pi-orchestrator-config 4.3.8 → 4.3.9

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -10,79 +10,79 @@ A [pi package](https://github.com/badlogic/pi-mono) that implements an **orchest
10
10
 
11
11
  Single extension that provides:
12
12
 
13
- | Feature | Description |
14
- |---------|-------------|
15
- | **Subagent tool** | Delegate tasks to specialist agents (single, parallel, chain, async modes); optional `model` param for model override |
16
- | **`list_models`** | List available models and providers for subagent model override |
17
- | **Async background agents** | Spawn agents in background with `async: true` — results surface automatically when complete. Fullscreen overlay dashboard with live output, keyboard nav, and kill support (`/async-status`, `/async-kill`). On **acpx** parents, optional async is coerced to sync; dream/cron need `internal_operations_provider` + `internal_operations_model`. `cli-*` providers support async natively (see `dev-docs/cli-provider.md`) |
18
- | **CLI / ACPX providers** | Optional `cli_agents` / `acpx_agents` register `cli-*` / `acpx-*` via native `createProvider` (pi ≥ 0.84): `/login cli-<agent>` or `/login acpx-<agent>`, model refresh, filter when unavailable. Missing context window / maxTokens / cost filled from models.dev (`~/.pi/pi-config/models.dev.json`, 1-day refresh). Cold-start default restore (#753): gates `startup`\|`new`, empty `enabledModels`, trusted project merge — see `dev-docs/cli-provider.md`. Headless Cursor `--approve-mcps` when `CLI_APPROVE_MCPS` is set (`false` opts out) or the process is a sidecar (`SIDECAR_PORT`; `startSidecar()` stamps it while running and restores it on `close()`). Executable sidecar consumers should call `bindSidecarListenExit()` so bind/`stopped` fatals exit 1. Headless Gemini defaults `GEMINI_CLI_TRUST_WORKSPACE=true` and preserves an explicit parent value (including `false`). |
19
- | **`/btw` command** | Quick side questions without polluting conversation history — ephemeral overlay |
20
- | **`/async-status` command** | Show status of background agents — select one for live output streaming |
21
- | **`/async-kill` command** | Kill async agents (overlay picker or by name/id) |
22
- | **`ask_user` tool** | Structured user input with options and free-text — used by workflows |
23
- | **Python/pip enforcement** | Auto-fixes `python`/`python3` → `uv run python`; blocks `pip`/`pip3` — requires `uv` |
24
- | **Git protection** | Blocks commits/pushes to main/master, merged branches, `--no-verify`, `git add .` |
25
- | **Remote script exec block** | Blocks `curl \| bash`, `eval $(curl)`, etc. Allows safe `VAR=$(curl ...)` variable assignments |
26
- | **Dangerous command gate** | Confirms `rm -rf`, `sudo`, `mkfs`, etc. |
27
- | **Rule injection** | Injects orchestrator routing rules into system prompt |
28
- | **Git status** | Live git status in status line with colored icons — updates after every tool call. Shows clickable `#N` when the branch has an open PR. Last-activity clock `⏱ HH:MM (Xm/Xh ago)` shows time since last response |
29
- | **Desktop notifications** | Notifies via `notify-send` on task completion, waiting for input, and action required |
30
- | **File preview** | Serves generated HTML/frontend files via HTTP for browser preview from container |
31
- | **Pidash dashboard** | Live web dashboard — multi-session monitoring, browser messaging, model switching, live session name updates, reasoning token display |
32
- | **Pidiff viewer** | Per-project diff viewer with review comments — branch diffs, file tree, inline comments, git-based ignore rules |
33
- | **Dreaming** | Background memory consolidation — extracts memories from sessions, deduplicates, maintains topic-based memory |
34
- | **Memory enforcement** | Code-enforced memory entries — triggers on bash/tool/file events, actions: block, run_after, warn. LLM cannot ignore enforced rules. Dreaming-safe via *(enforced)* marker |
35
- | **Upgrade changelog** | Shows release notes on session start after pi-config version upgrade |
36
- | **Task tracking** | Structured task lists for multi-step workflows — live widget, progress tracking, reminder nudges (owned, based on [@tintinweb/pi-tasks](https://github.com/tintinweb/pi-tasks) MIT) |
37
- | **Neovim integration** | Send changed files and review findings to nvim's quickfix list — only active when running inside nvim |
38
- | **Inter-agent communication** | P2P (`/coms`) agent communication — on-demand activation via slash command |
39
- | **Slash commands** | `/pr-review`, `/issue-review`, `/release`, `/review-local`, `/review-status`, `/query-db`, `/btw`, `/async-status`, `/async-kill`, `/status`, `/dream`, `/remember`, `/coms`, `/pi-config-settings`, `/mcpc connect` — with autocomplete argument hints |
40
- | **GitHub autocomplete** | Type `#` in the editor to get issue/PR suggestions from the current repo — lazy-loaded, 5min cache |
41
- | **Command arg completions** | Tab-complete arguments for slash commands — providers and models for `/external-ai`, branches for `/review-local`, PR numbers for `/pr-review`, and more |
42
