portable-agent-layer 0.39.0 → 0.41.0

package/README.md

@@ -2,7 +2,7 @@
 
 A cross-platform, cross-agent layer for portable AI workflows, memory, and accumulated knowledge.
 
-PAL lets you carry your agent context across **Windows**, **macOS**, and **Linux**, and work across different agent runtimes and interfaces such as **Claude Code**, **opencode**, **Cursor**, and **Codex**. Its core idea is simple: your knowledge and workflows should belong to **you**, not to a single machine, tool, or vendor.
+PAL lets you carry your agent context across **Windows**, **macOS**, and **Linux**, and work across different agent runtimes and interfaces such as **Claude Code**, **opencode**, **Cursor**, **GitHub Copilot**, and **Codex**. Its core idea is simple: your knowledge and workflows should belong to **you**, not to a single machine, tool, or vendor.
 
 > Inspired in part by [Daniel Miessler](https://danielmiessler.com)'s work on [Personal AI Infrastructure](https://github.com/danielmiessler/Personal_AI_Infrastructure). PAL is an independent open-source implementation focused on portability across platforms and agents. It is not affiliated with or endorsed by Daniel Miessler.
 
@@ -33,7 +33,7 @@ With PAL, you can:
 > **Bun is required.** PAL is built on [Bun](https://bun.sh) and will not work with Node.js or other runtimes. Install it with `curl -fsSL https://bun.sh/install | bash`.
 
 - [Bun](https://bun.sh) >= 1.3.0
-- At least one of: [Claude Code](https://claude.ai/code), [opencode](https://opencode.ai), [Cursor](https://cursor.com), or [Codex](https://openai.com/index/introducing-codex/)
+- At least one of: [Claude Code](https://claude.ai/code), [opencode](https://opencode.ai), [Cursor](https://cursor.com), [GitHub Copilot CLI](https://docs.github.com/en/copilot/github-copilot-in-the-cli), or [Codex](https://openai.com/index/introducing-codex/)
 
 ### Package mode (recommended)
 
@@ -98,31 +98,36 @@ pal cli install               # all available (default)
 
 ### Supported agents
 
-| Agent | Support | Skills | Hooks | AGENTS.md | Subagents |
-|-------|---------|--------|-------|-----------|-----------|
-| Claude Code | Full | Yes | Yes | Yes | Yes |
-| opencode | Full | Yes | Yes (plugin) | Yes | Yes |
-| Cursor | Full | Yes | Yes | Yes (injected via hook) | Yes |
-| GitHub Copilot | Full | Yes | Yes | Yes (via `~/.copilot/instructions/*.instructions.md`) | Yes |
-| Codex | Partial | Yes | No | Yes | No |
+| Agent | Support | Skills | Hooks | AGENTS.md | Subagents | Inference routing |
+|-------|---------|--------|-------|-----------|-----------|-------------------|
+| Claude Code | Full | Yes | Yes | Yes | Yes | `claude --print` |
+| opencode | Full | Yes | Yes (plugin) | Yes | Yes | `opencode run` |
+| Cursor | Full | Yes | Yes | Yes (injected via hook) | Yes | `cursor-agent` |
+| GitHub Copilot | Full | Yes | Yes | Yes (via `~/.copilot/instructions/*.instructions.md`) | Yes | `copilot` |
+| Codex | Full | Yes | Yes | Yes | No | `codex exec` |
+
+PAL's background inference (session naming, summaries, failure capture, etc.) runs through whichever subscription CLI is active — no API key required by default.
 
 ---
 
 ## Environment variables
 
-### Required
+### API keys (all optional)
+
+PAL routes inference through the host agent's subscription CLI by default. API keys are only needed as fallbacks when no CLI binary is available, or for skills that call non-Anthropic providers.
 
 | Variable | Description |
 |----------|-------------|
-| `PAL_ANTHROPIC_API_KEY` | Required for PAL's hook inference (sentiment analysis, session naming). Uses Haiku for low-cost background calls. |
+| `PAL_ANTHROPIC_API_KEY` | Fallback for hook inference when no native CLI is available. Uses Haiku. |
+| `PAL_OPENAI_API_KEY` | Fallback for hook inference when Codex is active without the `codex` binary, or when no Anthropic key is set. |
+| `PAL_GEMINI_API_KEY` | For YouTube video analysis and web search skill |
+| `PAL_XAI_API_KEY` | For Grok real-time research skill (X/web search) |
+| `PAL_PERPLEXITY_API_KEY` | For Perplexity deep research skill |
 
