manage-current-session-habits v0.3.4

manage-current-session-habits v0.3.4

manage-current-session-habits is a Codex skill for scanning the current conversation for user habit phrase candidates and confirming them in place through user-habit-pipeline.

This release turns the skill's install surface into a real dual-entrypoint path: Windows users can keep using PowerShell, while macOS and Linux users now get native shell entrypoints backed by the same shared implementation.

Highlights

• added native POSIX entrypoints for install, uninstall, bootstrap, backend invocation, and smoke validation
• moved install and wrapper logic onto shared Node-based core scripts instead of maintaining separate platform-specific implementations
• kept existing Windows ps1 entrypoints as thin wrappers so current usage does not break
• updated the skill README, lifecycle checklist, release checklist, runbook, and CI workflow to validate the public Windows and POSIX paths
• revalidated package-mode and repo-mode smoke against backend user-habit-pipeline v0.7.6

Why This Release Exists

Before v0.3.4, the repository already claimed cross-platform intent, but the most visible public install path still centered on PowerShell entrypoints.

That left a practical gap:

• Windows users had a direct path
• macOS and Linux users could still benefit from the shared logic, but the public install surface was not shaped around their default shell
• CI validated cross-platform behavior, but the external entrypoints were still more Windows-oriented than they needed to be

This release closes that gap by making the public skill surface match the project's actual cross-platform direction.

Typical Prompts

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

Product Boundary

This skill does:

• scan visible conversation context
• forward requests to the configured user-habit-pipeline backend
• keep explicit user confirmation for durable changes

This skill does not:

• auto-save habits during scan-only flows
• execute downstream workflows
• require users to locate private transcript files on disk

Validation

This release was validated with:

• backend user-habit-pipeline npm run release-check
• skill PowerShell smoke: pwsh -File .\scripts\ci-smoke.ps1 -BackendRepoPath E:\user-habit-pipeline
• skill POSIX smoke: bash ./scripts/ci-smoke.sh --backend-repo-path /e/user-habit-pipeline
• direct Windows wrapper chain validation for install.ps1 -> invoke-backend.ps1 -> uninstall.ps1

Compatibility Note

Current intended backend baseline for this skill release:

• user-habit-pipeline v0.7.6

v0.3.3

manage-current-session-habits v0.3.3

manage-current-session-habits is a Codex skill for scanning the current conversation for user habit phrase candidates and confirming them in place through user-habit-pipeline.

What changed

• Updated the skill release baseline to user-habit-pipeline v0.7.4.
• Aligned the public install and release story to the backend's packaged --host codex starter path.
• Kept the skill install surface focused on the in-thread scan / apply flow instead of transcript-file hunting.
• Revalidated package-mode install preview and installed wrapper smoke against the current backend.

Why it matters

The main value here is not a new skill command. The value is that the skill, the backend package, and the public install story now point at the same current-session integration path.

That makes the skill easier to trust and easier to explain: scan visible thread text, confirm explicitly, then write through the backend.

Typical prompts

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

Related entrypoints

• Backend repo: pingzi-crypto/user-habit-pipeline
• Backend npm package: user-habit-pipeline
• Clean outside-project demo: pingzi-crypto/user-habit-pipeline-codex-demo

Validation

This release was validated with:

• backend user-habit-pipeline npm run release-check
• skill scripts/install-skill.ps1 -CheckOnly
• skill scripts/check-install.ps1 -SmokeTest

manage-current-session-habits v0.3.2

manage-current-session-habits v0.3.2

manage-current-session-habits is a Codex skill for scanning the current conversation for user habit phrase candidates and confirming them in place through user-habit-pipeline.

This release is a small outward-facing refresh rather than a feature rewrite.
It updates the public skill surface so the repository, install path, and release materials better match how the skill is now meant to be discovered and used.

Highlights

• refreshed the public README positioning around Codex, current-session scanning, user habit memory, and in-thread confirmation
• tightened the skill description so the trigger surface is easier for Codex and users to understand
• updated skill-side release checklist and runbook baselines to the current backend user-habit-pipeline v0.7.0 contract
• kept the install flow focused on the normal Codex conversation UI instead of transcript-file hunting or manual backend commands

Why This Release Exists

Since v0.3.1, the biggest change has been public clarity rather than runtime behavior.

