -> 5 -->
diff --git a/docs/diagrams/system-overview.svg b/docs/diagrams/system-overview.svg
index 6712f976..67705c90 100644
--- a/docs/diagrams/system-overview.svg
+++ b/docs/diagrams/system-overview.svg
@@ -73,14 +73,14 @@
<text class="small" x="952" y="428" text-anchor="middle">DeploymentManager + WorkflowControlWorkflow (Start/Pause/Resume/Reset)</text>
<rect class="box p-purple" x="250" y="458" width="435" height="86" rx="8"/>
- <text class="h2" x="467" y="483" text-anchor="middle">Node plugins (140+, 34 palette groups)</text>
+ <text class="h2" x="467" y="483" text-anchor="middle">Node plugins (140+, 30+ palette groups)</text>
<text class="small" x="467" y="503" text-anchor="middle">server/nodes/&lt;group&gt;/&lt;plugin&gt;/ - one folder per node</text>
<text class="small" x="467" y="519" text-anchor="middle">Pydantic Params / Output = SSOT, served as NodeSpec</text>
- <text class="small" x="467" y="535" text-anchor="middle">78 skills in server/skills/ (SKILL.md playbooks)</text>
+ <text class="small" x="467" y="535" text-anchor="middle">80+ skills in server/skills/ (SKILL.md playbooks)</text>
<rect class="box p-purple" x="735" y="458" width="435" height="86" rx="8"/>
<text class="h2" x="952" y="483" text-anchor="middle">AuthService + credentials</text>
- <text class="small" x="952" y="503" text-anchor="middle">35 Credential classes (API keys, OAuth, CLI-managed logins)</text>
+ <text class="small" x="952" y="503" text-anchor="middle">35+ Credential classes (API keys, OAuth, CLI-managed logins)</text>
<text class="small" x="952" y="519" text-anchor="middle">Fernet + PBKDF2 encrypted credentials.db</text>
<text class="small" x="952" y="535" text-anchor="middle">optional JWT login (VITE_AUTH_ENABLED, AUTH_MODE single / multi)</text>