continue-codex-work — Claude Code skill: resuming Codex sessions
continue-codex-work is a Claude Code skill that resumes interrupted Codex sessions by Session ID. It activates when a run stalled, compacted, or transferred to another agent context.
MIT 344 installs ★ 1,540 updated 2026-11-22
npx skills add daymade/claude-code-skills --skill continue-codex-work -a claude-codeWhat continue-codex-work does
The tool recovers incomplete tasks from an interrupted Codex run and maps out the next direct action. It stays inactive if Codex has already natively resumed the same conversation.
First, it calls read-codex-history to verify the exact Session ID, fork lineage, and byte boundaries. The skill immediately fails closed on mismatched rollouts or broken snapshots without guessing files.
Next, it rebuilds a task contract tracking remaining deliverables and user corrections. Before modifying files, scripts/start_work.py checks the Git state and coordinates through the peer-message skill.
What it helps with
- resuming Codex sessions
- verify session rollout identity and fork lineage
- isolate unfulfilled business requirements
- reconcile in-flight jobs and Git workspace state
Example prompts for the agent
- Pick up interrupted Codex session 0192a83b-4c5d-7e8f-9012-3456789abcde and finish the task.
- Take over the previous Codex run that got stuck during compaction.
How to install continue-codex-work
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-codex-work -a claude-code2Manually: put the files in place
~/.claude/skills/continue-codex-work/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/continue-codex-work/
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-codex-work -a codex2Manually: put the files in place
.agents/skills/continue-codex-work/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-codex-work -a cursor2Manually: put the files in place
.cursor/skills/continue-codex-work/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install daymade/claude-code-skills daymade-claude-code/continue-codex-work/SKILL.md2Manually: put the files in place
.github/skills/continue-codex-work/SKILL.md1Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-codex-work -a gemini-cli2Platform CLI
gemini skills install https://github.com/daymade/claude-code-skills --path daymade-claude-code/continue-codex-workManually: put the files in place
.devin/skills/continue-codex-work/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add daymade/claude-code-skills --skill continue-codex-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-codex-work
How to install continue-codex-work in Claude Code?
Run npx skills add daymade/claude-code-skills --skill continue-codex-work -a claude-code.
What does continue-codex-work do?
It restores the context and original goal of an interrupted Codex run using its Session ID. The skill outlines the remaining tasks and prevents duplicating already finished work.
What is required for continue-codex-work to work?
It requires the read-codex-history skill for log verification, peer-message for task coordination, and Python 3 to execute scripts/start_work.py.
Source text
SKILL.md · MIT
---
name: continue-codex-work
description: >-
Continues interrupted Codex work only after read-codex-history verifies rollout identity and
fork/compaction lineage — restores the original outcome, unfulfilled requests, and the next
action. Use when given a Codex Session ID, asked to pick up a prior Codex run, told Codex was
interrupted, fused, compacted or stuck, or asked to take over without `codex resume`. Not for a
conversation Codex already natively resumed.
argument-hint: "[session-id]"
---
# Continue Codex Work
Continue means finish the still-unfulfilled business task. Reading history is a
mandatory evidence phase, not the deliverable.
## Entry gate: external takeover, not native resume
| Current situation | Action |
|---|---|
| Codex itself natively resumed this same conversation, and the restored turns or compaction state are already in the current context | Do not invoke this Skill or `read-codex-history`. Continue directly from the retained context, then verify current workspace and external state before changing anything. |
| A new/different Agent context must take over an earlier Codex rollout, or the user identifies another Session ID whose continuity is not already present here | Use this Skill. Step 1's verified history receipt remains mandatory. |
A restart is not enough to choose the second row. The deciding fact is whether
the current conversation is the host-restored continuation of the same Session,
not whether a Codex process was restarted.
## Step 1: Obtain a verified Codex read receipt
Invoke `daymade-claude-code:read-codex-history` for the exact Session ID and request
the full chronological Session evidence briefing. If the ID is unknown, use that
Skill's inventory or bounded search first.
The receipt must state:
- prompt-ledger, state-index, and rollout sources used;
- selected `session_meta.id` and whether it exactly matches the requested ID;
- root-to-child fork lineage and exact inherited byte boundaries;
- selected and inherited chronological user/assistant turns;
- compacted context, latest plan, end reason, unresolved calls, files, and gaps.
Fail closed on a mismatched rollout, missing child rollout, ambiguous physical copy,
missing parent, broken byte boundary, or continuation cue with no recoverable parent
objective. Never continue from the “closest” file.
**Isolated or remote `CODEX_HOME` staging**: When taking over a Codex session staged to
a separate directory (for example on a remote worker or an isolated handover bundle),
pass `CODEX_HOME=<staged-codex-home>` to `read-codex-history`
(`find_codex_session.py <session-id> --format json` and
`read_codex_session.py --session <session-id> --full`) rather than scanning the host's
default `~/.codex`. Require `status: "unique"` and `identity_verified: true` from the
staged index before proceeding.
## Step 2: Rebuild the continuation contract
Before choosing the next action or reporting a stage complete, fill this internal
contract from the verified read receipt and the project's current stage source.
