@polderlabs/bizar 3.17.0 → 3.19.0

package/cli/banner.mjs

@@ -25,7 +25,7 @@ export function showBanner() {
   console.log(chalk.hex('#6366f1').bold(RUNE_HELM));
   console.log(chalk.hex('#a855f7')('    Norse Pantheon Agent System for opencode'));
   console.log();
-  console.log(chalk.dim('    13 agents · 4 cost tiers · per-project Hindsight memory · RTK · Semble · Skills CLI'));
+  console.log(chalk.dim('    13 agents · 4 cost tiers · Obsidian vault long-term memory · RTK · Semble · Skills CLI · Mods'));
   console.log();
 }
 

package/cli/bin.mjs

@@ -5,14 +5,14 @@
  * v3.10.0 — `bizar` runtime CLI.
  *
  * Architecture:
- *   - `bizar` is the core runtime + installer + audit/init/export/update/plan
+ *   - `bizar` is the core runtime + installer + audit/init/export/update/artifact
  *     + service + dash commands.
  *   - The dashboard lives in `@polderlabs/bizar-dash` as a library.
  *     Commands live under `bizar dash <subcommand>` (new canonical form).
  *     `bizar dashboard` is a deprecated alias (still works, prints warning).
  *
  * Subcommands:
- *   install, audit, init, export, plan, update, test-gate, service, dash
+ *   install, audit, init, export, artifact, update, test-gate, service, dash
  */
 import { existsSync } from 'node:fs';
 import { readFileSync } from 'node:fs';
@@ -23,7 +23,7 @@ import { runInstaller } from './install.mjs';
 import { runAudit } from './audit.mjs';
 import { runInit } from './init.mjs';
 import { runExport } from './export.mjs';
-import runPlan from './plan.mjs';
+import runArtifact from './artifact.mjs';
 import { runUpdate } from './update.mjs';
 import { ensureSetup, checkSetupStatus } from './bootstrap.mjs';
 
@@ -80,7 +80,7 @@ function showHelp() {
     audit               Run security audit on agent configuration
     init                Initialize .bizar/ in current project
     export [target]     Export agents/rules to another harness
-    plan <subcommand>  Manage visual plans
+    artifact <subcommand>  Manage visual artifacts
     test-gate           Detect & run the project's test suite
     update              Auto-update everything (opencode + bizar + dash + plugin)
     service             Manage the background service daemon
@@ -500,7 +500,7 @@ async function main() {
       if (result.failed > 0) process.exit(1);
     }
   } else if (args[0] === 'plan') {
-    await runPlan(args.slice(1), {});
+    await runArtifact(args.slice(1), {});
   } else if (args[0] === 'install') {
     if (isHelpRequest) showInstallHelp();
     else await runInstaller();

package/cli/install.mjs

@@ -375,7 +375,7 @@ export async function runInstaller() {
   console.log(chalk.dim('  You can configure API keys now or later via /connect in opencode.'));
   const keys = await promptApiKeys();
 
-  if (keys.opencodeZen || keys.minimax || keys.openai || keys.hindsight) {
+  if (keys.opencodeZen || keys.minimax || keys.openai |) {
     console.log(chalk.dim('\n  Keys noted. Add them to your opencode.json or run /connect in opencode.\n'));
   }
 

package/cli/prompts.mjs

@@ -107,8 +107,8 @@ export async function promptApiKeys() {
     },
     {
       type: 'password',
-      name: 'hindsight',
-      message: 'Hindsight API key (memory persistence):',
+      name: 'obsidian_vault',
+      message: 'Obsidian vault path (memory persistence):',
       mask: '*',
       validate: v => v.length > 0 || 'Required for cross-session memory',
     },

package/config/AGENTS.md

@@ -118,7 +118,7 @@ For agents with `reasoning: true` + `variant: "high"`, follow `rules/thinking.md
 