-### Optional
+### Path overrides
 
 | Variable | Description |
 |----------|-------------|
-| `PAL_GEMINI_API_KEY` | For YouTube video analysis and web search skill  |
-| `PAL_XAI_API_KEY` | For Grok real-time research skill (X/web search) |
-| `PAL_PERPLEXITY_API_KEY` | For Perplexity deep research skill |
 | `PAL_HOME` | Override user state directory (default: `~/.pal` or repo root) |
 | `PAL_PKG` | Override package root |
 | `PAL_CLAUDE_DIR` | Override Claude config dir (default: `~/.claude`) |
@@ -132,6 +137,14 @@ pal cli install               # all available (default)
 | `PAL_CODEX_DIR` | Override Codex config dir (default: `~/.codex`) |
 | `PAL_AGENTS_DIR` | Override agents dir (default: `~/.agents`) |
 
+### Debug / test
+
+| Variable | Description |
+|----------|-------------|
+| `PAL_DEBUG` | Set to `1` to emit verbose hook logs to `memory/state/debug.log` |
+| `PAL_INFERENCE_DISABLED` | Set to `1` to disable all inference (used by the test suite to prevent real CLI spawns) |
+| `PAL_NOTIFICATIONS_DISABLED` | Set to `1` to suppress desktop notifications (used by the test suite) |
+
 ---
 
 ## Skills
@@ -142,16 +155,23 @@ PAL ships with built-in skills that extend your agent's capabilities:
 |-------|-------------|
 | `analyze-pdf` | Download and analyze PDF files |
 | `analyze-youtube` | Analyze YouTube videos using Gemini |
+| `consulting-report` | Generate consulting-style reports as PDFs |
 | `council` | Multi-perspective parallel debate on decisions |
+| `create-pdf` | Render structured content into a PDF |
 | `create-skill` | Scaffold a new skill from a description |
 | `extract-entities` | Extract people and companies from content |
 | `extract-wisdom` | Extract structured insights from content |
 | `first-principles` | Break down problems to fundamentals |
 | `fyzz-chat-api` | Query Fyzz Chat conversations via API |
+| `opinion` | Confirm or contradict tracked opinions (confidence-weighted) |
+| `presentation` | Build branded slide decks from outlines |
+| `projects` | Look up, resume, register, or manage tracked projects |
 | `reflect` | Diagnose why a PAL behavior didn't trigger |
 | `research` | Multi-agent parallel research |
 | `review` | Security-focused code review |
 | `summarize` | Structured summarization |
+| `telos` | Inspect or update goals, beliefs, strategies, narratives |
+| `think` | Structured first-pass reasoning on a problem |
 
 ---
 
@@ -178,7 +198,8 @@ Your setup should be able to travel with you.
 ## Features
 
 - **Cross-platform**: works on Windows, macOS, and Linux
-- **Cross-agent**: full support for Claude Code, opencode, and Cursor; partial support for Codex
+- **Cross-agent**: full support for Claude Code, opencode, Cursor, GitHub Copilot, and Codex (Codex still lacks subagents)
+- **Subscription-first inference**: background inference routes through whichever subscription CLI is active — no API key needed by default
 - **Portable knowledge**: export and import accumulated knowledge
 - **TypeScript-first**: built in TypeScript from day one
 - **Open source**: hackable, inspectable, extensible

