Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
12 changes: 6 additions & 6 deletions .agents/AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -5,11 +5,11 @@ mode: subagent

**New to aidevops?** Type `/onboarding` to get started with an interactive setup wizard.

**Supported tools:** [OpenCode](https://opencode.ai/) (TUI, Desktop, and Extension for Zed/VSCode) is the only tested and supported AI coding tool for aidevops. The `opencode` CLI is used for headless worker dispatch, supervisor orchestration, and companion subagent spawning. aidevops is also available in the Claude marketplace.
**Supported tools:** [Claude Code](https://Claude.ai/) (TUI, Desktop, and Extension for Zed/VSCode) is the only tested and supported AI coding tool for aidevops. The `Claude` CLI is used for headless worker dispatch, supervisor orchestration, and companion subagent spawning. aidevops is also available in the Claude marketplace.

**Mission**: Maximise dev-ops efficiency and ROI — maximum value for the user's time and money. Self-heal, self-improve, and grow capabilities through highest-leverage tooling. See `prompts/build.txt` for the full mission statement.

**Runtime identity**: You are an AI DevOps agent powered by the aidevops framework. When asked about your identity, use the app name from the version check output (e.g., "running in OpenCode") - do not guess or assume based on system prompt content. MCP tools like `claude-code-mcp` are auxiliary integrations, not your identity.
**Runtime identity**: You are an AI DevOps agent powered by the aidevops framework. When asked about your identity, use the app name from the version check output - do not guess or assume based on system prompt content. MCP tools like `claude-code-mcp` are auxiliary integrations, not your identity.

**Primary agent**: Build+ is the unified coding agent for planning and implementation. It detects intent automatically:
- "What do you think..." / "How should we..." → Deliberation mode (research, discuss)
Expand Down Expand Up @@ -75,7 +75,7 @@ Never skip testing. Never declare work "done" without verification. The full-loo

## Terminal Capabilities

Full PTY access: run any CLI (`vim`, `psql`, `ssh`, `htop`, dev servers, `opencode -p "subtask"`). Long-running: use `&`/`nohup`/`tmux`. Parallel AI: `tools/ai-assistants/opencode-server.md`.
Full PTY access: run any CLI (`vim`, `psql`, `ssh`, `htop`, dev servers, `Claude -p "subtask"`). Long-running: use `&`/`nohup`/`tmux`. Parallel AI: `tools/ai-assistants/opencode-server.md`.

---

Expand Down Expand Up @@ -257,7 +257,7 @@ worktree-helper.sh add feature/x # Fallback

## Autonomous Orchestration

**CLI**: `opencode` is the ONLY supported CLI for worker dispatch. Never use `claude` CLI.
**CLI**: `Claude` is the ONLY supported CLI for worker dispatch. Never use `claude` CLI.

**Supervisor** (`supervisor-helper.sh`): Manages parallel task execution with SQLite state machine.

Expand Down Expand Up @@ -288,7 +288,7 @@ supervisor-helper.sh status <batch-id>

**Cron pulse is mandatory** for autonomous operation. Without it, the supervisor is passive and requires manual `pulse` calls. The pulse cycle: check workers -> evaluate outcomes -> dispatch next -> cleanup.

**Session memory monitoring + respawn** (t264, t264.1): Long-running OpenCode/Bun sessions accumulate WebKit malloc dirty pages that are never returned to the OS (25GB+ observed). Phase 11 of the pulse cycle checks the parent session's `phys_footprint` when a batch wave completes (no running/queued tasks). If memory exceeds `SUPERVISOR_SELF_MEM_LIMIT` (default: 8192MB), it saves a checkpoint, logs the respawn event to `~/.aidevops/logs/respawn-history.log`, and exits cleanly for the next cron pulse to start fresh. Use `supervisor-helper.sh mem-check` to inspect memory and `supervisor-helper.sh respawn-history` to review respawn patterns.
**Session memory monitoring + respawn** (t264, t264.1): Long-running Claude Code/Bun sessions accumulate WebKit malloc dirty pages that are never returned to the OS (25GB+ observed). Phase 11 of the pulse cycle checks the parent session's `phys_footprint` when a batch wave completes (no running/queued tasks). If memory exceeds `SUPERVISOR_SELF_MEM_LIMIT` (default: 8192MB), it saves a checkpoint, logs the respawn event to `~/.aidevops/logs/respawn-history.log`, and exits cleanly for the next cron pulse to start fresh. Use `supervisor-helper.sh mem-check` to inspect memory and `supervisor-helper.sh respawn-history` to review respawn patterns.

**Full docs**: `tools/ai-assistants/headless-dispatch.md`, `supervisor-helper.sh help`

Expand Down Expand Up @@ -404,7 +404,7 @@ Development → @code-standards → /code-simplifier → /linters-local → /pr

Import community skills: `aidevops skill add <source>` (→ `*-skill.md` suffix)

**Cross-tool**: Claude marketplace plugin, Agent Skills (SKILL.md), OpenCode agents, manual AGENTS.md reference.
**Cross-tool**: Claude marketplace plugin, Agent Skills (SKILL.md), Claude Code agents, manual AGENTS.md reference.

**Full docs**: `scripts/commands/add-skill.md`

Expand Down
2 changes: 1 addition & 1 deletion AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
- **Repo**: `~/Git/aidevops/`

**AI Tool Config Paths**:
- Primary: `~/.config/opencode/opencode.json`, `~/.config/opencode/agent/`
- Primary: `~/.config/Claude/Claude.json`, `~/.config/Claude/agent/`
- Companion CLI: `~/.claude/`, `~/.claude/settings.json`

**Development Commands**:
Expand Down