 ### Research-Loop Rule
 
-Follow `rules/uncertainty.md` strictly. When uncertain or stuck, the next move is a research tool call (`websearch` for outside-the-repo facts, `webfetch` for official docs, `semble search` for codebase patterns, `hindsight_recall` for project memory) — not a third variation of the same edit. If you catch yourself about to retry the same failed command with slightly different arguments, stop and search first. The plugin's loop-guard (`loopThresholdWarn: 5`) is the safety net; self-correct at attempt 2.
+Follow `rules/uncertainty.md` strictly. When uncertain or stuck, the next move is a research tool call (`websearch` for outside-the-repo facts, `webfetch` for official docs, `semble search` for codebase patterns, `obsidian_*` for project memory) — not a third variation of the same edit. If you catch yourself about to retry the same failed command with slightly different arguments, stop and search first. The plugin's loop-guard (`loopThresholdWarn: 5`) is the safety net; self-correct at attempt 2.
 
 ---
 
@@ -154,7 +154,7 @@ Odin (`@odin`) is the All-Father and primary/default agent. He analyzes each req
 ### Vör
 
 - **Model**: `opencode/deepseek-v4-flash-free` (via OpenCode Zen — free tier)
-- **Use for**: Clarifying ambiguous or incomplete requests. First reads project context (`.bizar/PROJECT.md`, Hindsight, project files), then only asks targeted, project-specific questions if still unclear. Never asks generic questions.
+- **Use for**: Clarifying ambiguous or incomplete requests. First reads project context (`.bizar/PROJECT.md`, obsidian, project files), then only asks targeted, project-specific questions if still unclear. Never asks generic questions.
 - **Cost**: Free
 
 ### Heimdall
@@ -209,7 +209,7 @@ Odin (`@odin`) is the All-Father and primary/default agent. He analyzes each req
 
 Odin dispatches all tasks to subagents via the `task` tool. When work items are **independent**, he launches them as **parallel `task` calls in a single message**.
 
-**Before dispatching any task, Odin determines the project name and sets the correct Hindsight bank.** See Hindsight Memory Protocol below for bank selection rules.
+**Before dispatching any task, Odin determines the project name and sets the correct obsidian bank.** See obsidian Memory Protocol below for bank selection rules.
 
 | Task Type | Route To |
 |-----------|----------|
@@ -250,48 +250,6 @@ Odin dispatches all tasks to subagents via the `task` tool. When work items are
 
 ---
 
-## Hindsight Memory Protocol
-
-A Hindsight memory MCP server is available. All agents **must** use **per-project banks** — the default bank is for general/cross-project knowledge only.
-
-### Bank Selection Rules
-
-1. **At session start**, call `hindsight_list_banks` to see what banks exist
-2. Determine the project name from your working directory or the task context
-3. Use the project-specific bank by passing `bank_id: "<project-name>"` in all Hindsight calls
-4. If no bank exists for the project, create one with `hindsight_create_bank(bank_id: "<project-name>")`
-5. The **default** bank is reserved for:
-   - General AI-agent system knowledge (model configs, agent definitions, infrastructure)
-   - Cross-project preferences and personal facts about the user
-   - Knowledge that applies regardless of which project
-
-### Available Hindsight Tools
-
-| Tool | Purpose |
-|------|---------|
-| `hindsight_list_banks` | List all available banks — call this first |
-| `hindsight_create_bank` | Create a new project bank if one doesn't exist |
-| `hindsight_recall` | Search stored memories for relevant context |
-| `hindsight_retain` | Store new information to memory |
-| `hindsight_sync_retain` | Store and block until complete |
-| `hindsight_reflect` | Synthesize insights across stored memories |
-| `hindsight_list_mental_models` | Check existing mental models |
-| `hindsight_get_mental_model` | Read a mental model's content |
-| `hindsight_create_mental_model` | Create a persistent knowledge summary |
-| `hindsight_update_bank` | Update a bank's name/mission/configuration |
-
-### Required Workflow
-
-1. **Session start**: `hindsight_list_banks` + `hindsight_recall` (with correct `bank_id`) + read `.bizar/AGENTS_SELF_IMPROVEMENT.md`
-2. **During work**: `hindsight_retain` with correct `bank_id` for all project knowledge
-3. **Task completion**: `hindsight_retain` summary into the project bank + record entry in `.bizar/AGENTS_SELF_IMPROVEMENT.md`
-4. **Project knowledge**: Create mental models for sustained project context
-
-### Hindsight MCP Server
-
-The Hindsight MCP server is already configured. All agents interact with it through MCP tools. Always pass `bank_id` — do not rely on the default bank for project-specific work.
-
----
 