package/assets/templates/PAL/MEMORY_SYSTEM.md

@@ -1,26 +1,72 @@
 # Memory System
 
-PAL has its own memory system that persists across sessions AND across tools (Claude Code, opencode). Always prefer PAL memory over any tool-native memory system.
+PAL has its own memory system that persists across sessions AND across tools (Claude Code, opencode, Cursor, Copilot, Codex). Always prefer PAL memory over any tool-native memory system.
+
+## Layout
+
+All memory lives under `~/.pal/memory/`:
+
+```
+memory/
+├── state/                              # Runtime state (managed by hooks)
+│   ├── sessions.json                   # Session registry
+│   ├── counts.json                     # Cached counts for greeting
+│   ├── last-responses.json             # Cached responses for rating correlation
+│   ├── pending-failure.json            # Deferred failure capture
+│   ├── captured-learnings.json         # Session-intelligence dedup map
+│   ├── session-names.json              # Generated session titles
+│   ├── debug.log[.1-.5]                # Rotated hook execution logs
+│   └── token-usage.jsonl               # Per-call inference token usage
+│
+├── signals/
+│   └── ratings.jsonl                   # Append-only rating signals (explicit + implicit)
+│
+├── relationship/
+│   ├── YYYY-MM/YYYY-MM-DD.md           # Daily interaction notes (W / O / B)
+│   ├── opinions.json                   # Confidence-tracked opinions
+│   └── reflections/                    # Periodic reflection reports
+│
+├── self-model/
+│   └── current.md                      # Synthesized self-model (auto-imported into CLAUDE.md)
+│
+├── learning/
+│   ├── session/YYYY-MM/*.md            # Per-session learnings with frontmatter
+│   ├── failures/YYYY-MM/*.md           # Low-rating context captures
+│   ├── synthesis/YYYY-MM/*.md          # Weekly/monthly synthesis reports
+│   └── .retrieval-index.json           # Embedding index for retrieval
+│
+├── wisdom/
+│   ├── frames/<domain>.md              # Crystallized principles per domain
+│   └── state/                          # Wisdom graduation bookkeeping
+│
+└── projects/
+    └── <slug>/                         # Per-project history, threads, progress
+```
 
 ## Where to write
 
-- **Wisdom frames**: `~/.pal/memory/wisdom/frames/` — crystallized principles per domain (loaded every session)
-- **Relationship notes**: `~/.pal/memory/relationship/YYYY-MM/YYYY-MM-DD.md` — daily interaction observations (loaded every session)
-- **Session learnings**: `~/.pal/memory/learning/session/YYYY-MM/*.md` — reusable insights from sessions (loaded every session)
-- **Failure captures**: `~/.pal/memory/learning/failures/YYYY-MM/{timestamp}_{slug}/capture.md` — what went wrong and why
-- **Signals**: `~/.pal/memory/signals/ratings.jsonl` — append-only rating signal log (do not edit directly)
+Almost everything in `state/`, `signals/`, `relationship/opinions.json`, `learning/`, `wisdom/`, and `self-model/` is hook-managed — **never edit these directly**. Use the relevant skill or tool instead.
+
+| You want to capture | Go through | Lands at |
+|---------------------|------------|----------|
+| A new validated principle | `skill: opinion` (confirm) or wisdom frame edit | `wisdom/frames/<domain>.md` |
+| A daily observation about the user | Relationship note in algorithm LEARN phase | `relationship/YYYY-MM/YYYY-MM-DD.md` |
+| A reusable session insight | Session-intelligence handler (automatic) | `learning/session/YYYY-MM/*.md` |
+| A failure worth learning from | Failure-principle handler (automatic, low ratings) | `learning/failures/YYYY-MM/*.md` |
+| A rating signal | UserPromptOrchestrator rating capture | `signals/ratings.jsonl` |
+| A project handoff | Project skill / detached handlers | `projects/<slug>/` |
 
-## Format
+## Format conventions
 
-- **Wisdom frames**: One `.md` file per domain/topic. Each file contains bullet-point principles the user has validated or you've learned. Append new principles to existing files or create new domain files.
-- **Relationship notes**: Daily `.md` file with bullet-point observations about the interaction (tone, preferences, corrections).
-- **Session learnings**: One `.md` file per session with a `**Title:**` line summarizing what was learned.
-- **Failure captures**: One directory per failure, named `{YYYYMMDD-HHmmss}_{slug}/`, containing a `capture.md` with what went wrong and why.
+- **Wisdom frames**: One `.md` file per domain. Bullet-point principles, each tagged with a confidence value (e.g. `(85%)`). Append to existing files when possible; create a new domain only when none fits.
+- **Relationship notes**: Daily `.md` file. Each bullet is typed: `W` (world fact), `O(c=0.85)` (opinion), or `B(c=0.75)` (behavior pattern), with a confidence in `c=...`.
+- **Session learnings**: Frontmatter (`title`, `category`, `date`, `cwd`, optional `session`) + a body split into `## What Was Done` and `## Insights`.
+- **Failure captures**: One `.md` per failure, named `{YYYYMMDD-HHmmss}_{slug}.md`, describing what went wrong, why, and what to do differently.
+- **Synthesis reports**: One `.md` per period (`YYYY-MM-DD_weekly.md` etc.) aggregating recurring patterns from session learnings + failures.
 