- | **Discord bot** | Control pi sessions from your phone via Discord DMs — send prompts, answer ask_user dialogs, switch sessions |
13
+ | Feature | Description |
14
+ | ---------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
15
+ | **Subagent tool** | Delegate tasks to specialist agents (single, parallel, chain, async modes); optional `model` param for model override |
16
+ | **`list_models`** | List available models and providers for subagent model override |
17
+ | **Async background agents** | Spawn agents in background with `async: true` — results surface automatically when complete. Fullscreen overlay dashboard with live output, keyboard nav, and kill support (`/async-status`, `/async-kill`). On **acpx** parents, optional async is coerced to sync; dream/cron need `internal_operations_provider` + `internal_operations_model`. `cli-*` providers support async natively (see `dev-docs/cli-provider.md`) |
18
+ | **CLI / ACPX providers** | Optional `cli_agents` / `acpx_agents` register `cli-*` / `acpx-*` via native `createProvider` (pi ≥ 0.84): `/login cli-<agent>` or `/login acpx-<agent>`, model refresh, filter when unavailable. Missing context window / maxTokens / cost filled from models.dev (`~/.pi/pi-config/models.dev.json`, 1-day refresh). Cold-start default restore (#753): gates `startup`\|`new`, empty `enabledModels`, trusted project merge — see `dev-docs/cli-provider.md`. Headless Cursor `--approve-mcps` when `CLI_APPROVE_MCPS` is set (`false` opts out) or the process is a sidecar (`SIDECAR_PORT`; `startSidecar()` stamps it while running and restores it on `close()`). Executable sidecar consumers should call `bindSidecarListenExit()` so bind/`stopped` fatals exit 1. Headless Gemini defaults `GEMINI_CLI_TRUST_WORKSPACE=true` and preserves an explicit parent value (including `false`). |
19
+ | **`/btw` command** | Quick side questions without polluting conversation history — ephemeral overlay |
20
+ | **`/async-status` command** | Show status of background agents — select one for live output streaming |
21
+ | **`/async-kill` command** | Kill async agents (overlay picker or by name/id) |
22
+ | **`ask_user` tool** | Structured user input with options and free-text — used by workflows |
23
+ | **Python/pip enforcement** | Auto-fixes `python`/`python3` → `uv run python`; blocks `pip`/`pip3` — requires `uv` |
24
+ | **Git protection** | Blocks commits/pushes to main/master, merged branches, `--no-verify`, `git add .` |
25
+ | **Remote script exec block** | Blocks `curl \| bash`, `eval $(curl)`, etc. Allows safe `VAR=$(curl ...)` variable assignments |
26
+ | **Dangerous command gate** | Confirms `rm -rf`, `sudo`, `mkfs`, etc. |
27
+ | **Rule injection** | Injects orchestrator routing rules into system prompt |
28
+ | **Git status** | Live git status in status line with colored icons — updates after every tool call. Shows clickable `#N` when the branch has an open PR. Last-activity clock `⏱ HH:MM (Xm/Xh ago)` shows time since last response |
29
+ | **Desktop notifications** | Notifies via `notify-send` on task completion, waiting for input, and action required |
30
+ | **File preview** | Serves generated HTML/frontend files via HTTP for browser preview from container |
31
+ | **Pidash dashboard** | Live web dashboard — multi-session monitoring, browser messaging, model switching, live session name updates, reasoning token display |
32
+ | **Pidiff viewer** | Per-project diff viewer with review comments — branch diffs, file tree, inline comments, git-based ignore rules |
33
+ | **Dreaming** | Background memory consolidation — extracts memories from sessions, deduplicates, maintains topic-based memory |
34
+ | **Memory enforcement** | Code-enforced memory entries — triggers on bash/tool/file events, actions: block, run_after, warn. LLM cannot ignore enforced rules. Dreaming-safe via _(enforced)_ marker |
35
+ | **Upgrade changelog** | Shows release notes on session start after pi-config version upgrade |
36
+ | **Task tracking** | Structured task lists for multi-step workflows — live widget, progress tracking, reminder nudges (owned, based on [@tintinweb/pi-tasks](https://github.com/tintinweb/pi-tasks) MIT) |
37
+ | **Neovim integration** | Send changed files and review findings to nvim's quickfix list — only active when running inside nvim |
38
+ | **Inter-agent communication** | P2P (`/coms`) agent communication — on-demand activation via slash command |
39
+ | **Slash commands** | `/pr-review`, `/issue-review`, `/release`, `/review-local`, `/review-status`, `/query-db`, `/btw`, `/async-status`, `/async-kill`, `/status`, `/dream`, `/remember`, `/coms`, `/pi-config-settings`, `/mcpc connect` — with autocomplete argument hints |
40
+ | **GitHub autocomplete** | Type `#` in the editor to get issue/PR suggestions from the current repo — lazy-loaded, 5min cache |
41
+ | **Command arg completions** | Tab-complete arguments for slash commands — providers and models for `/external-ai`, branches for `/review-local`, PR numbers for `/pr-review`, and more |
42
+ | **Discord bot** | Control pi sessions from your phone via Discord DMs — send prompts, answer ask_user dialogs, switch sessions |
43
43
 