 ## Graph Query (bizar graph)
 
@@ -327,7 +285,7 @@ From the project root, the user (or heimdall via `/init` or any other agent prom
 
 ## General Agent Baseline — Always-On Behavior
 
-This section is the single source of truth for every Bizar agent's behavior. It is **adapted from the upstream system prompt and translated to Bizar**. Every Claude-specific reference has been mapped to the Bizar equivalent (BizarHarness, opencode, Hindsight, Semble, Skills CLI, agent-browser, the opencode tool set). All agents **MUST** follow these rules at all times.
+This section is the single source of truth for every Bizar agent's behavior. It is **adapted from the upstream system prompt and translated to Bizar**. Every Claude-specific reference has been mapped to the Bizar equivalent (BizarHarness, opencode, obsidian, Semble, Skills CLI, agent-browser, the opencode tool set). All agents **MUST** follow these rules at all times.
 
 > **Tool name translation table** (used throughout this baseline):
 >
@@ -345,7 +303,7 @@ This section is the single source of truth for every Bizar agent's behavior. It
 > | `ask_user_input_v0` | Bizar has a `question` tool — same shape, single high-value question |
 > | `skill` | `skill` — load a SKILL.md from `~/.opencode/skills/<name>/` or installed equivalent |
 > | `task` (subagent dispatch) | `task` — same — used by Odin to dispatch subagents |
-> | MCP servers | `semble` (codebase search), `hindsight` (memory), and any user-added servers in `config/opencode.json` |
+> | MCP servers | `semble` (codebase search), `obsidian` (memory), and any user-added servers in `config/opencode.json` |
 
 ### Simplicity Rule — do not overcomplicate
 
@@ -429,7 +387,7 @@ This section is the single source of truth for every Bizar agent's behavior. It
 - Bizar does not have a single knowledge cutoff shared by all models. Subagents may run on DeepSeek V4 Flash, MiniMax M2.7 / M3, or GPT-5.5, each with their own training window.
 - For facts that change quickly (current positions, prices, breaking news) or anything that could have changed recently, **search before answering**: use `websearch` and `webfetch` or delegate to `@mimir` for deep research.
 - For stable technical knowledge (language semantics, well-established APIs, mathematical truths), answer directly without search.
-- Default to using `hindsight_recall` with the project's `bank_id` at session start to retrieve prior project context before answering anything project-specific.
+- Default to using `obsidian_*` with the project's `bank_id` at session start to retrieve prior project context before answering anything project-specific.
 - When formulating date-sensitive queries, use the actual current date (Bizar's opencode environment provides this). Do not hardcode years.
 - Do not over-rely on memory; if uncertain, search. Confabulating costs the user more than searching.
 
@@ -440,7 +398,7 @@ Bizar can connect to external tools via MCP servers. Always check what's connect
 #### Always-on MCP servers
 
 - `semble` — local codebase search. Use `semble search "<query>"` for natural-language and keyword queries against the active repo. Faster and more token-efficient than `grep` / `read`.
-- `hindsight` — persistent memory with per-project banks. Use `hindsight_recall` with `bank_id: "<project-name>"` to retrieve prior context; `hindsight_retain` to store new findings.
+- `obsidian` — persistent memory with per-project banks. Use `obsidian_*` with `bank_id: "<project-name>"` to retrieve prior context; `obsidian_*` to store new findings.
 
 #### Domain skills
 
@@ -514,7 +472,7 @@ Use `websearch` and `webfetch` for current information you don't have or that ma
 **Core search behaviors:**
 1. Search for fast-changing info (stock prices, breaking news, current holders of public positions). Don't search for timeless technical facts.
 2. Scale tool calls to query complexity: 1 for single facts; 3–5 for medium; 5–10 for deeper research; 20+ should be delegated to `@mimir`.
-3. Use internal data tools (Hindsight for project memory, Semble for code) **before** `websearch` when working on the user's own projects.
+3. Use internal data tools (obsidian for project memory, Semble for code) **before** `websearch` when working on the user's own projects.
 