-## When to write
+## When NOT to write
 
-- When the user corrects you or gives feedback → wisdom frame
-- When you learn something about how the user prefers to work → relationship note
-- When a session produces reusable insights → session learning
-- When something fails significantly (rating < 6) → failure capture
-- Do NOT write memories about trivial exchanges or things already captured in TELOS.
+- Trivial exchanges (greetings, brief acknowledgments) → skip
+- Anything already captured in TELOS → skip
+- Anything that duplicates an existing wisdom frame or opinion → use the opinion skill to reinforce instead
+- Anything the user has not validated yet — record it as a relationship note first, let it graduate naturally

package/assets/templates/PAL/SYSTEM_ARCHITECTURE.md

@@ -227,12 +227,20 @@ Brief description.
 └─────────────────────┘
 
 ┌─────────────────────┐
-│       Stop          │──► StopOrchestrator.ts
+│       Stop          │──► StopOrchestrator.ts → runStopHandlers()
+│                     │    Detached (inference-bearing, fire-and-forget):
+│                     │    - Session intelligence (title, summary, insights, handoff)
+│                     │    - Failure principle extraction (low ratings)
+│                     │    - Auto-graduate wisdom (24h TTL, idempotent)
+│                     │    Concurrent (Promise.allSettled):
 │                     │    - Work session capture
-│                     │    - Relationship extraction (Haiku inference)
-│                     │    - Work learning capture
-│                     │    - Failure logging
+│                     │    - Project touch / history
+│                     │    - Persist last exchange (handoff seed)
+│                     │    - Cache last response (rating correlation)
 │                     │    - Reflect trigger check
+│                     │    - Self-model trigger check
+│                     │    - Synthesis pipeline
+│                     │    - Desktop notify
 │                     │    - Write context digests (semi-static sources)
 │                     │    - Auto-backup
 │                     │    - Count updates
@@ -243,9 +251,11 @@ Brief description.
 ### Design Principles
 
 - **Fail-open**: Hook errors never block the user's session
-- **Parallel execution**: Stop handlers run via `Promise.allSettled` — one failure doesn't block others
-- **Idempotent**: Handlers check for existing state before writing (e.g., session dedup)
-- **Timeout-aware**: Inference calls have hard timeouts (8s default)
+- **Detached inference**: Long inference calls (session intelligence, failure principle, auto-graduate) spawn detached child processes that survive the hook returning. The parent hook exits in milliseconds — cold-start latency for the subscription CLI no longer blocks the next prompt.
+- **Parallel execution**: Non-inference handlers run via `Promise.allSettled` — one failure doesn't block others
+- **Idempotent**: Handlers check for existing state before writing (auto-graduate uses 24h TTL + content dedup; session intelligence dedups by message count)
+- **Timeout-aware**: Inference calls have hard timeouts (60s default; longer because they run detached and don't block the user)
+- **Race-safe**: Detached claim files use atomic `rename()` with ENOENT treated as a benign "already claimed" signal (opencode fires `session.idle` + `session.diff` concurrently)
 - **Subagent-aware**: `LoadContext.ts` skips heavy context loading for subagent sessions
 