44
44
  ### Agents (26)
45
45
 
46
- | Category | Agents |
47
- |----------|--------|
48
- | Languages | python-expert, go-expert, ts-expert, java-expert, bash-expert |
49
- | Infrastructure | docker-expert, kubernetes-expert, jenkins-expert |
50
- | Dev workflow | git-expert, github-expert, test-runner, test-automator, debugger |
51
- | Documentation | technical-documentation-writer, api-documenter, docs-fetcher |
52
- | Code review | code-reviewer-quality, code-reviewer-guidelines, code-reviewer-security, code-reviewer-docs, code-reviewer-spec |
53
- | Security | security-auditor |
54
- | Workflow | scout, planner, worker, reviewer |
46
+ | Category | Agents |
47
+ | -------------- | --------------------------------------------------------------------------------------------------------------- |
48
+ | Languages | python-expert, go-expert, ts-expert, java-expert, bash-expert |
49
+ | Infrastructure | docker-expert, kubernetes-expert, jenkins-expert |
50
+ | Dev workflow | git-expert, github-expert, test-runner, test-automator, debugger |
51
+ | Documentation | technical-documentation-writer, api-documenter, docs-fetcher |
52
+ | Code review | code-reviewer-quality, code-reviewer-guidelines, code-reviewer-security, code-reviewer-docs, code-reviewer-spec |
53
+ | Security | security-auditor |
54
+ | Workflow | scout, planner, worker, reviewer |
55
55
 
56
56
  ### Prompt Templates
57
57
 
58
- | Prompt | Description |
59
- |--------|-------------|
60
- | `/implement <task>` | scout → planner → worker |
61
- | `/scout-and-plan <task>` | scout → planner |
62
- | `/implement-and-review <task>` | worker → 5 reviewers → worker |
63
- | `/pr-review [number\|url]` | Fetch PR diff, check past review comments, review with guidelines, post and track comments |
64
- | `/issue-review` | Review a GitHub issue spec and fix it |
65
- | `/release [flags]` | Create GitHub release with changelog and version bumping |
66
- | `/review-local [branch]` | Review local uncommitted or branch changes |
67
- | `/review-handler [url] [--autorabbit] [--autoqodo]` | Process PR review comments, fix approved items |
68
- | `/domain-model [focus area]` | Scan codebase and build/update a CONTEXT.md domain glossary for consistent AI vocabulary |
69
- | `/refine-review <url>` | Refine and improve existing PR review comments |
70
- | `/coderabbit-rate-limit [number\|url]` | Handle CodeRabbit rate limiting on PRs |
71
- | `/create-coms-feature-manager` | Generate a coms feature-manager prompt customized for the current project (source template: `templates/coms-feature-manager-prompt.md`) |
72
- | `/query-db <command>` | Query the review comments database |
73
- | `/external-ai <agent> [--model <model>] [--session-id <id>] [--fix\|--peer\|--resume] <prompt>` | Run prompts via AI CLIs directly (cursor, claude, gemini) — full model access |
74
- | `/external-ai-models-refresh` | Refresh cached AI CLI models for autocomplete |
75
- | `/dream` | Run memory consolidation — extract, deduplicate, maintain topic-based memory |
76
- | `/remember <what>` | Save a memory for future sessions |
77
- | `/dream-auto` | Toggle automatic memory dreaming (every 3h + session end) |
78
- | `/cron add\|list\|list-all\|remove` | Schedule recurring tasks within the pi session (e.g., `/cron add every 2h check for new issues`, `/cron add at 12:00 /review-handler`). `/cron list` and `/cron list-all` open overlay UI (view / remove; list-all = all sessions). Tasks run while pi is active, survive `/reload`, and stop on exit |
79
- | `/async-kill [name\|id\|all]` | Kill async agents (overlay picker or by name/id) |
80
- | `/pi-config-settings [project\|global]` | Interactive TUI settings editor — browse all pi-config settings with scope indicators, smart pickers for providers/models, and immediate save |
81
- | `/status` | Unified session snapshot — async agents, cron tasks, git branch, context usage |
82
- | `/nvim-changed-files` | Send git changed files to nvim's quickfix list (only inside nvim) |
83
- | `/pidiff start\|stop\|restart\|status` | Manage the pidiff diff viewer server (per-project) |
84
- | `/coms start\|stop\|status` | P2P local agent communication (Unix socket) |
85
- | `/mcpc connect` | Connect MCP servers from `~/.pi/pi-config/mcp.json` (`mcpc connect --stdio`). Run after editing that file. |
58
+ | Prompt | Description |
59
+ | ----------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
60
+ | `/implement <task>` | scout → planner → worker |
61
+ | `/scout-and-plan <task>` | scout → planner |
62
+ | `/implement-and-review <task>` | worker → 5 reviewers → worker |
63
+ | `/pr-review [number\|url]` | Fetch PR diff, check past review comments, review with guidelines, post and track comments |
64
+ | `/issue-review` | Review a GitHub issue spec and fix it |
65
+ | `/release [flags]` | Create GitHub release with changelog and version bumping |
66
+ | `/review-local [branch]` | Review local uncommitted or branch changes |
67
+ | `/review-handler [url] [--autorabbit] [--autoqodo]` | Process PR review comments, fix approved items |
68
+ | `/domain-model [focus area]` | Scan codebase and build/update a CONTEXT.md domain glossary for consistent AI vocabulary |
69
+ | `/refine-review <url>` | Refine and improve existing PR review comments |
70
+ | `/coderabbit-rate-limit [number\|url]` | Handle CodeRabbit rate limiting on PRs |
71
+ | `/create-coms-feature-manager` | Generate a coms feature-manager prompt customized for the current project (source template: `templates/coms-feature-manager-prompt.md`) |
72
+ | `/query-db <command>` | Query the review comments database |
73
+ | `/external-ai <agent> [--model <model>] [--session-id <id>] [--fix\|--peer\|--resume] <prompt>` | Run prompts via AI CLIs directly (cursor, claude, gemini) — full model access |
74
+ | `/external-ai-models-refresh` | Refresh cached AI CLI models for autocomplete |
75
+ | `/dream` | Run memory consolidation — extract, deduplicate, maintain topic-based memory |
76
+ | `/remember <what>` | Save a memory for future sessions |
77
+ | `/dream-auto` | Toggle automatic memory dreaming (every 3h + session end) |
78
+ | `/cron add\|list\|list-all\|remove` | Schedule recurring tasks within the pi session (e.g., `/cron add every 2h check for new issues`, `/cron add at 12:00 /review-handler`). `/cron list` and `/cron list-all` open overlay UI (view / remove; list-all = all sessions). Tasks run while pi is active, survive `/reload`, and stop on exit |
79
+ | `/async-kill [name\|id\|all]` | Kill async agents (overlay picker or by name/id) |
80
+ | `/pi-config-settings [project\|global]` | Interactive TUI settings editor — browse all pi-config settings with scope indicators, smart pickers for providers/models, and immediate save |
81
+ | `/status` | Unified session snapshot — async agents, cron tasks, git branch, context usage |
82
+ | `/nvim-changed-files` | Send git changed files to nvim's quickfix list (only inside nvim) |
83
+ | `/pidiff start\|stop\|restart\|status` | Manage the pidiff diff viewer server (per-project) |
84
+ | `/coms start\|stop\|status` | P2P local agent communication (Unix socket) |
85
+ | `/mcpc connect` | Connect MCP servers from `~/.pi/pi-config/mcp.json` (`mcpc connect --stdio`). Run after editing that file. |
86
86
 
