@radix-ai/ai-memory 0.1.0

package/templates/.ai/reference/environment-specs.json

@@ -0,0 +1,41 @@
+{
+  "version": 1,
+  "environments": [
+    {
+      "id": "cursor",
+      "name": "Cursor",
+      "detect": { "paths": [".cursor/", ".cursorrules"] },
+      "capabilities": {
+        "browser": { "native": true },
+        "screen_capture": { "platform": true }
+      }
+    },
+    {
+      "id": "claude-code",
+      "name": "Claude Code",
+      "detect": { "paths": [".claude/", ".claude-plugin/", "CLAUDE.md"] },
+      "capabilities": {
+        "browser": { "mcp": true },
+        "screen_capture": { "platform": true }
+      }
+    },
+    {
+      "id": "windsurf",
+      "name": "Windsurf",
+      "detect": { "paths": [".windsurfrules"] },
+      "capabilities": {
+        "browser": { "mcp": true },
+        "screen_capture": { "platform": true }
+      }
+    },
+    {
+      "id": "cline",
+      "name": "Cline",
+      "detect": { "paths": [".clinerules"] },
+      "capabilities": {
+        "browser": { "mcp": true },
+        "screen_capture": { "platform": true }
+      }
+    }
+  ]
+}

package/templates/.ai/sessions/archive/thread-archive.md

@@ -0,0 +1,15 @@
+# Thread Archive — Curated Session Summaries
+
+**Format:** One entry per session. Include proposals, accepted, rejected, and why.
+
+---
+
+<!-- Add entries below. Format:
+
+## YYYY-MM-DD — Session title
+
+**Proposed:** ...
+**Accepted:** ...
+**Rejected:** ... (reason)
+**Open:** ...
+-->

package/templates/.ai/sessions/open-items.md

@@ -0,0 +1,13 @@
+# Open Items — Live Registry
+
+**Compound Step 3c.** Track open/closed items. Update during compound; surface at session end.
+
+---
+
+## Open
+
+<!-- Add open items. Format: `- [ ] Brief description (source)` -->
+
+## Closed
+
+<!-- Add closed items. Format: `- [x] Brief description (resolved how)` -->

package/templates/.ai/toolbox/README.md

@@ -0,0 +1,5 @@
+# Toolbox — Technology/Capability Knowledge
+
+Add technology/capability docs here. Example: `shell.md`, `browser.md`, `hooks.md`, `mcp.md`, `context-engineering.md`.
+
+**Placement test:** Is this about THIS PROJECT's domain? If no → toolbox/. If yes → skills/.

package/templates/.ai/toolbox/browser.md

@@ -0,0 +1,30 @@
+---
+id: browser
+type: toolbox
+status: active
+---
+
+# Browser & Testing Patterns
+
+How to use browser tools (Playwright, Chrome, Puppeteer) with ai-memory.
+
+## Capture test results as memory
+
+After running browser tests:
+- **Failures** → write to `debugging.md` via `commit_memory` with symptom, screenshot path, and root cause
+- **New patterns** → write to `patterns.md` (e.g., "modal always needs 500ms delay after animation")
+- **Flaky tests** → tag as `[P1]` in debugging.md with reproduction steps
+
+## Screenshot as evidence
+
+When browser tools produce screenshots:
+- Reference the screenshot path in memory entries
+- Include the URL and viewport dimensions
+- Tag entries with the component/page name for searchability
+
+## Visual regression patterns
+
+When visual differences are detected:
+- Check memory for known acceptable changes (`search_memory` for the component)
+- If the change was intentional (matches a decision in `decisions.md`), skip
+- If unexpected, create a debugging entry with before/after paths

package/templates/.ai/toolbox/integrations.md

@@ -0,0 +1,44 @@
+---
+id: integrations
+type: toolbox
+status: active
+---
+
+# Integration Patterns
+
+How to use ai-memory with external platforms and automation triggers.
+
+## Slack → Agent → Memory
+
+When triggered from Slack (Cursor `@cursor` or custom webhook):
+1. Agent clones repo — `.ai/` is available because it's git-tracked
+2. Read IDENTITY.md and DIRECTION.md for project context
+3. Search memory for relevant decisions before starting work
+4. Use `claim_task` to prevent duplicate work
+5. After work: `publish_result` + `sync_memory` + push branch
+
+## GitHub PR → Review → Memory
+
+When triggered by PR creation (Bugbot, automation, CI):
+1. Get diff with `git diff origin/main...HEAD`
+2. Run `validate_context` against governance rules
+3. Search memory for patterns relevant to changed files
+4. Report violations and suggestions
+5. Record result with `publish_result`
+
+## Linear/Jira → Task → Memory
+
+When triggered by issue assignment:
+1. Read issue context from the trigger
+2. `claim_task` with the issue description
+3. Search memory for related decisions and patterns
+4. Do the work, capture learnings
+5. `publish_result` + `sync_memory`
+
+## Scheduled/Cron → Maintenance → Memory
+
+For recurring tasks (weekly audit, dependency check):
+1. Read current memory state
+2. Run `ai-memory eval` for health metrics
+3. Run `ai-memory prune` for stale entries
+4. Publish maintenance report to thread-archive

package/templates/.ai/toolbox/shell.md

@@ -0,0 +1,38 @@
+---
+id: shell
+type: toolbox
+status: active
+---
+
+# Shell & CLI Patterns
+
+How to use shell commands effectively with ai-memory.
+
+## Feed command output to memory
+
+| Command | Memory action |
+|---|---|
+| `git diff` | Feed to `validate_context` for governance checks |
+| `npm test` / `pytest` | Capture failures in `debugging.md` |
+| `npm run build` | Capture build errors and fixes |
+| Deploy scripts | Record deployment decisions in `decisions.md` |
+| `git log --oneline -20` | Use as context when writing session archive |
+
+## Memory-aware shell patterns
+
+Before running destructive commands:
+1. `search_memory` for relevant patterns or warnings
+2. Check `decisions.md` for any constraints on the operation
+3. After completion, `commit_memory` with what happened
+
+## CI/CD integration
+
+In CI pipelines, use the ai-memory CLI directly:
+```bash
+# Validate memory health
+npx @radix-ai/ai-memory eval --json
+
+# Check governance rules
+npx @radix-ai/ai-memory generate-harness
+npx @radix-ai/ai-memory validate
+```

package/templates/AGENTS.md

@@ -0,0 +1,4 @@
+# Agent Instructions
+
+See `.ai/IDENTITY.md` for behavioral constraints and `.ai/PROJECT_STATUS.md` for current project status.
+For ACP agent card, see `.ai/acp/manifest.json`.