SkillAgentSearch skills...

codex-sessions-manager

Use this skill when the user wants to inspect, search, export, verify,

Install / Use

npx skills add 1939869736luosi/codex-sessions-manager

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

71/100

Supported Platforms

OpenAI Codex

Our assessment of codex-sessions-manager

codex-sessions-manager scores 71/100 on our quality scale, 3793rd of 4,602 Development & Engineering skills we index.

Its SKILL.md is 5.5 KB long, well organised into 8 sections with 3 code examples: a solid amount of guidance for an agent.

It has no GitHub stars yet, so there is no community track record; judge it on its content.

Substance
26/30
Structure
18/20
Description
12/15
Adoption
0/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so codex-sessions-manager is actively maintained.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 80/100, with 2 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

codex-sessions-manager compared with similar skills

All 4 of these similar skills score higher than codex-sessions-manager; compare them before choosing.

SkillScoreStarsUpdatedFormat
codex-sessions-manager (this skill)by 1939869736luosi7102mo agoSKILL.md
Agent-Reachby Panniantong10097.0k5d agoCLAUDE.md
ai-job-searchby MadsLorentzen10046.6k5d agoCLAUDE.md
claude-howtoby luongnv8910041.9k3d agoCLAUDE.md
algorithmic-artby anthropics100177.9k21d agoSKILL.md

Frequently asked questions

How do I install codex-sessions-manager?
Run npx skills add 1939869736luosi/codex-sessions-manager. The install tabs above show the steps for each supported agent.
Which AI agents does codex-sessions-manager work with?
It is written for OpenAI Codex, as a SKILL.md file. Other agents that read the same format can often use it too.
Is codex-sessions-manager safe to use?
It declares no license and scores 80/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is codex-sessions-manager still maintained?
The repository was last updated about 2 months ago, so codex-sessions-manager is actively maintained.

name: codex-sessions-manager description: Use this skill when the user wants to inspect, search, export, verify, clean up, delete, restore, or purge local Codex sessions stored under ~/.codex. metadata: source: https://github.com/1939869736luosi/codex-sessions-manager

Codex Sessions Manager

Overview

CLI-first local Codex history audit and recovery tool. Use official Codex for normal task management and permanent deletion. Use this tool to verify what remains under ~/.codex, inspect legacy or damaged state, and perform previewable recoverable cleanup for confirmed residual IDs.

This is not a UI product, TUI, detail page, incremental scanner, or automatic cleanup service.

Setup

npm install -g codex-sessions-manager
codex-sessions --version

The default Codex root is ~/.codex. Use --root <path> for a different root.

For official Codex Skill discovery, place this Skill directory at either:

  • project scope: .agents/skills/codex-sessions-manager
  • user scope: $HOME/.agents/skills/codex-sessions-manager

The distributed Skill includes agents/openai.yaml inside the Skill directory.

When To Use

  • List/filter/search sessions (by project, time, status, source)
  • Show session detail or timeline
  • Inspect session family (parent/child/subagent/side/fork)
  • Audit and verify what remains locally after official deletion
  • Scan root for likely residue candidates
  • Preview batch delete scope without executing
  • Build explicit-ID delete plans
  • Export session backup
  • Delete sessions (permanent or recoverable trash)
  • Restore or purge trash entries
  • Verify post-delete cleanliness
  • Clean stale JSONL indexes

CLI Quick Reference

codex-sessions doctor [--json] [--details]
codex-sessions list [--limit N] [--project TEXT] [--status S] [--source-kind K]
codex-sessions sources [--json]
codex-sessions projects
codex-sessions show <id> [--json]
codex-sessions events <exact-id> [--output FILE]
codex-sessions family <id> [--children|--parents|--subagents|--impact|--full]
codex-sessions audit <id> [--json]
codex-sessions audit-root [--limit 50] [--status S...] [--source S...]
codex-sessions preview-root [--limit 50] [--status S...] [--source S...]
codex-sessions monthly-review [--limit 50] [--status S...] [--source S...] [--details]
codex-sessions export <id> [--output ./backup.json]
codex-sessions plan-delete <id...> [--include-children] [--include-family] [--json]
codex-sessions plan-delete --source-kind K --limit N [--json]
codex-sessions preview-plan <file> [--json]
codex-sessions delete <full-uuid...> [--trash] [--yes] [--allow-active]
codex-sessions trash-list
codex-sessions restore <exact-trash-id> --yes
codex-sessions purge <exact-trash-id> --yes
codex-sessions cleanup-index <full-uuid...> [--yes] [--allow-active]
codex-sessions cleanup-stale [--yes]
codex-sessions verify <id...>
codex-sessions recovery-status [--json]
codex-sessions recover <exact-operation-id> --yes

Safety Rules

  • All destructive commands require --yes (CLI) or confirm=true (MCP); without it you get preview only.
  • Always run a separate preview before any destructive action, then require explicit user confirmation.
  • preview, plan-delete, family, impact, audit-root, preview-root, monthly-review are read-only. They are never deletion authorization.
  • Permanent delete removes only exact thread-linked logs. Trash retains them; restore leaves them unchanged; final purge removes them unless that ID is live again. Memory is always retained.
  • Delete never recursively adds parent, child, or family sessions. List every intended ID explicitly.
  • Allowlisted exact-key global-state refs are removable only through explicit-ID confirmed delete.
  • Unknown global-state refs are warnings only; do not delete them automatically.
  • Confirmed session mutations require full canonical UUIDs. Active-session deletion also requires --allow-active / allowActive=true.
  • An interrupted write keeps an exclusive lock and durable recovery record. Inspect it before running the exact confirmed recovery operation.
  • A committed operation with partial or failed verification remains committed and must be reported that way.

MCP (Optional, Advanced)

codex-sessions-mcp                       # default: read-only profile (16 bounded tools)
codex-sessions-mcp --profile admin       # all 22 tools including destructive ops

Invalid --profile values exit with code 1. Default is read-only for safety.

get_session defaults to detail=compact (20 items / 64 KiB / 1 MiB source read). detail=full is still bounded (200 items / 256 KiB / 8 MiB source read). Session metadata is bounded too. Both return explicit completeness metadata; sourceCompleteness preserves parse/unsupported status when the MCP envelope is also truncated_limit, and itemsKnown=null means the reader stopped before EOF. list_sessions defaults to 50 concise records, accepts at most 200, and caps the response at 256 KiB. Use CLI JSON for complete local result sets and export for a JSON recovery bundle whose embedded text or base64 data can reconstruct source-file bytes.

All other MCP structured responses also have a final 256 KiB / 200-items-per-collection cap. Explicit session operations accept at most 50 IDs; list_trash defaults to 50 entries. Limit hits return responseCompleteness and responseOmittedReason. Exact backup export is CLI-only.

Detailed References

For full CLI/MCP parameter reference, safety model, and advanced rules:

Related Skills

View on GitHub
GitHub Stars0
CategoryDevelopment
Updated2mo ago
Forks0

Trust signals

80/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 medium1 low