87
87
  ### Inter-Agent Communication (coms)
88
88
 
@@ -98,12 +98,12 @@ Two systems for Pi agents to communicate with each other, activated on-demand vi
98
98
 
99
99
  **Tools available once activated:**
100
100
 
101
- | Tool | Description |
102
- |------|-------------|
103
- | `*_list` | List peer agents with names, models, context usage, queue depth |
104
- | `*_send` | Send a prompt to a peer agent |
105
- | `*_get` | Non-blocking poll for a response |
106
- | `*_await` | Block until response arrives |
101
+ | Tool | Description |
102
+ | --------- | --------------------------------------------------------------- |
103
+ | `*_list` | List peer agents with names, models, context usage, queue depth |
104
+ | `*_send` | Send a prompt to a peer agent |
105
+ | `*_get` | Non-blocking poll for a response |
106
+ | `*_await` | Block until response arrives |
107
107
 
108
108
  Forked from [disler/pi-vs-claude-code](https://github.com/disler/pi-vs-claude-code) coms extensions. We own these files — FIFO message queue, structured task delegation via coms protocol.
109
109
 
@@ -216,25 +216,44 @@ Run scout and planner in a chain to analyze the auth module
216
216
 
217
217
  ```bash
218
218
  # Discover available models
219
- list_models(provider="litellm")
219
+ list_models(provider="my-provider")
220
220
 
221
221
  # Run agent with explicit model
222
- subagent(agent="worker", task="...", model="litellm/claude-opus-4-6-1m")
222
+ subagent(agent="worker", task="...", model="my-provider/model-id")
223
223
 
224
224
  # Parallel tasks with per-task models
225
- subagent(tasks=[{agent: "worker", task: "...", model: "litellm/claude-opus-4-6-1m", cwd: "..."}])
225
+ subagent(tasks=[{agent: "worker", task: "...", model: "my-provider/model-id", cwd: "..."}])
226
226
  ```
227
227
 
228
+ ### OpenAI-compatible model discovery
229
+
230
+ Set `discoverModels: true` on a static `openai-completions` provider in `models.json` to load
231
+ its `/v1/models` catalog at interactive session start. Discovery never writes `models.json`. It
232
+ augments the configured provider in Pi's runtime registry, so Pi's ordinary `/model` picker
233
+ displays each result under the exact configured provider key, for example
234
+ `chatgpt-image-latest [litellm]`. The selected model keeps that same provider key and streams
235
+ through the configured provider. Existing static models remain available. Only exact duplicate
236
+ returned IDs are deduplicated.
237
+
238
+ After discovery and runtime registration both succeed, Pi appends a durable transcript entry such
239
+ as `Providers: litellm (242)`. The key and count come from that session's configured provider and
240
+ discovery response. Failed discovery or registration appends no entry, so a prior session's count
241
+ is never presented as current.
242
+
243
+ Each returned string ID remains selectable, including opaque, blank, and whitespace-preserved IDs.
244
+ Discovery materializes a complete Pi `Model` object for every result. It uses source metadata for
245
+ `name`, `reasoning`, `input`, `cost`, `contextWindow`, and `maxTokens` when the response provides
246
+ it. Missing or invalid metadata uses Pi's static-model compatibility defaults: `reasoning: false`,
247
+ zero costs, `contextWindow: 128000`, and `maxTokens: 16384`. Unknown input declares
248
+ `['text', 'image']` so Pi can safely inspect `model.input`. That declaration only satisfies Pi's
249
+ model contract. It does not filter discovery results or claim that the endpoint accepts image input.
250
+
228
251
  ## Code Review Loop
229
252
 
