ghosttrap-cli 0.3.20__tar.gz → 0.3.22__tar.gz

{ghosttrap_cli-0.3.20 → ghosttrap_cli-0.3.22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ghosttrap-cli
-Version: 0.3.20
+Version: 0.3.22
 Summary: Watch for errors streaming from ghosttrap.io
 Project-URL: Homepage, https://github.com/alex-rowley/ghosttrap-cli
 Requires-Python: >=3.10
@@ -52,18 +52,19 @@ If you want to manually flag a caught exception or a non-exception condition, ca
 | `ghosttrap peek --clear` | Skip outstanding errors, then wait for the next one |
 | `ghosttrap last` | Fetch the most recent error and exit (no waiting) |
 | `ghosttrap last --clear` | Fetch the most recent error and skip everything older |
-| `ghosttrap watch` | Stream all errors continuously |
+| `ghosttrap watch` | Deprecated — `peek` reconnects until an error arrives, which covers the streaming case |
 | `ghosttrap list [n]` | Print a numbered summary of the most recent `n` errors (default 10, max 50). Doesn't move the cursor. |
 | `ghosttrap show <i>` | Full details for row `i` from the last `list` (1-based). Doesn't move the cursor. |
 | `ghosttrap clear` | Skip all outstanding errors |
 | `ghosttrap nuke` | Permanently delete every server-side row for the current repo (errors + token). Requires typed confirmation. |
 
-`peek`, `watch`, `last`, and `clear` accept `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise they detect the repo from cwd. `nuke` is intentionally cwd-locked.
+Every command except `setup` and `nuke` accepts `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise the repo is detected from cwd. `nuke` is intentionally cwd-locked.
 
 ## How it works
 
 - **Setup** authenticates with GitHub (via the active `gh` account) to prove you have access to the repo, then saves a repo token locally. If your active `gh` account can't see the repo, setup fails with a clear message; switch with `gh auth switch` and retry.
-- **Peek** and **watch** connect to ghosttrap.io using that token — no GitHub auth needed after setup
+- **Peek** connects to ghosttrap.io using that token — no GitHub auth needed after setup
+- If the connection drops or the server closes an idle socket, peek reconnects with a 60-second backoff for as long as it runs — it only exits once it has delivered an error (or hit a real failure, which it reports on stderr)
 - Errors that arrive while you're offline are replayed on next connect (cursor-based, no duplicates)
 - Repos are tracked by GitHub's immutable repo id, so a rename or transfer doesn't require any action — the next connect picks up the new `owner/name` and your token keeps working
 - Local state is stored in `~/.ghosttrap/config.json`, keyed by GitHub repo id

{ghosttrap_cli-0.3.20 → ghosttrap_cli-0.3.22}/README.md

@@ -43,18 +43,19 @@ If you want to manually flag a caught exception or a non-exception condition, ca
 | `ghosttrap peek --clear` | Skip outstanding errors, then wait for the next one |
 | `ghosttrap last` | Fetch the most recent error and exit (no waiting) |
 | `ghosttrap last --clear` | Fetch the most recent error and skip everything older |
-| `ghosttrap watch` | Stream all errors continuously |
+| `ghosttrap watch` | Deprecated — `peek` reconnects until an error arrives, which covers the streaming case |
 | `ghosttrap list [n]` | Print a numbered summary of the most recent `n` errors (default 10, max 50). Doesn't move the cursor. |
 | `ghosttrap show <i>` | Full details for row `i` from the last `list` (1-based). Doesn't move the cursor. |
 | `ghosttrap clear` | Skip all outstanding errors |
 | `ghosttrap nuke` | Permanently delete every server-side row for the current repo (errors + token). Requires typed confirmation. |
 
-`peek`, `watch`, `last`, and `clear` accept `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise they detect the repo from cwd. `nuke` is intentionally cwd-locked.
+Every command except `setup` and `nuke` accepts `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise the repo is detected from cwd. `nuke` is intentionally cwd-locked.
 
 ## How it works
 
 - **Setup** authenticates with GitHub (via the active `gh` account) to prove you have access to the repo, then saves a repo token locally. If your active `gh` account can't see the repo, setup fails with a clear message; switch with `gh auth switch` and retry.
-- **Peek** and **watch** connect to ghosttrap.io using that token — no GitHub auth needed after setup
+- **Peek** connects to ghosttrap.io using that token — no GitHub auth needed after setup
+- If the connection drops or the server closes an idle socket, peek reconnects with a 60-second backoff for as long as it runs — it only exits once it has delivered an error (or hit a real failure, which it reports on stderr)
 - Errors that arrive while you're offline are replayed on next connect (cursor-based, no duplicates)
 - Repos are tracked by GitHub's immutable repo id, so a rename or transfer doesn't require any action — the next connect picks up the new `owner/name` and your token keeps working
 - Local state is stored in `~/.ghosttrap/config.json`, keyed by GitHub repo id