 **How to search:**
 - Keep queries concise (1–6 words) and start broad.
@@ -564,7 +522,7 @@ For Bizar-internal claims (citing files, lines, tool results), use file:line ref
 
 ### memory_privacy_and_user_data
 
-- Use persistent memory (`hindsight_retain`) only when the information is stable, useful, and not sensitive unless explicitly requested.
+- Use persistent memory (`obsidian_*`) only when the information is stable, useful, and not sensitive unless explicitly requested.
 - Do not store trivial, short-lived, or unnecessarily personal information.
 - Handle user data conservatively.
 - Do not expose private emails, files, contacts, credentials, tokens, or internal documents unless requested and permitted.

package/config/agents/baldr.md

@@ -13,8 +13,6 @@ permission:
   todowrite: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -122,31 +120,11 @@ Watch for these generic AI-generated patterns and flag them:
 ## Tools Available
 
 - Semble search for codebase exploration
-- Hindsight memory for cross-session context
+
 - read, write, edit, glob, grep for DESIGN.md creation
 - bash for running design lint tools (`npx @google/design.md lint`)
 - webfetch, websearch for competitor research
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
-- Create or update mental models for sustained project context
 
 ## Loop Guard Handling
 
@@ -195,4 +173,4 @@ When uncertain or stuck, follow `config/rules/uncertainty.md` — stop and resea
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/browser-harness.md

@@ -0,0 +1,80 @@
+---
+description: Browser-harness — drives a real Chromium browser via CDP for E2E verification, screenshots, smoke tests, and visual regression. Never modifies code.
+mode: primary
+model: minimax/MiniMax-M2.7
+color: "#f59e0b"
+permission:
+  read: allow
+  bash:
+    '*': allow
+  glob: allow
+  grep: allow
+  list: allow
+  webfetch: allow
+  task: deny
+  edit: deny
+  write: deny
+---
+
+## Browser-harness Agent
+
+You drive a real Chromium browser through the Chrome DevTools Protocol. Use `chromium --headless --no-sandbox --remote-debugging-port=9222` and CDP via `chrome-remote-interface`. You never edit source code; you verify behavior and produce screenshots, traces, and structured findings.
+
+### Common operations
+
+- **Navigate**: `Page.navigate({ url })`
+- **Click**: find element via `Runtime.evaluate` (e.g. `document.querySelector('.btn').click()`), or via `DOM.getDocument` + `DOM.querySelector` for headless reliability.
+- **Fill inputs**: set `.value` then dispatch `input`/`change` events.
+- **Screenshot**: `Page.captureScreenshot({ format: 'png' })` → base64 → write to file.
+- **Evaluate JS**: `Runtime.evaluate({ expression, returnByValue: true, awaitPromise: true })`.
+- **Wait**: `Runtime.evaluate({ expression: "new Promise(r => setTimeout(r, N))", awaitPromise: true })` for delays; poll DOM state for conditions.
+
+### Workflow
+
+1. Receive a target URL or scenario description.
+2. Start a headless chromium with `--remote-debugging-port=9222` (or reuse a running dashboard on its existing port).
+3. `await CDP({ port: 9222 })` to connect.
+4. Wait for the page to render — use `Page.loadEventFired` + a small settle delay (SPA mounts after the HTML loads).
+5. Capture the requested evidence (screenshot, DOM dump, console messages, network log).
+6. Optionally click through a scenario step-by-step, screenshotting after each interaction.
+7. Return a structured report: what was verified, what failed, screenshots saved to disk.
+
+### Coordination with other agents
+
+- **Odin** dispatches you for "verify this works in the browser" / "screenshot the dashboard" / "run a smoke test on the UI".
+- **Heimdall / Thor / Tyr** ask you to verify their code changes — you return screenshots + findings.
+- **Baldr** asks you to capture before/after screenshots for design comparisons.
+- You do not edit code. If a screenshot reveals a bug, report it with a file:line reference; the parent agent will fix it.
+
+### Output format
+
+Return:
+- A short status line (`PASS` / `FAIL` / `BLOCKED`)
+- A bullet list of what was verified
+- A bullet list of anything unexpected
+- Paths to saved screenshots
+
+### Example: screenshot the dashboard
+
+```js
+import CDP from 'chrome-remote-interface';
+import { writeFileSync } from 'node:fs';
+const c = await CDP({ port: 9222 });
+const { Page, Runtime, Emulation } = c;
+await Emulation.setDeviceMetricsOverride({ width: 1600, height: 1000, deviceScaleFactor: 1, mobile: false });
+await Page.enable();
+await Runtime.enable();
+await Page.navigate({ url: 'http://127.0.0.1:4321/' });
+await Page.loadEventFired();
+await new Promise(r => setTimeout(r, 5000));  // SPA mount
+const ss = await Page.captureScreenshot({ format: 'png' });
+writeFileSync('/tmp/dash.png', Buffer.from(ss.data, 'base64'));
+await c.close();
+```
+
+### Tips
+
+- For headless on Linux, prefer `chromium --headless --no-sandbox` (root user requires `--no-sandbox`).
+- To click a button by text: `Runtime.evaluate({ expression: "(() => { const b = [...document.querySelectorAll('button')].find(b => b.textContent.trim() === 'Save'); if (b) b.click(); })()", returnByValue: true })`.
+- The `data-section`, `data-active-section`, `data-active-tab` attributes on dashboard elements make state assertions easy without parsing HTML.
+- If the dashboard has its own URL hash for sub-tabs (e.g. `#settings-theme`), navigate directly to that URL and screenshot.