230
- After any code change, the orchestrator runs 6 agents **in parallel** (5 reviewers + test-automator):
253
+ `review_loop_enforcement` controls automatic review-agent dispatch and commit enforcement.
231
254
 
232
- 1. **code-reviewer-quality** — Code quality & maintainability
233
- 2. **code-reviewer-guidelines** — Project guidelines adherence
234
- 3. **code-reviewer-security** — Bugs, logic errors, security
235
- 4. **code-reviewer-docs** — Documentation quality, completeness, accuracy
236
- 5. **code-reviewer-spec** — Code/PR/issue spec alignment
237
- 6. **test-automator** — Runs project tests (pytest, node tests, pre-commit)
255
+ - `true` automatically dispatches 6 agents in parallel (5 reviewers + `test-automator`) after code changes and requires a clean review before commits.
256
+ - `false` disables automatic review-agent and test-agent dispatch. Manual reviews and test runs are optional. No review state, cycle, or result is required before commit.
238
257
 
239
258
  When `review_loop_enforcement` is enabled, the loop stops once all reviewers approve with 0 findings and tests pass
240
259
  (`tests_passed: true` in `pi-config-review-state.jsonl`), OR after `review_loop_max_cycles` total cycles (default `3`,
@@ -275,10 +294,10 @@ See `dev-docs/project-settings.md` for the full settings table.
275
294
 
276
295
  These are set automatically by the orchestrator when spawning reviewer agents:
277
296
 
278
- | Variable | Description |
279
- |----------|-------------|
280
- | `PI_REVIEW_BASE_BRANCH` | Base branch for diff comparison (auto-detected from PR or falls back to main) |
281
- | `PI_HAS_PR` | `true` if a PR exists for the current branch, `false` for pre-push reviews. When `false`, reviewers skip Review History and `code-reviewer-spec` runs a reduced flow (issue-only checks) |
297
+ | Variable | Description |
298
+ | ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
299
+ | `PI_REVIEW_BASE_BRANCH` | Base branch for diff comparison (auto-detected from PR or falls back to main) |
300
+ | `PI_HAS_PR` | `true` if a PR exists for the current branch, `false` for pre-push reviews. When `false`, reviewers skip Review History and `code-reviewer-spec` runs a reduced flow (issue-only checks) |
282
301
 
283
302
  #### Per-Project Resource Management
284
303
 
@@ -300,11 +319,11 @@ Priority: project > user > package (bundled).
300
319
 
301
320
  The orchestrator loads rules from three directories (later layers override same-filename entries):
302
321
 
303
- | Layer | Path | Scope | Number range |
304
- |-------|------|-------|--------------|
305
- | Package | `<pi-config>/rules/` | All users, all projects | `00-69` |
306
- | User | `~/.pi/agent/rules/` | All projects for this user | `70-89` |
307
- | Project | `<project>/.pi/rules/` | Current project only | `90-99` |
322
+ | Layer | Path | Scope | Number range |
323
+ | ------- | ---------------------- | -------------------------- | ------------ |
324
+ | Package | `<pi-config>/rules/` | All users, all projects | `00-69` |
325
+ | User | `~/.pi/agent/rules/` | All projects for this user | `70-89` |
326
+ | Project | `<project>/.pi/rules/` | Current project only | `90-99` |
308
327
 
309
328
  To add a custom rule, create a `.md` file in the appropriate directory:
310
329
 
@@ -363,10 +382,10 @@ The `generate_image` tool creates images from structured descriptions via Gemini
363
382
 
364
383
  Set `image_model` in `pi-config-settings.json` or use `PI_IMAGE_MODEL` env var.
365
384
 
366
- | Setting / Variable | Description |
367
- |----------|-------------|
368
- | `image_model` / `PI_IMAGE_MODEL` | Google/Gemini image model for `generate_image` (Settings TUI filters to google; empty = disabled). |
369
- | `GEMINI_API_KEY` or `GOOGLE_API_KEY` | Gemini API key (env only) |
385
+ | Setting / Variable | Description |
386
+ | ------------------------------------ | -------------------------------------------------------------------------------------------------- |
387
+ | `image_model` / `PI_IMAGE_MODEL` | Google/Gemini image model for `generate_image` (Settings TUI filters to google; empty = disabled). |
388
+ | `GEMINI_API_KEY` or `GOOGLE_API_KEY` | Gemini API key (env only) |
370
389
 
371
390
  **Usage:** Ask naturally — "generate an image of a sunset" — or use structured params: `subject`, `action`, `scene`, `composition`, `lighting`, `style`, `text`, `aspect_ratio`.
372
391
 