The repository now does a better job of explaining:

• what the skill is
• who it is for
• which prompts should trigger it
• what the backend compatibility baseline is today

That makes the skill easier to install, easier to recognize, and easier to recommend without changing its explicit-confirmation model.

Typical Prompts

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

Product Boundary

This skill does:

• scan visible conversation context
• forward requests to the configured user-habit-pipeline backend
• keep explicit user confirmation for durable changes

This skill does not:

• auto-save habits during scan-only flows
• execute downstream workflows
• require users to locate private transcript files on disk

Validation

This release was validated with:

• backend user-habit-pipeline npm run release-check
• skill install.ps1 -CheckOnly
• skill scripts/check-install.ps1 -SmokeTest
• skill repo-override preview through install.ps1 -BackendRepoPath <path> -CheckOnly

Compatibility Note

Current intended backend baseline for this skill release:

• user-habit-pipeline v0.7.0

manage-current-session-habits v0.3.1

manage-current-session-habits is a Codex skill for scanning the current conversation for user habit phrase candidates and confirming them in place through user-habit-pipeline.

It is designed for the lowest-friction flow inside the Codex app:

• scan the current thread
• review candidate phrases
• add / ignore / list / remove habits
• stay inside the normal conversation UI

Best fit

• Codex users who want current-session habit scanning
• AI chat workflows that should stay inside the conversation UI
• in-thread user preference capture without transcript-file hunting
• explicit, reviewable habit confirmation instead of hidden auto-learning

What shipped in v0.3.1

• release-readiness alignment with the backend bridge contract
• stronger install smoke coverage for cached follow-up apply, not only scan and list
• skill-side release checklist and release runbook
• safer packaging for published install and repo-override preview flows

Typical prompts

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

Product boundary

This skill does:

• scan visible conversation context
• forward requests to the configured user-habit-pipeline backend
• keep explicit user confirmation for durable changes

This skill does not:

• auto-save habits during scan-only flows
• execute downstream workflows
• require users to locate private transcript files on disk

Quick start

$repo = Join-Path $HOME ".codex/repos/manage-current-session-habits"
git clone https://github.com/pingzi-crypto/manage-current-session-habits.git $repo
& (Join-Path $repo "install.ps1")

manage-current-session-habits v0.3.0

manage-current-session-habits v0.3.0

This release upgrades the public install surface from a repo-internal script flow to a fuller install lifecycle:

• one-step install from the repo root
• bootstrap install for cold-start setup
• safer clone fallback when GitHub HTTPS clone is blocked
• safe uninstall for generated local skill/runtime state
• a public lifecycle acceptance checklist for install, upgrade, uninstall, and reinstall

Highlights

• added install.ps1 as the main one-step install entrypoint
• added bootstrap-install.ps1 for cold-start setup from outside the repo
• added SSH fallback for bootstrap clone when HTTPS GitHub clone fails but SSH is available
• added uninstall.ps1 to remove the installed skill link and generated local runtime files without deleting the repo checkout
• added INSTALL-LIFECYCLE-CHECKLIST.md as the public acceptance path for install lifecycle validation
• updated README.md to prefer the most reliable clone-based quick start and document uninstall/reinstall paths

Why This Release Exists

v0.2.0 made the skill package-first for the backend install path.

This release focuses on the next practical gap:

• first-time setup should be simpler
• reinstall should not require manual cleanup
• uninstall should be explicit and safe
• release-time acceptance should follow one public checklist instead of scattered ad hoc steps

Validation

This release was validated with:

• package-mode install and smoke through install.ps1
• repo-override install and smoke through install.ps1 -BackendRepoPath <path>
• uninstall preview through uninstall.ps1 -CheckOnly
• uninstall and reinstall round trip through uninstall.ps1 followed by install.ps1
• cold-start clone/install in a temporary directory using the public README quick-start path

manage-current-session-habits v0.2.0

Suggested GitHub Release Title

manage-current-session-habits v0.2.0

Suggested GitHub Release Body

manage-current-session-habits packages the current-thread habit flow into a shareable Codex skill, now with a package-first backend install path.

