479a027e4b
Renames: - platform/ → workspace-server/ (Go module path stays as "platform" for external dep compat — will update after plugin module republish) - workspace-template/ → workspace/ Removed (moved to separate repos or deleted): - PLAN.md — internal roadmap (move to private project board) - HANDOFF.md, AGENTS.md — one-time internal session docs - .claude/ — gitignored entirely (local agent config) - infra/cloudflare-worker/ → Molecule-AI/molecule-tenant-proxy - org-templates/molecule-dev/ → standalone template repo - .mcp-eval/ → molecule-mcp-server repo - test-results/ — ephemeral, gitignored Security scrubbing: - Cloudflare account/zone/KV IDs → placeholders - Real EC2 IPs → <EC2_IP> in all docs - CF token prefix, Neon project ID, Fly app names → redacted - Langfuse dev credentials → parameterized - Personal runner username/machine name → generic Community files: - CONTRIBUTING.md — build, test, branch conventions - CODE_OF_CONDUCT.md — Contributor Covenant 2.1 All Dockerfiles, CI workflows, docker-compose, railway.toml, render.yaml, README, CLAUDE.md updated for new directory names. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
52 lines
2.1 KiB
Python
52 lines
2.1 KiB
Python
"""Helpers for the workspace's one-shot initial_prompt.
|
|
|
|
Kept as a standalone module (no heavy imports like uvicorn) so the marker
|
|
logic is unit-testable without standing up the full workspace runtime.
|
|
|
|
Background: the workspace runtime supports an `initial_prompt` that runs once
|
|
on first boot (clone the repo, set git hooks, read CLAUDE.md, commit_memory).
|
|
A marker file `.initial_prompt_done` prevents the prompt from re-running on
|
|
subsequent boots.
|
|
|
|
Prior behaviour wrote the marker AFTER the prompt completed successfully. If
|
|
the prompt crashed mid-execution (e.g. ProcessError from a stale Claude
|
|
session), the marker was never written; every subsequent container boot
|
|
replayed the same failing prompt, cascading into "every message crashes until
|
|
an operator intervenes." See GitHub issue #71.
|
|
|
|
Fix (2026-04-12): write the marker BEFORE firing the prompt. If the prompt
|
|
fails, operators re-send it manually via chat — cheap and available — instead
|
|
of trapping the workspace in a crash loop.
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
import os
|
|
|
|
|
|
def resolve_initial_prompt_marker(config_path: str) -> str:
|
|
"""Return the path where the `.initial_prompt_done` marker should live.
|
|
|
|
Prefers ``<config_path>/.initial_prompt_done`` when the directory is
|
|
writable; falls back to ``/workspace/.initial_prompt_done`` for containers
|
|
where ``/configs`` is read-only.
|
|
"""
|
|
if os.access(config_path, os.W_OK):
|
|
return os.path.join(config_path, ".initial_prompt_done")
|
|
return "/workspace/.initial_prompt_done"
|
|
|
|
|
|
def mark_initial_prompt_attempted(marker_path: str) -> bool:
|
|
"""Write the marker best-effort. Return True on success, False on I/O error.
|
|
|
|
Called BEFORE the initial-prompt self-message is sent. If the attempt
|
|
later fails, the marker is still present — so the next container boot
|
|
does NOT replay the same failing prompt. Operators retry manually via
|
|
the chat interface instead of relying on auto-replay.
|
|
"""
|
|
try:
|
|
with open(marker_path, "w") as f:
|
|
f.write("attempted")
|
|
return True
|
|
except OSError:
|
|
return False
|