@@ -629,64 +648,64 @@ PI_PIDIFF_ENABLE=false pi
629
648
 
630
649
  ### Optional mounts
631
650
 
632
- | Mount | Purpose |
633
- |---|---|
634
- | `-v "$HOME/.agents":"$HOME/.agents":rw` | User-level skills (install/uninstall from container) |
635
- | `-v "$HOME/.config/gcloud/application_default_credentials.json":"$HOME/.config/gcloud/application_default_credentials.json":ro` | Google Cloud ADC (for Claude via Vertex AI) |
636
- | `-v "$HOME/.config/cursor/auth.json":"$HOME/.config/cursor/auth.json":ro` | Cursor CLI auth (for acpx-cursor / cli-cursor models) |
637
- | `-v "$HOME/.claude/credentials.json":"$HOME/.claude/credentials.json":ro` | Claude CLI auth (for acpx-claude / cli-claude models) |
638
- | `-v "$HOME/.config/glab-cli":"$HOME/.config/glab-cli":ro` | GitLab CLI config (auth tokens, host settings) |
639
- | `-v "$HOME/.coderabbit":"$HOME/.coderabbit":rw` | CodeRabbit CLI auth and review data |
640
- | `-v "$HOME/screenshots":"$HOME/screenshots"` | Share screenshots/images with the agent |
641
- | `-v /var/run/docker.sock:/var/run/docker.sock:ro` + `--group-add $(stat -c '%g' /var/run/docker.sock)` | Docker container inspection via `docker-safe` |
642
- | `-v /var/run/podman/podman.sock:/var/run/podman/podman.sock:ro` | Podman container inspection via `docker-safe` |
651
+ | Mount | Purpose |
652
+ | ------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------- |
653
+ | `-v "$HOME/.agents":"$HOME/.agents":rw` | User-level skills (install/uninstall from container) |
654
+ | `-v "$HOME/.config/gcloud/application_default_credentials.json":"$HOME/.config/gcloud/application_default_credentials.json":ro` | Google Cloud ADC (for Claude via Vertex AI) |
655
+ | `-v "$HOME/.config/cursor/auth.json":"$HOME/.config/cursor/auth.json":ro` | Cursor CLI auth (for acpx-cursor / cli-cursor models) |
656
+ | `-v "$HOME/.claude/credentials.json":"$HOME/.claude/credentials.json":ro` | Claude CLI auth (for acpx-claude / cli-claude models) |
657
+ | `-v "$HOME/.config/glab-cli":"$HOME/.config/glab-cli":ro` | GitLab CLI config (auth tokens, host settings) |
658
+ | `-v "$HOME/.coderabbit":"$HOME/.coderabbit":rw` | CodeRabbit CLI auth and review data |
659
+ | `-v "$HOME/screenshots":"$HOME/screenshots"` | Share screenshots/images with the agent |
660
+ | `-v /var/run/docker.sock:/var/run/docker.sock:ro` + `--group-add $(stat -c '%g' /var/run/docker.sock)` | Docker container inspection via `docker-safe` |
661
+ | `-v /var/run/podman/podman.sock:/var/run/podman/podman.sock:ro` | Podman container inspection via `docker-safe` |
643
662
 
644
663
  ### What's in the image
645
664
 
646
- | Tool | Purpose |
647
- |---|---|
648
- | `pi` | Coding agent |
649
- | `git` | Version control |
650
- | `gh` | GitHub CLI (PRs, issues) |
651
- | `glab` | GitLab CLI (MRs, issues, pipelines) — install [gitlab-cli-skills](https://github.com/vince-winkintel/gitlab-cli-skills) for full agent support |
652
- | `gcc` | C compiler — needed for building Python C extensions (e.g. ovirt-engine-sdk-python) |
653
- | `uv` / `uvx` | Python execution (enforced by orchestrator) |
654
- | `go` | Go development and code review |
655
- | `libxml2-dev` | libxml2 headers — needed for building Python C extensions (e.g. ovirt-engine-sdk-python) |
656
- | `mcpc` | MCP CLI (`npm: @apify/mcpc`). Put servers in `~/.pi/pi-config/mcp.json` (already on the `~/.pi` mount). Pi runs `mcpc connect … --stdio` on start. After editing the file, `/mcpc connect`. `entrypoint.sh` reinstalls `@apify/mcpc` on every start so `PI_HOST_USER` host `.npm-global` mapping does not hide the image copy. |
657
- | `myk-pi-tools` | PR review, release, and other CLI utilities |
658
- | `prek` | Pre-commit hook runner |
659
- | `acpx` | Agent proxy for remote models |
660
- | `kubectl` / `oc` | Kubernetes and OpenShift CLI |
661
- | `agent-browser` | Browser automation CLI (navigate, click, screenshot, forms) |
662
- | `procps` | Process utilities (ps, top, pgrep, pkill) |
663
- | `passwd` | `usermod`/`groupmod` — init remaps user `node` to `PI_HOST_UID`/`PI_HOST_GID` |
664
- | `docker` / `podman` | Container CLIs (used via `docker-safe` read-only wrapper) |
665
- | `docker-safe` | Restricted Docker/Podman wrapper — container only (ps, logs, inspect, top, stats) |
666
- | `jq` | JSON processing |
667
- | `rg` (ripgrep) | Fast recursive search |
668
- | `cr` | CodeRabbit CLI — local AI code reviews |
669
- | `curl` | HTTP requests |
665
+ | Tool | Purpose |
666
+ | ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
667
+ | `pi` | Coding agent |
668
+ | `git` | Version control |
669
+ | `gh` | GitHub CLI (PRs, issues) |
670
+ | `glab` | GitLab CLI (MRs, issues, pipelines) — install [gitlab-cli-skills](https://github.com/vince-winkintel/gitlab-cli-skills) for full agent support |
671
+ | `gcc` | C compiler — needed for building Python C extensions (e.g. ovirt-engine-sdk-python) |
672
+ | `uv` / `uvx` | Python execution (enforced by orchestrator) |
673
+ | `go` | Go development and code review |
674
+ | `libxml2-dev` | libxml2 headers — needed for building Python C extensions (e.g. ovirt-engine-sdk-python) |
675
+ | `mcpc` | MCP CLI (`npm: @apify/mcpc`). Put servers in `~/.pi/pi-config/mcp.json` (already on the `~/.pi` mount). Pi runs `mcpc connect … --stdio` on start. After editing the file, `/mcpc connect`. `entrypoint.sh` reinstalls `@apify/mcpc` on every start so `PI_HOST_USER` host `.npm-global` mapping does not hide the image copy. |
676
+ | `myk-pi-tools` | PR review, release, and other CLI utilities |
677
+ | `prek` | Pre-commit hook runner |
678
+ | `acpx` | Agent proxy for remote models |
679
+ | `kubectl` / `oc` | Kubernetes and OpenShift CLI |
680
+ | `agent-browser` | Browser automation CLI (navigate, click, screenshot, forms) |
681
+ | `procps` | Process utilities (ps, top, pgrep, pkill) |
682
+ | `passwd` | `usermod`/`groupmod` — init remaps user `node` to `PI_HOST_UID`/`PI_HOST_GID` |
683
+ | `docker` / `podman` | Container CLIs (used via `docker-safe` read-only wrapper) |
684
+ | `docker-safe` | Restricted Docker/Podman wrapper — container only (ps, logs, inspect, top, stats) |
685
+ | `jq` | JSON processing |
686
+ | `rg` (ripgrep) | Fast recursive search |
687
+ | `cr` | CodeRabbit CLI — local AI code reviews |
688
+ | `curl` | HTTP requests |
670
689
 