 ### Configuration
@@ -404,6 +414,66 @@ Everything else loads via the routing table in CLAUDE.md. The AI reads files onl
 
 ---
 
+## Inference Routing Architecture
+
+### Subscription-First Model
+
+PAL's background inference (session naming, summaries, failure capture, wisdom graduation, etc.) routes through whichever subscription CLI is currently hosting the session. No API key is required by default — the user already pays for the subscription that comes with their agent.
+
+### Routing Table
+
+The active agent is detected via `getActiveAgent()` in `src/hooks/lib/agent.ts` (driven by the `PAL_AGENT` env var set per-target). The dispatcher in `src/hooks/lib/inference.ts` then routes to the matching CLI:
+
+| Active agent | CLI invoked | Notes |
+|--------------|-------------|-------|
+| `claude` | `claude --print` | Inherits the user's Claude Code subscription |
+| `opencode` | `opencode run` | Output parsed via `extractOpencodeText()` |
+| `cursor` | `cursor-agent` | argv-only, no stdin |
+| `copilot` | `copilot` | GitHub Copilot CLI |
+| `codex` | `codex exec` | Falls back to `PAL_OPENAI_API_KEY` if the `codex` binary is missing |
+
+If no CLI binary is available, the dispatcher falls back to `PAL_ANTHROPIC_API_KEY` (Haiku via the Anthropic API) or `PAL_OPENAI_API_KEY` (OpenAI API). `canInfer()` returns `false` only when both routes are unavailable — handlers then skip silently.
+
+### Recursion Defense (Three Layers)
+
+Spawning a CLI from inside a hook risks the child re-triggering the same hooks, leading to an infinite spawn loop. Three defenses, all active at once:
+
+1. **CLI flags** — Mirrors PAI's pattern: `--setting-sources ''` and `--tools ''` strip the child of its hook config and tool access entirely (claude, opencode equivalents).
+2. **Sentinel env var** — `PAL_SPAWNED_INFERENCE=1` is set on every spawn. Hook entry points call `exitIfSpawnedInference()` and bail before doing any work.
+3. **Depth counter** — `PAL_SPAWN_DEPTH` increments per hop; the circuit breaks at `MAX_DEPTH=1`. Belt-and-suspenders for the above.
+
+The `CLAUDECODE` env var is unset *scoped to the child process only* (not deleted from the parent's environment) so the spawned `claude` CLI doesn't think it's nested inside another Claude session.
+
+### Detached Inference Pattern
+
+Long inference calls (session intelligence, failure principle extraction, auto-graduate) run via `spawnDetachedInference()`. The parent hook writes the transcript to a tmp file, spawns a detached child with `--run <args>`, and returns immediately. The child reads the tmp file, runs inference, writes results to disk, and unlinks the tmp file.
+
+```
+StopOrchestrator
+  │
+  ├─► detachSessionIntelligence(transcript, sessionId)
+  │     └─► [child] inference → session-learning/YYYY-MM/...
+  │
+  ├─► detachFailurePrinciple(transcript)
+  │     ├─► Atomic claim: rename pending-failure.json → tmp/pending.json
+  │     │   (ENOENT means another concurrent Stop hook claimed it — benign)
+  │     └─► [child] inference → failures/YYYY-MM/...
+  │
+  └─► detachAutoGraduate()
+        └─► [child] 24h TTL + content dedup → wisdom/frames/...
+```
+
+The orchestrator returns in milliseconds; subscription-CLI cold-start latency no longer blocks the user's next prompt.
+
+### Resilience
+
+- **Empty-abort retry** — If a CLI returns an empty stdout (burst-concurrency race), retry once with 500–1500ms jitter
+- **Test kill-switch** — `PAL_INFERENCE_DISABLED=1` (preloaded by `test/setup.ts`) short-circuits all inference so tests never spawn real CLIs
+- **Per-call attribution** — Every inference call logs `caller=X sessionId=Y` so debug.log makes routing trivially auditable
+- **Doctor probes** — `pal cli doctor` actually invokes each route and reports which agents are correctly wired
+
+---
+
 ## Security Architecture
 