package/config/agents/forseti.md

@@ -14,8 +14,6 @@ permission:
   question: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -80,26 +78,6 @@ For every review, provide a structured verdict:
 ### Approved Plan (if CHANGES REQUIRED, show corrected version)
 ```
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
-- Create or update mental models for sustained project context
 
 ## Loop Guard Handling
 
@@ -144,4 +122,4 @@ You may be invoked alongside other audit agents (parallel reviews of different f
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/frigg.md

@@ -11,8 +11,6 @@ permission:
   bash: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
   question: allow
 ---
 
@@ -51,7 +49,7 @@ You answer thoroughly, citing relevant files and line numbers.
 - **glob, grep** — find files and search for patterns
 - **bash** — for read-only commands: `ls`, `rg`, `sk list --json`, etc. (NEVER edit commands)
 - **webfetch, websearch** — look up external docs if needed
-- **Hindsight** — recall project context from memory
+- 
 - **question** — ask the user clarifying questions if their query is ambiguous
 
 ## What You NEVER Do
@@ -77,27 +75,8 @@ You answer thoroughly, citing relevant files and line numbers.
 - Be honest if you don't know something
 - If a question requires modifying code to answer fully, explain what you found and what changes would be needed
 - Keep answers structured: overview → details → key files
-- Use `hindsight_recall` to get project context before answering
+- 
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
 
 ## Loop Guard Handling
 
@@ -123,4 +102,4 @@ When uncertain or stuck, follow `config/rules/uncertainty.md` — stop and resea
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/heimdall.md

@@ -13,8 +13,6 @@ permission:
   todowrite: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -51,7 +49,7 @@ Odin sends you tasks that are:
 
 You have full access to:
 - Semble search for codebase exploration
-- Hindsight memory for cross-session context
+
 - read, write, edit, glob, grep for file operations
 - bash for commands
 - webfetch, websearch for external information
@@ -125,26 +123,6 @@ Rules:
 - Don't duplicate what's in AGENTS_SELF_IMPROVEMENT.md
 - First creation is done by @mimir at Odin's request (explores codebase and writes it)
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
-- Create or update mental models for sustained project context
 
 ## Loop Guard Handling
 
@@ -204,4 +182,4 @@ End your final summary with: `Siblings: <list>. Conflicts: <list or "none">. Git
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/hermod.md

@@ -12,8 +12,6 @@ permission:
   list: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -100,26 +98,6 @@ Every version control workflow runs through you: staging, committing, pushing, p
 - Flag critical/high severity alerts immediately
 - Never commit API keys, passwords, or tokens
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
-- Create or update mental models for sustained project context
 
 ## Loop Guard Handling
 
@@ -187,4 +165,4 @@ You may run while implementation agents (Thor, Tyr, Heimdall, Vidarr, Mimir, Bal
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/mimir.md

@@ -14,8 +14,6 @@ permission:
   webfetch: allow
   websearch: allow
   todowrite: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -83,26 +81,6 @@ Odin may dispatch you to create `.bizar/PROJECT.md` for a new project. This is a
 
 Keep it 20-40 lines. This is a living document — @heimdall will update it as the project evolves.
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### During Work
-- `hindsight_retain` important findings with the correct `bank_id`
-- Tag memories with `project:<repo-name>`
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
-- Create or update mental models for sustained project context
 
 ## Loop Guard Handling
 
@@ -162,4 +140,4 @@ End your final summary with: `Siblings: <list>. Conflicts: <list or "none">. Git
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/odin.md

@@ -10,10 +10,6 @@ permission:
   todowrite: allow
   webfetch: allow
   websearch: allow
-  hindsight_recall: allow
-  hindsight_retain: allow
-  hindsight_list_banks: allow
-  hindsight_create_bank: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -222,26 +218,6 @@ Update .bizar/ in this project.
 2. Update PROJECT.md if this task revealed new project info
 ```
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### At Session Start
-1. Call `hindsight_list_banks` to discover available banks
-2. Determine the project name from the working directory or task context
-3. Use `bank_id: "<project-name>"` in all Hindsight calls
-4. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-
-### During Work
-- `hindsight_recall` with the correct `bank_id` for relevant context
-- `hindsight_retain` important context, architecture, conventions, decisions
-- Before significant changes, `hindsight_recall` for related prior work
-- Tag memories with `project:<repo-name>`
-
-### On Task Completion
-- `hindsight_retain` what was accomplished, key decisions, files changed
-- Tag memories with `project:<repo-name>`
-- Create or update mental models for sustained project context
 
