research-epic-manager / skills
AGENTS.md0 starsChanged 2 months agoMITView on GitHub
This file is the tracker contract for a file-based issue tracker (epics -> issues -> comments). Treat it as the entry point: read it before adding epics/issues/comments or working any issue. This repo's skills and scripts are relocatable: absolute machine-local paths are written as WS_ROOT / EPICS_ROOT prefixes, never hard-coded values, so the same files work on any machine. - WS_ROOT - the workspace/harness root (default /workspace; user-editable). If your layout differs, edit this definition. - EPICS_ROOT - the tracker…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository6 other files it gives its agents.
- add-commentSkill
- add-epicSkill
- add-issueSkill
- issue-managerSkill
- issue-subagent-orchestrationSkill
- research-project-epic-managerSkill
The file
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 1 is distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 14
If your layout differs, edit this definition.
- Very specific: line 62
Working an issue: - follow the issue-manager skill.
≈ “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.
Files
EPICS_ROOT - the issue tracker
This file is the tracker contract for a file-based issue tracker (epics -> issues -> comments). Treat it as the entry point: read it before adding epics/issues/comments or working any issue.
Paths
This repo's skills and scripts are relocatable: absolute machine-local paths
are written as WS_ROOT / EPICS_ROOT prefixes, never hard-coded values, so
the same files work on any machine.
WS_ROOT- the workspace/harness root (default/workspace; user-editable). If your layout differs, edit this definition.EPICS_ROOT- the tracker root where the epics/issues/comments live (default/workspace/epics; user-editable; expected to sit under WS_ROOT but may be anywhere). If your layout differs, edit this definition.
Layout
EPICS_ROOT/
|-- AGENTS.md # this file - the tracker contract
|-- <epic>/
|-- EPIC.md # scope, goals, issues index, auto issue generation
`-- issues/
`-- <issue>/
|-- ISSUE.md # the issue: metadata, description, acceptance criteria
`-- comments/ # history - one file per comment, chronological, newest lastProject skills
The tracker's skills live in the shared harness-level discovery root at
WS_ROOT/.dsh/skills/, which DSH discovers automatically when the current
working directory is under EPICS_ROOT/:
WS_ROOT/.dsh/skills/
|-- add-epic/ # create an epic
|-- add-issue/ # create an issue (draft-first lifecycle)
|-- add-comment/ # append an anonymous chronological comment
|-- issue-manager/ # protocol for working an issue
|-- issue-subagent-orchestration/ # DSH-native sub-agent dispatch mechanics
`-- research-project-epic-manager/ # 5-phase project orchestration (built on these)Workflow
- Epic - a new body of related work starts with
add-epic.EPIC.mddescribes the scope (Goals, Non-Goals, Scope); its Auto Issue Generation section will describe how issues for the epic are generated automatically. Keep the Issues table inEPIC.mdcurrent. - Issue - a unit of work inside an existing epic, via
add-issue. The default is draft-first (author the description, check acceptance criteria, remove placeholders, then promote toopen). For a one-shot creation pass--status opento skip the draft stage; use--bodyto write the full content inline. - Comments - history only, via
add-comment. The## Commentssection insideISSUE.mdis reserved - never write ad-hoc notes there; comments are files incomments/. - Working an issue - follow the
issue-managerskill. Summary:- The chat working the issue is the issue manager.
- Only the issue manager changes issue state (ISSUE.md) and appends comments.
- Only the issue manager dispatches agents.
- Coding/reviewer sub-agents are isolated from issue concepts (no tracker paths, no lifecycle vocabulary) and talk only to the issue manager - results return to the dispatcher.
- Dispatch mechanics: the
issue-subagent-orchestrationskill.
- Lifecycle -
draft -> open -> resolved/closed(withblockedas a side state that returns toopen). Status lives inISSUE.mdMetadata.
Conventions
- Comments are author-anonymous (shared
**Agent**tag - no human or agent names), chronological with newest last, one comment = one point, and reserved for the narrative of solving the work (decisions, trade-offs, blockers) - see.dsh/skills/add-comment/references/comments-best-practices.md. - A coder's self-report is a claim: the issue manager re-runs verification
and gets a reviewer verdict before marking anything
resolved. - Git: the top-level git repository of
WS_ROOTtracks the tracker (EPICS_ROOT) and theWS_ROOT/.dsh/discovery roots. Workspaces (the project directories next toEPICS_ROOT) are gitignored here: each workspace is tracked by its own local-only git repository with no remote.EPICS_ROOTcontains no child git repositories. Commit tracker changes in theWS_ROOTrepo; commit workspace work in the workspace's own repo.
Validation batteries
Each tracker skill carries a stdlib-only validation script in its own
scripts/ directory - add-epic/scripts/test_add_epic.py,
add-issue/scripts/test_add_issue.py, add-comment/scripts/test_add_comment.py,
and issue-manager/scripts/test_lifecycle.py. The first three drive the real
skill scripts against a self-deleting scratch epic under EPICS_ROOT
(test-scratch-* dirs, removed on exit unless --keep); test_lifecycle.py
additionally enforces cross-cutting invariants (discovery roots, git hygiene,
layout invariants over every epic/issue, and the ASCII-only rule for all .md
files). Run all with exit code 0 as the end-to-end sanity check for the
tracker (the canonical result log is kept on the tracker's own test issue).
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
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.