Use the latest human request when no project stage source exists.
| Field | Evidence required |
|---|---|
| Original business outcome | Earliest still-governing human request across inherited and selected timelines |
| Current explicit request | Latest human request that is not only a continuation cue |
| Current delivery stage / non-goals | Latest human correction narrowing what to deliver now and what to defer; do not promote the original long-term ambition back above it |
| Already completed | Independent completion evidence, not Agent narration or an assistant final. Bind it to artifact/session identity and input conditions; reuse it while those conditions are unchanged. Read volatile runtime and delivery state fresh. |
| Still unfulfilled | Requested result without completion evidence |
| User corrections / do-not-repeat | Human messages rejecting a route, assumption, or output |
| Proven assets and successful routes | Existing code, documents, Skills, outputs, commands, and prior successful experiments |
| Next direct action | One action whose success visibly reduces the unfulfilled business result |
If the child contains only “继续,” “continue,” or `/fork`, inherit the task from the
verified parent snapshot. A local cue is not a standalone goal.
## Step 3: Reconcile with current reality
1. Read the current target project's `AGENTS.md`/`CLAUDE.md`.
2. Confirm cwd, branch, working tree, relevant files, and external state. When continuing
on a remote host or a fresh pool checkout, inspect `git status` and `git rev-parse HEAD`
directly rather than assuming uncommitted working-tree state from the source machine
exists in the target checkout.
3. Before independent takeover edits, reruns or dispatch, execute this Skill's
`scripts/start_work.py` against the loaded `peer-message` Skill. Retain the
recovered stable task key and owned paths across Forks, but obtain the actor
from the current host; never adopt a parent/history/context-pack ID. Native
parent/worker work retains the host task tree and skips independent registration.
```bash
python3 '<loaded-continue-codex-work>/scripts/start_work.py' \
--peer-skill '<absolute-loaded-peer-message>' -- \
--repo '<absolute-target-checkout>' --path '<owned-repo-relative-file-or-directory>' \
--task '<stable-issue-or-deliverable>' --actor '<exact-current-host-address>'
```
The wrapper runs the existing `peer.py coord start` and preserves its result
and exit code. On exit 3, use its existing owner/result lead before starting
replacement work. An empty registry retains unknown occupancy. Resolve a
missing dependency through the existing Skill owner; do not create a second
board. Follow peer-message's current work-entry/reconciliation contract for
non-Git resources, changed identities and released declarations. This checks
the invoked entry; it does not enforce all host file writes.
4. Verify that recorded patches, commits, pushes, downloads, jobs, or Agent outputs
actually landed; rollouts record attempted actions too.
5. Retrieve and reuse prior successful assets before creating a replacement.
6. For transient failures such as usage limits or service errors, verify whether the
original process later resumed and finished before duplicating work.
For an existing native Goal, use [Native Goal reconciliation](references/native-goal-reconciliation.md)
to distinguish unfinished business, runtime status, supported recovery and observed
automatic continuation. An absent recovery tool in one tool menu is not proof that
the live host has no recovery control.
7. Check what was **in flight** when the session stopped. A long task may survive
its controller; a start record proves neither completion nor process death.
Read its existing receipt/output and verify the current process before deciding
whether to monitor, finish archival or rerun. If the process is gone and the
result is unconfirmed, give an authorized rerun a detached
result-to-disk channel (e.g. remote `( cmd > log 2>&1 </dev/null; echo $? > rc ) &`
polled through the rc file — prefer `setsid`/`nohup` where the target has them,
since a pty-backed session can still SIGHUP a bare subshell) so the next
interruption cannot take the result again.
**In-flight CI example:** An existing `result.json` names commit `H`, owner `O`
and a running test phase; its lifecycle marker records PID `P`. Before starting
another CI, match `P`'s live argv to `H` and the exact output path, match its cwd
and child test phase to that receipt, and observe log growth or a child completing
the documented timeout fixture. A matching live operation should be monitored
through its original result, archive and cleanup, keeping its owner and accumulated
budget. A vanished controller or an old receipt mtime does not authorize a second
run; PID existence alone does not establish identity. Reconcile any final-head
delta separately and rerun only the affected checks. This process check follows
Step 1 for external takeover; native continuation still uses the entry gate above.
## Step 4: Execute the business task
Perform the next direct action and close its feedback loop. Reviews, Skill work,
infrastructure cleanup, format polish, and extra safety machinery remain subordinate
unless they directly unblock the original result or the user explicitly promoted them.
Before every material branch, ask: “If this succeeds, which unfulfilled result in
the current authorized stage becomes smaller?” No answer means the branch is not
continuation work. A blocked source, login or provider route blocks only dependent
work; execute other necessary authorized work before declaring the whole task blocked.
Verify the result at the user's consumption surface: searchable bytes alone do not
prove a readable page, and generated suggestions do not prove saved human decisions.
Reuse passed checks for unchanged inputs; expand verification only for new changes,
failures or unresolved evidence gaps.
Do not run `codex resume` or `codex --continue`. Do not overwrite unrelated changes.
## Step 5: Report against the contract
- **Recovered goal**: exact Session lineage and original business outcome.
- **Executed**: actions and independent verification.
- **Now complete**: contract rows actually closed.
- **Remaining**: unresolved rows and exact blockers.
- **Not repeated**: relevant rejected routes deliberately avoided.
Context recovery, parser success, a green review, or a finished subprocess is not by
itself successful continuation. The user's business result is the completion unit.
© daymade · MIT · daymade/claude-code-skills