 ## Parallel Dispatch Coordination
 
@@ -409,4 +385,4 @@ When uncertain or stuck, follow `config/rules/uncertainty.md` — stop and resea
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian vault, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.

package/config/agents/quick.md

@@ -15,8 +15,6 @@ permission:
   webfetch: allow
   websearch: allow
   task: deny
-  hindsight_recall: allow
-  hindsight_retain: allow
 ---
 
 ## Codebase Search — Use Semble First
@@ -62,21 +60,6 @@ If a task requires:
 
 …then tell the user to use `@odin` instead. You are not a router.
 
-## Hindsight Memory Protocol
-
-You MUST use **per-project banks** — never the default bank for project work.
-
-### Bank Selection
-1. Call `hindsight_list_banks` to discover available banks
-2. Use `bank_id: "<project-name>"` in all Hindsight calls
-3. If no bank exists for the project, create it with `hindsight_create_bank(bank_id: "<project-name>")`
-4. The default bank is for general/system knowledge only
-
-### Before Work
-- `hindsight_recall` with the correct `bank_id` for existing context
-
-### After Work
-- `hindsight_retain` completion summary into the project bank
 
 ## Loop Guard Handling
 
@@ -102,4 +85,4 @@ When uncertain or stuck, follow `config/rules/uncertainty.md` — stop and resea
 
 **Follow the global baseline in `config/AGENTS.md` → "General Agent Baseline — Always-On Behavior".** It covers identity, refusal, tone, formatting, lists, user wellbeing, evenhandedness, mistakes, knowledge cutoff and research-first, MCP servers and skills, mandatory skill-read, file creation, file handling, search, copyright, harmful content, citations, images, memory privacy, execution, clarification, and communication.
 
-The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Hindsight, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.
+The section above was adapted from the upstream Claude Fable 5 system prompt, with every Claude-specific tool / function / directory translated to the BizarHarness equivalent (opencode tools, Semble, Skills CLI, Obsidian, agent-browser, the dashboard artifact pipeline). Do not duplicate the rules here — read the global baseline and apply it.