671
690
  #### `myk-pi-tools` subcommands
672
691
 
673
- | Command | Description |
674
- |---------|-------------|
675
- | `myk-pi-tools pr diff` | Fetch PR diff and metadata |
676
- | `myk-pi-tools pr info` | Fetch PR information as structured JSON |
677
- | `myk-pi-tools pr post-comment` | Post inline comments to a PR |
678
- | `myk-pi-tools pr store-pr-review` | Store posted PR review comments to `pr-reviews.db` |
679
- | `myk-pi-tools pr update-resolution` | Update resolution status for a previously posted review comment. Used by Phase 1c of `/pr-review` to persist LLM evaluation verdicts |
680
- | `myk-pi-tools pr get-review-history` | Return complete review history for a PR as JSON (posted, skipped, resolved findings). Used by reviewers to avoid re-raising dismissed findings |
681
- | `myk-pi-tools pr get-skipped-comments` | Get previously skipped review comments for a PR |
682
- | `myk-pi-tools pr claude-md` | Fetch `CLAUDE.md` and `AGENTS.md` content for a PR's repository |
683
- | `myk-pi-tools release create` | Create a GitHub release with changelog |
684
- | `myk-pi-tools db` | Review database query commands |
685
- | `myk-pi-tools reviews` | Review handling commands |
686
- | `myk-pi-tools memory` | Project memory commands — persistent per-repo learning |
687
- | `myk-pi-tools ai-cli` | AI CLI commands (cursor, claude, gemini) |
688
- | `myk-pi-tools coderabbit` | CodeRabbit commands (check, trigger, retry) |
689
- | `myk-pi-tools settings get [key ...]` | Resolve settings (project → global → env → default). No args = all keys |
692
+ | Command | Description |
693
+ | -------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- |
694
+ | `myk-pi-tools pr diff` | Fetch PR diff and metadata |
695
+ | `myk-pi-tools pr info` | Fetch PR information as structured JSON |
696
+ | `myk-pi-tools pr post-comment` | Post inline comments to a PR |
697
+ | `myk-pi-tools pr store-pr-review` | Store posted PR review comments to `pr-reviews.db` |
698
+ | `myk-pi-tools pr update-resolution` | Update resolution status for a previously posted review comment. Used by Phase 1c of `/pr-review` to persist LLM evaluation verdicts |
699
+ | `myk-pi-tools pr get-review-history` | Return complete review history for a PR as JSON (posted, skipped, resolved findings). Used by reviewers to avoid re-raising dismissed findings |
700
+ | `myk-pi-tools pr get-skipped-comments` | Get previously skipped review comments for a PR |
701
+ | `myk-pi-tools pr claude-md` | Fetch `CLAUDE.md` and `AGENTS.md` content for a PR's repository |
702
+ | `myk-pi-tools release create` | Create a GitHub release with changelog |
703
+ | `myk-pi-tools db` | Review database query commands |
704
+ | `myk-pi-tools reviews` | Review handling commands |
705
+ | `myk-pi-tools memory` | Project memory commands — persistent per-repo learning |
706
+ | `myk-pi-tools ai-cli` | AI CLI commands (cursor, claude, gemini) |
707
+ | `myk-pi-tools coderabbit` | CodeRabbit commands (check, trigger, retry) |
708
+ | `myk-pi-tools settings get [key ...]` | Resolve settings (project → global → env → default). No args = all keys |
690
709
 