It is built for users who prefer to stay inside the normal Codex conversation UI instead of hunting for transcript files or stitching together backend commands by hand.

With this release, you can:

• scan the current Codex thread for habit candidates
• review explicit phrase definitions and repeated shorthand
• confirm or ignore candidates with short follow-up prompts such as 添加第1条 and 忽略第1条
• list or remove saved user habit phrases
• stop low-value cleanup quickly with 停
• install the skill against the published user-habit-pipeline npm package by default
• keep using a local backend checkout when you explicitly prefer that path

What is included

• Codex skill entry through SKILL.md and agents/openai.yaml
• portable local install flow through scripts/install-skill.ps1
• install verification and smoke check through scripts/check-install.ps1 -SmokeTest
• package-first backend resolution through the published user-habit-pipeline npm package
• backward-compatible local checkout resolution through -BackendRepoPath
• backend alignment with the current-session bridge contract in user-habit-pipeline v0.4.1
• outward-facing README quick start and embedded demo GIF
• release, publishing, and demo-material templates for future sharing

Why this exists

The goal is simple: let the user stay in the active Codex thread and say things like:

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

instead of manually locating conversation files or invoking backend scripts directly.

Important product boundary

This skill does not auto-save habits during scan-only flows.
It does not execute downstream workflow actions.
It forwards visible current-thread context into the backend, and the user still explicitly confirms durable changes.

Install

& .\scripts\install-skill.ps1
& .\scripts\check-install.ps1 -SmokeTest

Optional local checkout mode:

& .\scripts\install-skill.ps1 -BackendRepoPath <path-to-user-habit-pipeline>

Suggested release assets

• attach the README demo GIF: assets/readme-short-demo.gif
• reuse outward-facing copy from references/publishing-kit.md
• reuse the recording checklist from references/demo-material-requirements.md

Optional closing paragraph

This release turns the skill into a simpler public install surface: package-first by default, still explicit and reviewable, and still aligned to the same current-thread confirmation boundary instead of hidden automation.

manage-current-session-habits v0.1.0

manage-current-session-habits packages the current-thread habit flow into a shareable Codex skill.

It is built for users who prefer to stay inside the normal Codex conversation UI instead of hunting for transcript files or stitching together backend commands by hand.

With this release, you can:

• scan the current Codex thread for habit candidates
• review explicit phrase definitions and repeated shorthand
• confirm or ignore candidates with short follow-up prompts such as 添加第1条 and 忽略第1条
• list or remove saved user habit phrases
• stop low-value cleanup quickly with 停

What is included

• Codex skill entry through SKILL.md and agents/openai.yaml
• portable local install flow through scripts/install-skill.ps1
• install verification and smoke check through scripts/check-install.ps1 -SmokeTest
• backend alignment with the current-session bridge contract in user-habit-pipeline
• outward-facing README quick start and embedded demo GIF
• release, publishing, and demo-material templates for future sharing

Why this exists

The goal is simple: let the user stay in the active Codex thread and say things like:

• 扫描这次会话里的习惯候选
• 添加第1条
• 忽略第1条
• 列出用户习惯短句

instead of manually locating conversation files or invoking backend scripts directly.

Important product boundary

This skill does not auto-save habits during scan-only flows.
It does not execute downstream workflow actions.
It forwards visible current-thread context into the backend, and the user still explicitly confirms durable changes.

Install

& .\scripts\install-skill.ps1 -BackendRepoPath <path-to-user-habit-pipeline>
& .\scripts\check-install.ps1 -SmokeTest

Suggested release assets

• attach the README demo GIF: assets/readme-short-demo.gif
• reuse outward-facing copy from references/publishing-kit.md
• reuse the recording checklist from references/demo-material-requirements.md

Optional closing paragraph

This is the first public packaging pass for the current-session habit flow. It focuses on natural in-thread UX, explicit confirmation, portable install behavior, and lightweight release validation rather than broad workflow automation.