agentleFS
AlenroyfeildAlenroyfeild/ai-dev-workspace/llms-full.txt

ai-dev-workspace

llms.txt0 starsChanged 26 days agoMITView on GitHub

Shared Markdown task memory for developers who switch between Claude Code and Codex. Save decisions, next actions and dead ends in files you own; configured local hooks provide a task brief at session start. Install pipx install ai-dev-workspace --pip-args='--pre', then run ws start in a Git repo to preview setup. Python 3.9+ CLI (ws) and local stdio MCP server; MIT licensed. - Local Markdown task records, explicit ws remember, ws checkpoint and ws brief. - Seven core MCP tools by…

  • Commits and pushes

The file

We split this llms-full.txt into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Only saved information is remembered.

    line 11

≈ “Common” is an estimate: at least 1 in 5 of the 3,592 llms.txt files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
llms-full.txt
# AI Dev Workspace — expanded context

> Shared Markdown task memory for developers who switch between Claude Code and Codex. Save decisions, next actions and dead ends in files you own; configured local hooks provide a task brief at session start. Install `pipx install ai-dev-workspace --pip-args='--pre'`, then run `ws start` in a Git repo to preview setup. Python 3.9+ CLI (`ws`) and local stdio MCP server; MIT licensed.

## Stable — published 0.1.0b7

- Local Markdown task records, explicit `ws remember`, `ws checkpoint` and `ws brief`.
- Seven core MCP tools by default; other tools remain callable by name.
- Claude Code and Codex have live checks on macOS. Cursor, Copilot and Gemini
  integrations are fixture-tested; Linux/Windows have CI, not live-client proof.
- Only saved information is remembered; automatic transcript capture is best-effort.
- The optional guards pack is a safety net, not a security boundary.

## Stable workflow — published 0.1.0b7

```sh
pipx install ai-dev-workspace --pip-args='--pre'
cd path/to/your/git-repo
ws start
ws remember "Prices are integer cents"
ws checkpoint --status in_progress --next "Add the rate-limit check"
ws brief
```

`ws start` lists its files and asks before applying repo-local setup. Private
memory lives outside the code repo; generated local wiring is Git-excluded.
Opening the repo in a configured client supplies the task brief. Rerun `ws start`
for setup status; `ws start --undo` removes unchanged files that setup owns and
keeps user edits for review. Do not commit private memory or local client config.

Memory uses no background model calls. A plain handoff file can be sufficient;
historical synthetic results are not a claim of general productivity gains.
See the released Measurements page for later runs and the plain-HANDOFF caveat.

## Preview installation — Claude Code plugin

With the CLI installed and `ws start` completed in the repo, run in Claude Code:

```text
/plugin marketplace add Alenroyfeild/ai-dev-workspace
/plugin install ai-dev-workspace@ai-dev-workspace
```

This is our GitHub marketplace, not Anthropic's official directory. The beta.5
plugin has a live Claude check; setup and duplicate-wiring limits are described
in the [plugin guide](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/ADVANCED.md#claude-code-plugin-preview).
Team sharing is also preview: share only reviewed notes with a private team repo.

## Released in beta.7

- Stable: `ws lesson review [--days N]` lists old lesson date stamps for manual
  review; it is read-only and never changes or deletes saved lessons.
- Preview: Codex SessionStart catch-up recovers changed transcript content for the
  locally claimed task, including resumed turns without a callback; unsafe and
  ambiguous sources are skipped, and it is fixture-tested, not live-tested with
  Codex resume.
- Preview: MCP plugin servers discover a ws-started repo from client roots and stay
  inactive for unlinked or invalid roots; Codex offers no roots.
- Stable: guards parse git push option grammar, including abbreviations of
  --delete, --prune and --mirror; they remain a safety net, not a security boundary.

## Released in beta.6

- Stable: bounded compact MCP recall, read-only stack audit and the advisory behavior pack.
- Preview: opt-in digest advice is size-only; tool output is not compressed automatically.
- Preview: worker fallback runs only declared providers and read-only roles;
  gateway delegation is checked only with a local fake server and sends a redacted brief.
- Stable: lesson promotion drafts advice only; skills packaging is format-checked,
  without a verified third-party installer run.
- Preview: claude-mem import is fixture-tested against a checkpointed SQLite schema;
  newer schemas may not import.
- Preview: Antigravity rules, MCP and SessionStart/PreToolUse hooks are fixture-tested;
  there is no signed-in live proof or automatic capture.
- Preview: Claude Code plugin needs the CLI and `ws start`; not in Anthropic's
  directory, Windows untested. Codex plugin hooks need a trusted manual desktop install;
  no MCP in the plugin, no live hook proof. Cursor plugin is fixture-tested only,
  not submitted, with no verified install command. MCP plugin server roots discovery (Preview) activates only from client roots for already started workspaces and refuses others. [Plugin limits](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/ADVANCED.md#plugin-runtime-packaging-preview).

## Coming next

Common Agent Plugins root manifest (PLG3) and Codex plugin MCP through roots (PLG5)
are parked and unavailable for Codex; codex-cli 0.155.1 offers no roots and cannot expand
plugin MCP paths, so use `ws start` for MCP. Clients that offer roots get the beta.7 Preview.

## Released documentation

- [Share memory between Claude Code and Codex](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/guides/share-memory-claude-code-codex.md)
- [Keep decisions without a background AI](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/guides/keep-decisions-without-background-ai.md)
- [Stop retrying failed approaches](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/guides/stop-retrying-failed-approaches.md)

- [README](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/README.md): install, supported clients and results with limits
- [Setup](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/SETUP.md) and [Tutorial](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/TUTORIAL.md)
- [Commands](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/COMMANDS.md) and [MCP](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/MCP.md)
- [Client verification](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/CLIENT-VERIFICATION.md): fixture and live limits
- [Overview](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/OVERVIEW.md), [Team](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/TEAM-WORKSPACE.md) and [Measurements](https://github.com/Alenroyfeild/ai-dev-workspace/blob/main/docs/MEASUREMENTS.md)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.