691
710
  ### What's protected
692
711
 
@@ -702,7 +721,7 @@ PI_PIDIFF_ENABLE=false pi
702
721
  **Network** — `--network host` shares the host network stack,
703
722
  so the container can reach any service your host can (LAN, localhost),
704
723
  and the host can reach services started inside the container.
705
- This is required for local MCP servers, LiteLLM proxy, and file preview
724
+ This is required for local MCP servers, OpenAI-compatible proxy, and file preview
706
725
  (agents serve generated HTML/frontend files via HTTP for browser access).
707
726
  If your LLM provider is cloud-based, you don't use local MCPs,
708
727
  and you don't need file preview, you can omit `--network host`.
@@ -785,15 +804,15 @@ Control your pi sessions from your phone via Discord DMs.
785
804
 
786
805
  ### Discord Commands
787
806
 
788
- | Command | Description |
789
- |---------|-------------|
790
- | `!sessions` | List active pi sessions |
791
- | `!watch N` | Watch/switch to session N |
792
- | `!status` | Show current session info |
793
- | `!abort` | Abort current operation |
794
- | `!help` | Show commands |
795
- | Any text | Send as prompt to watched session |
796
- | `/command` | Forward slash command to pi session |
807
+ | Command | Description |
808
+ | ----------- | ----------------------------------- |
809
+ | `!sessions` | List active pi sessions |
810
+ | `!watch N` | Watch/switch to session N |
811
+ | `!status` | Show current session info |
812
+ | `!abort` | Abort current operation |
813
+ | `!help` | Show commands |
814
+ | Any text | Send as prompt to watched session |
815
+ | `/command` | Forward slash command to pi session |
797
816
 
798
817
  ### How It Works
799
818
 
@@ -823,10 +842,10 @@ See [DEVELOPMENT.md](DEVELOPMENT.md) for tips on testing extensions locally, run
823
842
 
824
843
  A standalone HTTP sidecar wrapping the Pi SDK, living under `packages/pi-sidecar/`.
825
844
 
826
- | Package | Language | Registry |
827
- |---------|----------|----------|
828
- | `@myk-org/pi-sidecar` | TypeScript | npm |
829
- | `pi-sidecar-client` | Python | PyPI |
845
+ | Package | Language | Registry |
846
+ | --------------------- | ---------- | -------- |
847
+ | `@myk-org/pi-sidecar` | TypeScript | npm |
848
+ | `pi-sidecar-client` | Python | PyPI |
830
849
 
831
850
  CLI commands: `npx pi-sidecar` (start server), `npx pi-sidecar-start` (background start/stop).
832
851
 
@@ -837,9 +856,9 @@ See [`packages/pi-sidecar/CONSUMER-GUIDE.md`](packages/pi-sidecar/CONSUMER-GUIDE
837
856
 
838
857
  Google Vertex AI Claude provider for Pi, living under `packages/pi-vertex-claude/`.
839
858
 
840
- | Package | Language | Registry |
841
- |---------|----------|----------|
842
- | `@myk-org/pi-vertex-claude` | TypeScript | npm |
859
+ | Package | Language | Registry |
860
+ | --------------------------- | ---------- | -------- |
861
+ | `@myk-org/pi-vertex-claude` | TypeScript | npm |
843
862
 
844
863
  See [`packages/pi-vertex-claude/README.md`](packages/pi-vertex-claude/README.md) for full documentation.
845
864
 
@@ -55,7 +55,36 @@ Before ANY `git commit`, if `review_loop_enforcement` is `true`:
55
55
  ## Protection Rules
56
56
 
57
57
  - NEVER commit or push to main/master branch (unless `allow_push_to_protected_branches` is `true` — see Project Settings)
58
- - NEVER commit to already-merged branches
58
+ - Before every commit on a non-main/master branch, check whether that branch was merged:
59
+
60
+ ```bash
61
+ current_branch=$(git branch --show-current)
62
+ remote_ref="refs/remotes/origin/$current_branch"
63
+ default_ref=$(git symbolic-ref --quiet refs/remotes/origin/HEAD 2>/dev/null || true)
64
+ if ! git show-ref --verify --quiet "$default_ref"; then
65
+ if git show-ref --verify --quiet refs/remotes/origin/main; then
66
+ default_ref=refs/remotes/origin/main
67
+ elif git show-ref --verify --quiet refs/remotes/origin/master; then
68
+ default_ref=refs/remotes/origin/master
69
+ else
70
+ echo "BLOCK: cannot resolve origin's default branch"
71
+ exit 1
72
+ fi
73
+ fi
74
+
75
+ if git show-ref --verify --quiet "$remote_ref" \
76
+ && git merge-base --is-ancestor "$remote_ref" "$default_ref" \
77
+ && [ "$(git rev-parse "$remote_ref")" != "$(git rev-parse "$default_ref")" ]; then
78
+ echo "BLOCK: origin/$current_branch is already merged into ${default_ref#refs/remotes/origin/}"
79
+ exit 1
80
+ fi
81
+ ```
82
+
83
+ Resolve the default branch from `refs/remotes/origin/HEAD`. If that symbolic ref is unavailable
84
+ or invalid, fall back to `refs/remotes/origin/main`, then `refs/remotes/origin/master`; block the
85
+ commit if neither exists. Block only when `refs/remotes/origin/<current-branch>` exists, its tip
86
+ is an ancestor of the resolved default tip, and the two tips are not equal. A fresh pushed branch
87
+ at the default tip is allowed. If the matching remote ref does not exist, continue with the commit.
59
88
  - NEVER use `--no-verify` flag
60
89
  - Branch prefixes: `feature/`, `fix/`, `hotfix/`, `refactor/`
61
90