vibe

Vibe Orchestrator

/vibeon | /vibeoff | /vibestatus

Toggle auto-delegate mode — Vibe automatically handles coding tasks without requiring /vibe each time.

| Command | Action | |---------|--------| | /vibeon | touch ~/.local/share/vibe-auto.flag → confirm "Auto-vibe ON" | | /vibeoff | rm -f ~/.local/share/vibe-auto.flag → confirm "Auto-vibe OFF" | | /vibestatus | report auto-mode (ON/OFF) and active model override |

For /vibestatus, run both checks and print two lines:

Auto-vibe: ON | OFF
Model: <alias>  (override)  OR  Model: deepseek-flash  (config default)

Auto-mode pre-filter (when flag is set)

When vibe-auto.flag exists, apply this gate before loading the full skill:

| Task signal | Action | |---|---| | 1 file, ≤10 lines, exact location already known | Edit directly — do NOT invoke the skill | | Logic non-trivial, location unclear, multiple files, HTML/JS content, or >1 change | Invoke /vibe as normal |

/vibe-report

If the user invokes /vibe-report, run ~/tools/delegate-report with any flags extracted from the arguments, display output verbatim, and stop.

| User says | Flag | |-----------|------| | "last 7 days", "7d" | --since 7 | | "last 30 days", "30d" | --since 30 | | "project foo" | --project foo | | "only failures", "fails", "bugs" | --fails | | (nothing) | (no flags — full report) |

/vibe-model-pick | /vibe-model-clear

Override the Vibe model for all subsequent delegations without touching ~/.vibe/config.toml. Works via VIBE_ACTIVE_MODEL env var, which Vibe respects over the config file.

| Command | Action | |---------|--------| | /vibe-model-pick <alias> | echo <alias> > ~/.local/share/vibe-model.flag → confirm | | /vibe-model-clear | rm -f ~/.local/share/vibe-model.flag → confirm "back to config default" |

Available aliases (from ~/.vibe/config.toml):

| Alias | Model | Provider | Notes | |-------|-------|----------|-------| | deepseek-flash | deepseek-v4-flash | DeepSeek | Default — fast, cheap | | mistral-medium-3.5 | mistral-vibe-cli-latest | Mistral | Stronger reasoning | | devstral-small | devstral-small-latest | Mistral | Lighter Mistral model | | local | devstral (llamacpp) | Local | Requires local server on :8080 |

Run the bash command, print one confirmation line showing the active model, and stop.

When the user invokes /vibe <instruction>, Claude delegates the implementation to Mistral Vibe via its programmatic mode, supervises in real time, and reports.

Known Limits

Hard constraints of Mistral Vibe CLI — not config options.

1. UTF-8 / special chars cause search_replace failures

Vibe's search_replace tool matches byte-for-byte. Accented chars, curly quotes, or emoji in old_string → silent match failure, no write. Workaround: use python3 str.replace() for those edits, or restructure the prompt to avoid them.

2. Code duplication bug

Vibe sometimes re-inserts a block it has already written (off-by-one in its diff logic). Check for duplicate function definitions or repeated class bodies after every run.

3. Orchestration chain has 6 independent failure points

The delegation pipeline is: vibe CLI → pseudo-TTY (script) → Python stream parser → TOML pricing lookup → git diff → JSON log. Each link can fail independently:

| Link | Failure mode | Symptom | |------|-------------|---------| | Vibe CLI | Auth expired, update broke API | Immediate exit, no output | | pseudo-TTY (script) | Platform difference (GNU vs BSD flags) | Hangs silently or garbled output | | Stream parser | Vibe changes its JSON schema | Tool calls not detected, wrong token count | | TOML pricing | config.toml missing or renamed | Falls back to Mistral Medium 3.5 rates | | git diff | Not a git repo, or Vibe committed mid-run | Wrong file count, misleading stat | | JSON log | ~/.local/share/ not writable | Silent log skip, /vibe-report misses the run |

When a run produces unexpected results, check these links in order from top to bottom.

4. Never pass source code through a bash heredoc

Nested quotes, f-strings, or backslashes in inline bash << 'PYEOF' mangle escaping.

• ASCII code: use search_replace directly.
• Content too long for inline: ask Vibe to write to /tmp/new.py, then search_replace via open('/tmp/new.py').read().
• Never write a helper script whose sole job is str.replace() on another file.

5. HTML tags in the prompt body cause shell redirect errors (exit 127)

<div>, </div>, <span> etc. embedded in the prompt string are interpreted by bash as file redirections. div is not a command → "No such file or directory", exit 127. Rule: if the prompt references or contains any HTML — even in a quoted description — write the content to a temp file first:

cat > /tmp/new_block.html << 'EOF'
<div class="map-sidebar">...</div>
EOF

Then reference it in the prompt: "Replace the sidebar block in templates/map.html with the content of /tmp/new_block.html." Vibe reads it with read_file, no shell parsing.

This also applies to JS files containing template literals (backticks) or JSX-like syntax.

Step 1 — Detect workdir

1. git rev-parse --show-toplevel in the current directory.
2. If ambiguous or no git repo → ask with AskUserQuestion.

Step 2 — Decompose the task

Critical rule: Vibe is optimized for atomic, focused tasks. Its system prompt literally says "Most tasks need <150 words."

| Size | Definition | Max turns | Approach | |------|-----------|-----------|----------| | Trivial | 1 file, change is obvious and located | — | Skip delegation — edit directly | | Simple | 1 file, non-trivial logic or unknown location | 5–8 | 1 vibe call | | Medium | 2–3 related files, 1 objective | 8–12 | 1 structured vibe call | | Complex | >3 files OR business logic OR DB migrations | — | Break into sub-tasks |

Decomposition for complex tasks:

Sub-task 1: Explore / read relevant files (read-only, 5 turns)
Sub-task 2: Implement change A in file X (8 turns)
Sub-task 3: Implement change B in file Y (8 turns)
Sub-task 4: Verify / test (5 turns)

→ Check git diff between each sub-task before launching the next.

Step 3 — Write the Vibe prompt

Vibe has no context from the parent conversation. The prompt must be self-contained.

Structure of a good Vibe prompt:

Stack: Python/Flask, SQLAlchemy, SQLite
Key files: app.py (routes + fetch), models.py (Entry)

TASK: [one single thing to do, stated as an imperative]

CONSTRAINTS:
- [what must not break]
- [expected format if relevant]

VERIFY: grep for "def function_name" in file.py and confirm it exists.

Formulation rules:

• One task per prompt — never "also do X and Y"
• Name the exact files to modify
• Include a grep-based verification criterion (not a file re-read)
• Language: English (better Mistral performance)

⚠️ Shell safety: if the prompt contains UTF-8 accented chars, emojis,

: in Python/YAML code, or typographic apostrophes — the vibe-delegate script

passes them safely via a temp file (printf %q). Never interpolate such a prompt

directly into a bash heredoc.

Verification — always use grep, not file re-read:

VERIFY: grep for "def extract_labels" in app.py and confirm it exists.

A grep is reliable. A file re-read may miss content outside the read window.

Step 4 — Launch Vibe

~/tools/vibe-delegate "<workdir>" "<prompt>" [max-turns] [agent] [timeout-secs]

| Argument | Default | Notes | |----------------|----------|-------------------------------------------------| | workdir | — | Absolute path, must exist | | prompt | — | Self-contained task description | | max-turns | 10 | Mistral turn limit — hard cap at 12, never more | | agent | (none) | See agent t