菜单
Recover an unreachable or crashed standalone Pharo MCP image while preserving logs, image files, changes, repository state, and Epicea history. Use for local MCP restart, replacement-image creation, and image-side change recovery.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Reproduce Pharo smalltalkCI or GitHub Actions failures locally, preferably in Docker; preserve logs; narrow flaky tests; and distinguish project, compatibility, VM/native, and external-service failures.
Reconcile live Pharo, Iceberg, Tonel, and Git state before exporting, reviewing, committing, or pushing image-side changes. Use for dirty or detached Iceberg state, linked worktrees, package-membership mismatches, and image-to-Git delivery.
Load or smoke-test a Pharo project from Tonel, Metacello, or Iceberg and diagnose baseline, branch, source-directory, group, dependency, or version-specific load failures. Use for fresh-image loads and CI-equivalent project loading.
Use PharoCompatibility to load and test a project across supported Pharo versions. Use for missing or renamed Pharo globals, classes, selectors, packages, notices, or semantics; baseline surface selection; and reusable cross-version shims.
基于 SOC 职业分类
| name | pharo-mcp-recovery |
| description | Recover an unreachable or crashed standalone Pharo MCP image while preserving logs, image files, changes, repository state, and Epicea history. Use for local MCP restart, replacement-image creation, and image-side change recovery. |
Use project-managed lifecycle tools when they exist. This workflow is for a standalone image without a safer environment-specific recovery surface.
isRunning, isListening, and
localUrlString; restart the existing MCP instance when possible.pharo-project-load to restore MCP and the target project.pharo-image-git-handoff before repository attachment or Tonel export.Ask before deleting images, resetting repositories, discarding changes, checking out a branch that changes loaded packages, or applying ambiguous history entries.
Report the failure boundary, evidence preserved, image used, restart/recovery actions, recovered behavior, checks run, and unrecovered state.