메뉴
Patterns for Claude Code workflow resilience — state persistence, crash recovery, command failure handling, subagent failure handling, and artefact durability. Not about application-level error handling.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | mav-claude-code-recovery |
| description | Patterns for Claude Code workflow resilience — state persistence, crash recovery, command failure handling, subagent failure handling, and artefact durability. Not about application-level error handling. |
| user-invocable | false |
| disable-model-invocation | true |
Patterns for making Claude Code workflows resilient to failures. This covers Claude Code's own recovery mechanisms, not error handling in generated application code.
The maverick-task-progress marker on the GitHub issue is the single
state surface for issue-driven work (see
mav-github-issue-workflow): phase, branch, and artefact
comment ids, written via uv run maverick task-progress set after
each milestone completes — never before. Writing after completion means
the marker always reflects a consistent, completed milestone: if a
failure occurs mid-phase, the marker points at the last completed phase
and the incomplete phase re-executes from scratch.
Local files (worktrees, shortlists, diffs) are derivable caches. When a local artefact and GitHub disagree, GitHub wins.
Post artefacts to GitHub immediately after they are produced, via
uv run maverick issue comment post --kind <design|plan|tasks|completion>.
| Artefact | When |
|---|---|
| Solution design | Immediately after design is approved/produced |
| Implementation plan | Immediately after plan is approved/produced |
| Completion summary | After all steps pass verification |
| Blocking questions | When blocked and unable to resolve |
Batching artefacts to the end means a crash loses all intermediate work, the human has no visibility, and recovery re-does the design/planning.
When starting a session that may be resuming previous work, do not reconstruct the position by hand:
uv run maverick coord resume-point <repo> <issue>
The command reads the task-progress marker, locates the PR via the recorded branch, inspects CI and the review verdict, and returns where to resume with evidence. Then verify the physical prerequisites for that resume point before acting:
| Check | Command |
|---|---|
| Branch exists locally / remotely | git branch --list $BRANCH / git ls-remote --heads origin $BRANCH |
| Expected commits present | git log --oneline -5 |
| Issue still open | gh issue view $ISSUE --json state -q '.state' |
Common reconciliations: branch on remote but not local → fetch and check
out; branch nowhere but marker says branch → re-create from the story
base; issue closed → check whether a PR already merged (work may be
complete). When resuming, briefly report what was recovered and where you
are continuing from.
For epic recovery — DAG, epic-state, claims, in-flight block walks,
worktree reconciliation — follow mav-durability-on-gh
(cold-start hydration order) and mav-block-propagation
(resume an interrupted walk with uv run maverick bprop run before any
other epic work).
When a tool or command fails: read the error output, diagnose the root cause, then fix — applying the canonical retry budget defined in mav-plan-execution (two attempts → systematic debugging → escalate).
--force, --no-verify, or || true to make a failing command succeed. Fix the root cause.When a subagent fails or returns incomplete results:
Scan a project for missing best-practice areas and implement the top recommendation for each gap (linting, unit testing) — installs tools, writes configs, verifies, and commits. Pass 'recommend' to stop after writing recommendations without implementing (replaces the old do-recommend skill).
Implement a focused code change. Use this skill as the wrapper for any implementation work so the Maverick workflow report captures what was done and so the agent applies the project's coding standards before editing. Intended to be invoked once per task from inside a do-issue-* or do-epic phase, not standalone.
Audit a codebase for security risks in one of two modes. In full-audit mode it scans the entire codebase and writes a findings report to docs/security-audit.md (run as part of do-init or on demand). In update mode it reviews only a diff plus the code it could impact, returning a structured findings list as a pre-push gate for do-issue-solo and do-issue-guided. Covers secrets exposure, dependency vulnerabilities, authentication and authorisation patterns, input validation, transport security, and common OWASP risks.
Create, restructure, or update technical documentation. Handles greenfield projects, refactoring non-compliant docs, and incremental updates after code changes.
Work on a multi-story GitHub epic end-to-end. Builds a DAG from the child stories, groups them into waves, runs waves in parallel via per-story worktrees, ejects PRs that fail agent-code-review for human handling, and propagates blocks to downstream stories. Requires git worktrees.
Initialise a project for use with Maverick — verifies the GitHub App, installs the CLI if needed, writes the project config with integration tracking, scaffolds docs, generates project skills, runs an initial cybersecurity audit, then commits the changes and opens a PR.