@counterposition/skill-pi 0.74.0 → 0.75.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@counterposition/skill-pi",
-  "version": "0.74.0",
+  "version": "0.75.0",
   "description": "Pi coding agent reference skill — extensions, settings, providers, SDK/RPC, packages, and .pi/ config",
   "keywords": [
     "ai-skill",

package/skills/pi/references/packages.md

@@ -23,7 +23,7 @@ pi update <source>        # update one package
 pi config
 ```
 
-By default, `install` and `remove` write to global settings (`~/.pi/agent/settings.json`). Use `-l` to write to project settings (`.pi/settings.json`) instead.
+By default, `install` and `remove` write to user settings (`~/.pi/agent/settings.json`). Use `-l` to write to project settings (`.pi/settings.json`) instead.
 
 Project installs are shareable with a team. Pi will install missing project packages automatically on startup.
 
@@ -38,7 +38,7 @@ npm:@scope/pkg@1.2.3
 npm:pkg
 ```
 
-- Global installs use `npm install -g`
+- User installs go under `~/.pi/agent/npm/`
 - Project installs go under `.pi/npm/`
 - Version-pinned installs are skipped by `pi update`
 - Use `npmCommand` in settings if you need a wrapper such as `mise` or `asdf`

package/skills/pi/references/providers.md

@@ -49,7 +49,8 @@ Use `/logout` to clear stored OAuth credentials. Pi 0.71.0 removed built-in Goog
 {
   "anthropic": { "type": "api_key", "key": "sk-ant-..." },
   "openai": { "type": "api_key", "key": "!op read 'OpenAI API Key'" },
-  "google": { "type": "api_key", "key": "GEMINI_API_KEY" }
+  "google": { "type": "api_key", "key": "GEMINI_API_KEY" },
+  "together": { "type": "api_key", "key": "TOGETHER_API_KEY" }
 }
 ```
 
@@ -167,7 +168,12 @@ Common `api` values:
 
 - `thinkingLevelMap` (Pi 0.72) replaces `compat.reasoningEffortMap`. Map pi levels (`off`/`minimal`/`low`/`medium`/`high`/`xhigh`) to provider values; use `null` to hide a level.
 - `openRouterRouting` is forwarded as-is in the OpenRouter `provider` field (fallbacks, ZDR, ignore lists, throughput/latency).
-- Advanced `compat` flags exist for proxy quirks (`cacheControlFormat`, `supportsLongCacheRetention`, `supportsEagerToolInputStreaming`, `sendSessionIdHeader`, `sendSessionAffinityHeaders`). See [Pi `docs/models.md`](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/docs/models.md) when a proxy rejects pi's defaults.
+- `compat.thinkingFormat` supports OpenAI-compatible reasoning variants: `openrouter` sends `reasoning: { effort }`, `together` sends `reasoning: { enabled }` plus `reasoning_effort` when supported, and `qwen-chat-template` targets local Qwen-compatible servers that read `chat_template_kwargs.enable_thinking`.
+- Advanced `compat` flags exist for proxy quirks (`cacheControlFormat`, `supportsReasoningEffort`, `supportsLongCacheRetention`, `supportsEagerToolInputStreaming`, `sendSessionIdHeader`, `sendSessionAffinityHeaders`). See [Pi `docs/models.md`](https://github.com/earendil-works/pi/blob/main/packages/coding-agent/docs/models.md) when a proxy rejects pi's defaults.
+
+### Context Overflow Recovery
+
+Pi can auto-compact and retry when a provider fails with a recognized context-window overflow. For custom providers whose overflow text Pi does not recognize, use a provider-scoped `message_end` extension handler to rewrite the finalized assistant `errorMessage` so it starts with `context_length_exceeded`.
 
 ## Custom Providers via Extensions
 

package/skills/pi/references/settings.md

@@ -114,7 +114,7 @@ Notes:
 }
 ```
 
-- `npmCommand` is argv-style and is used for npm lookup/install operations, including git-package installs.
+- `npmCommand` is argv-style and is used for npm lookup/install operations, including git-package installs. User-scoped npm packages install under `~/.pi/agent/npm/`; project-scoped npm packages install under `.pi/npm/`.
 - `terminal.showTerminalProgress` (default `false` since Pi 0.70.0) toggles OSC 9;4 progress reporting in supporting terminals (iTerm2, WezTerm, Windows Terminal, Kitty).
 - `terminal.imageWidthCells` (Pi 0.68.1) caps inline tool-output image width in terminal cells.
 

package/skills/pi/references/skills.md

@@ -1,6 +1,6 @@
 # Skills
 
-Skills are self-contained capability packages that Pi loads on demand. Pi follows the [Agent Skills specification](https://agentskills.io/specification) and stays fairly lenient, warning on many violations instead of hard-failing.
+Skills are self-contained capability packages that Pi loads on demand. Pi follows the [Agent Skills specification](https://agentskills.io/specification) and stays fairly lenient, warning on most violations instead of hard-failing. Pi allows skill names to differ from parent directory names so shared skill directories can work across multiple agent harnesses.
 
 ## How Skills Work
 
@@ -83,7 +83,7 @@ Read `references/api.md` for the full details.
 
 | Field | Required | Notes |
 |-------|----------|-------|
-| `name` | Yes | 1-64 chars, lowercase letters/numbers/hyphens, must match parent dir |
+| `name` | Yes | 1-64 chars, lowercase letters/numbers/hyphens; may differ from parent dir |
 | `description` | Yes | Max 1024 chars; primary trigger text |
 | `license` | No | License name or bundled reference |
 | `compatibility` | No | Environment requirements |
@@ -113,7 +113,6 @@ To reuse Claude Code or Codex skill directories, add them to settings:
 
 Pi warns on most issues but still loads the skill when possible:
 
-- Name and parent-directory mismatch
 - Name too long or invalid
 - Description too long
 - Duplicate names, where first discovered wins