{ghosttrap_cli-0.3.20 → ghosttrap_cli-0.3.22}/ghosttrap_cli/cli.py

@@ -4,6 +4,7 @@ import argparse
 import asyncio
 import json
 import os
+import signal
 import subprocess
 import sys
 import tempfile
@@ -13,7 +14,20 @@ import urllib.request
 
 import websockets
 
-__version__ = "0.3.20"
+
+def _harden_signals():
+    """Explicitly ignore SIGURG so no process supervisor can nudge peek out
+    with an out-of-band signal. POSIX default is already ignore; this makes
+    it defensive against layers that change the disposition.
+    """
+    s = getattr(signal, "SIGURG", None)
+    if s is not None:
+        try:
+            signal.signal(s, signal.SIG_IGN)
+        except (OSError, ValueError):
+            pass
+
+__version__ = "0.3.22"
 
 GHOSTTRAP_SERVER = "wss://ghosttrap.io/stream/"
 CONFIG_DIR = os.path.expanduser("~/.ghosttrap")
@@ -53,8 +67,9 @@ description: Production error monitoring via ghosttrap.io. Trigger when starting
 # Ghosttrap
 
 Read `~/.ghosttrap/config.json` for state. It contains:
-- `repos`: map keyed by GitHub repo id (stringified int) to `{"github_id": int, "owner": str, "name": str, "token": "t_xxx", "sdk_installed": bool, "sdk_version": str, "init_file": str}`.
+- `repos`: map keyed by GitHub repo id (stringified int) to `{"github_id": int, "owner": str, "name": str, "token": "t_xxx", "sdk_installed": bool, "sdk_version": str, "init_file": str}`. Entries may also carry `recent`, the error ids cached by the last `ghosttrap list` (managed by the CLI — leave it alone).
 - `cursor`: last seen error ID
+- `skill_baseline`: the previous release's skill text, used to 3-way-merge skill updates with local edits. Never edit or delete it.
 
 ## On session start
 
@@ -75,17 +90,19 @@ For caught exceptions or non-exception conditions the user explicitly wants repo
 
 ## Other commands
 
-- `ghosttrap last` — fetch the single most recent error and exit immediately, no waiting. Useful when the user wants to look at the latest error without starting a watch. Add `--clear` to also skip everything older in one shot.
+- `ghosttrap last` — fetch the single most recent error and exit immediately, no waiting. Useful when the user wants to look at the latest error without blocking on a peek. Add `--clear` to also skip everything older in one shot.
 - `ghosttrap list [n]` — print a numbered summary of the most recent `n` errors (default 10, max 50). Does not move the cursor. Caches the ordered ids in config so a follow-up `ghosttrap show <i>` returns full details for that row.
 - `ghosttrap show <i>` — full details for the i-th row from the most recent `ghosttrap list`. Does not move the cursor.
 - `ghosttrap clear` — manually skip outstanding errors without waiting. Useful if the user explicitly wants to drop the queue.
 - `ghosttrap nuke` — permanently delete every server-side row for the current repo (errors + the Repo row + its token). Requires the user to type the repo name `owner/name` to confirm. Only run if the user explicitly asks to wipe server data — never proactively. After it succeeds the token is dead; the user would need to `ghosttrap setup` again to use this repo.
 
+`peek` and every command above except `nuke` accept `--repo owner/name` to target another claimed repo when the cwd isn't inside it (e.g. `ghosttrap list --repo owner/name`).
+
 ## Rules
 
 - Always `run_in_background: true` for peek — it blocks.
 - Don't run multiple peeks at once.
-- If peek exits without output (connection lost), restart it.
+- Peek reconnects by itself (60s backoff) when the connection drops — a quiet peek is waiting, not hung. It only exits after printing an error event, or with a message on stderr if something is actually wrong; restart it only in that second case.
 - After installing/updating the SDK, write the state back to config.json.
 """
 
@@ -416,6 +433,31 @@ async def setup(server_url, token):
         sys.exit(1)
 
 
+# Exception types we knowingly retry on. All represent a transient network/transport
+# blip rather than a semantic error from the server:
+#   - ConnectionClosed: peer closed the WebSocket (either side of the handshake)
+#   - InvalidStatus:    non-101 HTTP response during upgrade (e.g. 502 from a proxy)
+#   - ConnectionError:  builtin — refused, reset, unreachable
+#   - OSError:          DNS failure, transient socket errors (gaierror is a subclass)
+# Anything else escapes and prints a diagnostic line first so we can add it here
+# in the next release. Semantic rejections from the server ({"type": "rejected"})
+# raise SystemExit, which we deliberately do NOT catch — those are real errors.
+_RETRYABLE = (
+    websockets.ConnectionClosed,
+    websockets.InvalidStatus,
+    ConnectionError,
+    OSError,
+)
+
+
+def _log_unexpected(e):
+    print(
+        f"unexpected {type(e).__module__}.{type(e).__name__}: {e} — "
+        f"not currently in the retry list; please report so we can add it.",
+        file=sys.stderr,
+    )
+
+
 async def watch(server_url, token):
     config = _load_config()
     print(f"connecting to {server_url}...", file=sys.stderr)
@@ -424,8 +466,11 @@ async def watch(server_url, token):
         try:
             await _connect_and_handle(server_url, token, config, once=False)
             print("connection closed by server, reconnecting...", file=sys.stderr)
-        except (websockets.ConnectionClosed, websockets.InvalidStatus, ConnectionError, OSError):
+        except _RETRYABLE:
             print("connection lost, reconnecting...", file=sys.stderr)
+        except Exception as e:
+            _log_unexpected(e)
+            raise
         await asyncio.sleep(60)
 