 ### Fail-Open Design
@@ -450,10 +520,13 @@ src/targets/
 ├── copilot/         # GitHub Copilot specific
 │   ├── install.ts   # Write instruction files + update VS Code settings
 │   └── uninstall.ts
+├── codex/           # Codex specific
+│   ├── install.ts   # Merge hooks into ~/.codex/hooks.json + skills + AGENTS.md
+│   └── uninstall.ts
 └── lib.ts           # Shared: JSON read/write, settings merge, TELOS scaffold
 ```
 
-Codex support is partial — AGENTS.md is symlinked to `~/.codex/AGENTS.md` automatically (no dedicated target installer needed).
+All five targets install hooks, skills, and AGENTS.md. Codex does not yet support subagents.
 
 ### Path Resolution
 

package/assets/templates/hooks.copilot.json

@@ -10,23 +10,23 @@
     "userPromptSubmitted": [
       {
         "type": "command",
-        "bash": "bun run {{PKG_ROOT}}/src/hooks/UserPromptOrchestrator.ts"
+        "bash": "PAL_AGENT=copilot bun run {{PKG_ROOT}}/src/hooks/UserPromptOrchestrator.ts"
       }
     ],
     "preToolUse": [
       {
         "type": "command",
-        "bash": "bun run {{PKG_ROOT}}/src/hooks/SecurityValidator.ts"
+        "bash": "PAL_AGENT=copilot bun run {{PKG_ROOT}}/src/hooks/SecurityValidator.ts"
       },
       {
         "type": "command",
-        "bash": "bun run {{PKG_ROOT}}/src/hooks/SkillGuard.ts"
+        "bash": "PAL_AGENT=copilot bun run {{PKG_ROOT}}/src/hooks/SkillGuard.ts"
       }
     ],
     "agentStop": [
       {
         "type": "command",
-        "bash": "bun run {{PKG_ROOT}}/src/hooks/StopOrchestrator.ts"
+        "bash": "PAL_AGENT=copilot bun run {{PKG_ROOT}}/src/hooks/StopOrchestrator.ts"
       }
     ]
   }

package/assets/templates/settings.claude.json

@@ -30,7 +30,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/LoadContext.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/LoadContext.ts"
           }
         ]
       },
@@ -39,7 +39,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/CompactRecover.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/CompactRecover.ts"
           }
         ]
       }
@@ -50,7 +50,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/PreCompactPersist.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/PreCompactPersist.ts"
           }
         ]
       }
@@ -61,7 +61,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/UserPromptOrchestrator.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/UserPromptOrchestrator.ts"
           }
         ]
       }
@@ -72,7 +72,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/SecurityValidator.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/SecurityValidator.ts"
           }
         ]
       },
@@ -81,7 +81,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/SkillGuard.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/SkillGuard.ts"
           }
         ]
       }
@@ -92,7 +92,7 @@
         "hooks": [
           {
             "type": "command",
-            "command": "bun run {{PKG_ROOT}}/src/hooks/StopOrchestrator.ts"
+            "command": "PAL_AGENT=claude bun run {{PKG_ROOT}}/src/hooks/StopOrchestrator.ts"
           }
         ]
       }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "portable-agent-layer",
-  "version": "0.39.0",
+  "version": "0.41.0",
   "description": "PAL — Portable Agent Layer: persistent personal context for AI coding assistants",
   "type": "module",
   "bin": {
@@ -42,6 +42,8 @@
     "check-write": "biome check --write",
     "knip": "knip-bun",
     "klint": "klint",
+    "jscpd": "jscpd src --noTips --silent",
+    "jscpd:report": "jscpd src --noTips",
     "lint-staged": "lint-staged",
     "prepare": "bun .husky/install.mjs",
     "install:all": "bun run src/cli/index.ts cli install",
@@ -66,21 +68,22 @@
     "@opencode-ai/plugin": "latest",
     "@semantic-release/changelog": "^6.0.3",
     "@semantic-release/git": "^10.0.1",
-    "@semantic-release/github": "^12.0.6",
+    "@semantic-release/github": "^12.0.8",
     "@types/adm-zip": "^0.5.8",
     "@types/bun": "latest",
     "@types/node": "latest",
     "husky": "^9.1.7",
-    "knip": "^6.9.0",
+    "jscpd": "^4.2.3",
+    "knip": "^6.14.1",
     "lint-staged": "^15.5.2",
     "semantic-release": "^25.0.3",
     "typescript": "^5.9.3"
   },
   "dependencies": {
-    "@clack/prompts": "^1.3.0",
+    "@clack/prompts": "^1.4.0",
     "@konvert7/klint": "^0.2.0",
     "adm-zip": "^0.5.17",
     "marked": "^15.0.12",
-    "playwright": "^1.59.1"
+    "playwright": "^1.60.0"
   }
 }