continue-claude-code-work — Claude Code skill: resume work
continue-claude-code-work restores interrupted Claude Code work from session history, identifies unfinished requests, and selects the next verified action.
MIT 216 installs ★ 1,540 updated 2026-10-25
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a claude-codeWhat continue-claude-code-work does
The skill reads the selected session through daymade-claude-code:read-claude-code-history and asks for a chronological evidence brief. It checks the session identity, project, covered time, compaction boundaries, queued requests, unresolved calls, touched files, and known gaps.
It then compares that record with the current project. The workflow checks AGENTS.md or CLAUDE.md, the working directory, branch, working tree, and the actual result of earlier writes or commands. From there it chooses a direct action tied to the unfinished outcome and verifies what happened. It does not run claude --resume or claude --continue, and it leaves unrelated working-tree changes alone.
A Session ID is needed, unless the history search can locate it. This skill is for Claude Code sessions; Codex sessions require continue-codex-work.
What it helps with
- recover the goal of an interrupted session
- identify requests that remain unfinished
- verify whether earlier changes actually landed
- choose the next action from session evidence
- continue work without
claude --resume
Example prompts for the agent
- Continue Claude Code work from Session ID
[session-id], read the history first, and complete the unfinished task. - Pick up the interrupted Claude Code session, check the current project state, and avoid repeating completed changes.
- The session was compacted. Find its history, recover the original goal, and take the next direct action.
How to install continue-claude-code-work
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a claude-code2Manually: put the files in place
~/.claude/skills/continue-claude-code-work/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/continue-claude-code-work/
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a codex2Manually: put the files in place
.agents/skills/continue-claude-code-work/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a cursor2Manually: put the files in place
.cursor/skills/continue-claude-code-work/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install daymade/claude-code-skills daymade-claude-code/continue-claude-code-work/SKILL.md2Manually: put the files in place
.github/skills/continue-claude-code-work/SKILL.md1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a gemini-cli2Platform CLI
gemini skills install https://github.com/daymade/claude-code-skills --path daymade-claude-code/continue-claude-code-workManually: put the files in place
.devin/skills/continue-claude-code-work/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about continue-claude-code-work
How do I install continue-claude-code-work in Claude Code?
Run npx skills add daymade/claude-code-skills --skill continue-claude-code-work -a claude-code, then check ~/.claude/skills/continue-claude-code-work/SKILL.md.
What does continue-claude-code-work do?
It reconstructs unfinished Claude Code work from session history, compares that record with the current project, and carries out the next verified action.
Does it work with Codex sessions?
No. This skill targets Claude Code sessions; Codex sessions use continue-codex-work.
Source text
SKILL.md · MIT
--- name: continue-claude-code-work description: >- Continues interrupted Claude Code work after local history is read and reconciled — reconstructs the original outcome, unfulfilled requests, and the next action. Use when given a Claude Code Session ID, asked to pick up prior Claude work, told a run was interrupted or compacted, or asked to take over without `claude --resume`. Not for Codex sessions (use continue-codex-work). argument-hint: "[session-id]" --- # Continue Claude Code Work Continue means execute the still-unfinished business task. It does not mean summarize the transcript, perfect the history tool, or start a new experiment. ## Step 1: Obtain a verified read receipt Invoke `daymade-claude-code:read-claude-code-history` for the exact Session. Ask that Skill for a chronological evidence briefing, not separate “last user” and “last assistant” buckets. If the Session ID is unknown, use its search/inventory route first. Do not act until the receipt states: - exact Session ID and project; - sources and time coverage; - compaction and raw-chronology boundaries; - every retained human request, including queued prompts; - end reason, unresolved calls, files touched, and explicit gaps. If a clipped section could contain the objective or a correction, re-read it with the reader's full mode. If identity, ancestry, or authorship is unresolved, stop; continuing from a guessed Session is worse than asking for the missing evidence. ## Step 2: Rebuild the continuation contract Before the first project-changing tool call, write this compact internal contract from evidence. Do not invent a value to fill a blank. | Field | Evidence required | |---|---| | Original business outcome | Earliest still-governing human request, not a later implementation detail | | Current explicit request | Latest human request that is not merely “继续” | | Already completed | Independent current-state verification, not the old Agent's claim | | Still unfulfilled | Requested result with no completion evidence | | User corrections / do-not-repeat | Human messages that rejected a route, output, or assumption | | Proven assets and successful routes | Existing code, docs, Skills, outputs, or commands that previously worked | | Next direct action | One action whose success visibly reduces the unfulfilled business result | When the latest message is only “继续,” inherit the objective from the verified chronology. Never treat the cue itself as the specification. ## Step 3: Reconcile with current reality 1. Read the target project's current `AGENTS.md`/`CLAUDE.md` and authoritative project state; historical instructions may be stale. 2. Confirm the current cwd, branch, working tree, and relevant files. 3. Verify whether prior writes, commits, external operations, or background jobs actually landed. A transcript records attempts as well as successes. 4. Retrieve the proven asset or prior successful path before authoring a replacement. 5. If another process may still complete the same transiently failed work, verify its current result before duplicating it. ## Step 4: Execute the business task Take the next direct action from the contract and verify its result. Skill edits, review expansion, documentation systems, and infrastructure cleanup remain subordinate unless the user explicitly made one of them the business outcome or they are strictly required to unblock it. Before changing direction, ask mechanically: “If this succeeds, which original unfulfilled result becomes smaller?” If the answer is none, record it as a later proposal and return to the task. Do not run `claude --resume` or `claude --continue`. Do not overwrite unrelated working-tree changes. ## Step 5: Report against the contract - **Recovered goal**: Session and original business outcome. - **Executed**: actions and independent verification. - **Now complete**: contract rows that are actually closed. - **Remaining**: unfulfilled rows and exact blockers. - **Not repeated**: prior rejected routes deliberately avoided when relevant. Do not say “continued successfully” when only context extraction or a process step completed; the business result is the completion unit.
© daymade · MIT · daymade/claude-code-skills