 
@@ -438,8 +483,11 @@ async def peek(server_url, token):
             if got_error:
                 return
             print("connection closed by server, reconnecting...", file=sys.stderr)
-        except (websockets.ConnectionClosed, websockets.InvalidStatus, ConnectionError, OSError):
+        except _RETRYABLE:
             print("connection lost, reconnecting...", file=sys.stderr)
+        except Exception as e:
+            _log_unexpected(e)
+            raise
         await asyncio.sleep(60)
 
 
@@ -676,6 +724,7 @@ def last(do_clear=False, requested=None):
 
 
 def main():
+    _harden_signals()
     parser = argparse.ArgumentParser(prog="ghosttrap", description="Watch for errors from ghosttrap.io")
     sub = parser.add_subparsers(dest="command")
 
@@ -684,7 +733,7 @@ def main():
     clear_parser = sub.add_parser("clear", help="Skip all outstanding errors")
     clear_parser.add_argument("--repo", help="Target repo as owner/name (overrides cwd detection)")
 
-    watch_parser = sub.add_parser("watch", help="Stream errors in real time")
+    watch_parser = sub.add_parser("watch", help="Deprecated: stream errors in real time (use peek)")
     watch_parser.add_argument("--server", default=GHOSTTRAP_SERVER, help="WebSocket server URL")
     watch_parser.add_argument("--repo", help="Target repo as owner/name (overrides cwd detection)")
 
@@ -715,6 +764,11 @@ def main():
     elif args.command == "clear":
         clear(requested=args.repo)
     elif args.command == "watch":
+        print(
+            "warning: 'watch' is deprecated and may be removed in a future release — "
+            "'peek' now reconnects until an error arrives.",
+            file=sys.stderr,
+        )
         _require_setup()
         _refresh_skill_if_stale()
         config = _load_config()

{ghosttrap_cli-0.3.20 → ghosttrap_cli-0.3.22}/ghosttrap_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ghosttrap-cli
-Version: 0.3.20
+Version: 0.3.22
 Summary: Watch for errors streaming from ghosttrap.io
 Project-URL: Homepage, https://github.com/alex-rowley/ghosttrap-cli
 Requires-Python: >=3.10
@@ -52,18 +52,19 @@ If you want to manually flag a caught exception or a non-exception condition, ca
 | `ghosttrap peek --clear` | Skip outstanding errors, then wait for the next one |
 | `ghosttrap last` | Fetch the most recent error and exit (no waiting) |
 | `ghosttrap last --clear` | Fetch the most recent error and skip everything older |
-| `ghosttrap watch` | Stream all errors continuously |
+| `ghosttrap watch` | Deprecated — `peek` reconnects until an error arrives, which covers the streaming case |
 | `ghosttrap list [n]` | Print a numbered summary of the most recent `n` errors (default 10, max 50). Doesn't move the cursor. |
 | `ghosttrap show <i>` | Full details for row `i` from the last `list` (1-based). Doesn't move the cursor. |
 | `ghosttrap clear` | Skip all outstanding errors |
 | `ghosttrap nuke` | Permanently delete every server-side row for the current repo (errors + token). Requires typed confirmation. |
 
-`peek`, `watch`, `last`, and `clear` accept `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise they detect the repo from cwd. `nuke` is intentionally cwd-locked.
+Every command except `setup` and `nuke` accepts `--repo owner/name` to target a specific claimed repo when you're not inside its working tree (e.g. `ghosttrap peek --repo alex-rowley/ghosttrap-cli`). Otherwise the repo is detected from cwd. `nuke` is intentionally cwd-locked.
 
 ## How it works
 
 - **Setup** authenticates with GitHub (via the active `gh` account) to prove you have access to the repo, then saves a repo token locally. If your active `gh` account can't see the repo, setup fails with a clear message; switch with `gh auth switch` and retry.
-- **Peek** and **watch** connect to ghosttrap.io using that token — no GitHub auth needed after setup
+- **Peek** connects to ghosttrap.io using that token — no GitHub auth needed after setup
+- If the connection drops or the server closes an idle socket, peek reconnects with a 60-second backoff for as long as it runs — it only exits once it has delivered an error (or hit a real failure, which it reports on stderr)
 - Errors that arrive while you're offline are replayed on next connect (cursor-based, no duplicates)
 - Repos are tracked by GitHub's immutable repo id, so a rename or transfer doesn't require any action — the next connect picks up the new `owner/name` and your token keeps working
 - Local state is stored in `~/.ghosttrap/config.json`, keyed by GitHub repo id

{ghosttrap_cli-0.3.20 → ghosttrap_cli-0.3.22}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ghosttrap-cli"
-version = "0.3.20"
+version = "0.3.22"
 description = "Watch for errors streaming from ghosttrap.io"
 readme = "README.md"
 requires-python = ">=3.10"