agentleFS
SuShuhengSuShuheng/making-things/AGENTS.md

making-things

AGENTS.md1 starsChanged 4 months agoMITView on GitHub

This repository packages a tool-agnostic Electron workflow plugin. It is not an Electron application and it contains no project-specific workflow records. - A project using this plugin stores its PRD, contract, plans, research, workstreams, work items, handoffs, and capability evidence only in the sibling Git control repository ../.making/<project-slug>/. - Never create .making/ in a project code repository. local.yaml and *.local.yaml belong only in the control repository and remain ignored. - Do not record secrets, raw user prompts, or absolute machine…

The file

We split this AGENTS.md into the rules it gives an agent

6 rules. About 1 is common, 2 are distinctive, and 3 are very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Never create .making/ in a project code repository.

    line 11
  2. Common:

    Do not record secrets, raw user prompts, or absolute machine paths in the control repository.

    line 13
  3. Very specific:

    Read the sidecar PROJECT.md, relevant TRACK.md, TASK.md, and the applicable Skill before changing a workstream record.

    line 22
  4. Very specific:

    Keep renderer code browser-only.

    line 42
  5. Distinctive:

    Verify privileged IPC senders and inputs at runtime.

    line 47
  6. Distinctive:

    Use test-first development for behavior changes and preserve unrelated user changes.

    line 51

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
AGENTS.md

Making Things plugin instructions

This repository packages a tool-agnostic Electron workflow plugin. It is not an Electron application and it contains no project-specific workflow records.

Control-repository boundary

  • A project using this plugin stores its PRD, contract, plans, research, workstreams, work items, handoffs, and capability evidence only in the sibling Git control repository ../.making/<project-slug>/.
  • Never create .making/ in a project code repository. local.yaml and *.local.yaml belong only in the control repository and remain ignored.
  • Do not record secrets, raw user prompts, or absolute machine paths in the control repository.
  • Packaged Skills, templates, native manifests, adapters, and optional hooks remain source assets in this repository; they are not project records.

Prompt-first workstreams

  • A full lifecycle starts only after an explicit user invocation of making-things.
  • Read the sidecar PROJECT.md, relevant TRACK.md, TASK.md, and the applicable Skill before changing a workstream record. Never auto-start a workflow.
  • There are exactly three workstream classes: frontend, backend, and integration. A summary is lowercase YYYY-MM-DD-english-summary.
  • A responsible parent Agent manually creates and enters the code branch {lane}/{date-summary} and .worktrees/{lane}/{date-summary}/. The plugin does not create branches, worktrees, commits, or pushes.
  • Frontend and backend may proceed in parallel only after an approved contract assigns non-overlapping scopes. Integration begins after its declared inputs are complete. TDD subtasks never receive separate worktrees.
  • mts-one is parent-led sequential TDD. mts-more may use one writing native subagent at a time only when the adapter records WRITE_SUBAGENT: VERIFIED; otherwise route to mts-one.
  • handoff and handon pause and resume a whole workstream, not an individual TDD subtask. Hooks are optional notices only.

Electron safety and evidence

  • Keep renderer code browser-only. Expose narrowly scoped preload APIs with contextBridge; never expose raw privileged Electron or Node APIs.
  • Preserve nodeIntegration: false, contextIsolation: true, sandbox: true, and webSecurity: true unless a documented, approved exception exists.
  • Verify privileged IPC senders and inputs at runtime. Do not weaken security or tests to hide a failure.
  • Capability claims are only VERIFIED, DEGRADED, or UNSUPPORTED and require dated smoke evidence. Do not simulate unsupported host behavior.
  • Use test-first development for behavior changes and preserve unrelated user changes. Push, publish, sign, or create a release only with explicit user authorization and fresh evidence.

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.