Skip to content

docs: clarify task-agent context boundaries #2152

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
NubsCarson wants to merge 4 commits into from
+7 −3
Closed

docs: clarify task-agent context boundaries #2152

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
51499e7
docs: clarify task-agent context boundaries
NubsCarson May 15, 2026
dbf9f5a
Merge remote-tracking branch 'origin/develop' into nubs/agent-context…
NubsCarson May 19, 2026
308b52e
Merge remote-tracking branch 'origin/develop' into codex/refresh-2152
Dexploarer May 25, 2026
fa14282
docs: refresh task-agent context docs
Dexploarer May 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 7 additions & 3 deletions AGENTS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,8 @@
# Milady — Project Codex Guidance

Milady is a local-first AI assistant built on [elizaOS](https://github.com/elizaOS). It wraps the elizaOS runtime with a CLI, desktop app (Electrobun), web dashboard, and platform connectors (Telegram, Discord, WeChat, etc.). Inherits `~/.codex/AGENTS.md`.
Milady is a local-first AI assistant built on [elizaOS](https://github.com/elizaOS). It wraps the elizaOS runtime with a CLI, desktop app (Electrobun), web dashboard, and platform connectors (Telegram, Discord, WeChat, etc.).

Codex agents read this `AGENTS.md` from the workspace. When spawned by the agent orchestrator, they also receive generated task context with the current room, task, repo routes, and loopback bridge endpoints. Keep stable project rules here; keep deployment/persona policy in the runtime character/config.

Check notice on line 5 in AGENTS.md

View check run for this annotation

Codacy Production / Codacy Static Code Analysis

AGENTS.md#L5 

Overly complex sentence (43 words). Break into shorter instructions.

@codacy-production codacy-production Bot May 25, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codacy has a fix for the issue: Overly complex sentence (43 words). Break into shorter instructions.

Suggested change
Codex agents read this `AGENTS.md` from the workspace. When spawned by the agent orchestrator, they also receive generated task context with the current room, task, repo routes, and loopback bridge endpoints. Keep stable project rules here; keep deployment/persona policy in the runtime character/config.
Split into multiple lines. Config files should read like command lists, not paragraphs.


## Cloud frontend visual review (REQUIRED for any UI change)

Expand Down Expand Up @@ -196,9 +198,11 @@

When Eliza Cloud is enabled or requested, prefer it as the managed backend (app registration, auth, hosted APIs, usage tracking, billing, app domains, creator monetization, Docker container deployments) before inventing custom infrastructure. Cloud monetization (inference markups, purchase-share, redeemable earnings) is a first-class product constraint. If docs disagree with the current schema/UI/API in this repo, the repo wins.

### 2. Claude Code project skills (out-of-band for Codex)
### 2. Claude Code project context (out-of-band for Codex)

`.claude/agents/`, `.claude/commands/`, `.claude/plugins/`, and `.claude/hooks/` are Claude Code project assets for humans or Claude Code sessions. They are not Codex instructions.

`.claude/skills/<name>/SKILL.md` files exist in this repo (currently `phase-review`) but they are **Claude Code slash commands** — they only fire when the user is running Claude Code, not Codex. Codex agents can ignore them. Documented here so you don't mistake them for elizaOS runtime skills if you encounter them in a directory listing.
`.claude/skills/<name>/SKILL.md` files are Claude Code slash commands. Codex agents can ignore them. Codex uses this `AGENTS.md` plus the orchestrator's generated task context; spawned Claude Code sessions use their own Claude project assets.

## Dependencies on elizaOS

Expand Down
Loading