Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/lucasduys/forge/watchgit clone --depth 1 https://github.com/LucasDuys/forgeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00011 | $0.01491 |
| Opus 5 | $0.00005 | $0.00745 |
| Sonnet 5 | $0.00002 | $0.00298 |
| Haiku 4.5 | $0.00001 | $0.00149 |
Grade A, and why
watch scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Forge Watch
Launch the autonomous implementation loop with an interactive TUI dashboard. Identical to /forge execute except that it spawns claude with --output-format stream-json --verbose and renders a live dashboard showing the current phase, active agent + tool, frontier progress bar, token counters, restart meter, and a scrolling transcript pane.
Use this when you want eyes-on visibility into what Forge is doing in real time. Use /forge execute when you want plain stdout (better for CI logs, redirected output, or small terminals).
Pre-flight Check
Identical to /forge execute:
-
Verify
.forge/exists. If it does not, stop and tell the user:.forge/not found. Run/forge brainstormfirst to generate specifications, then/forge planto create task frontiers. -
Verify
.forge/specs/contains at least one spec file. If not, stop and tell the user:No specs found. Run
/forge brainstormfirst. -
Verify
.forge/plans/contains at least one*-frontier.mdfile. If not, stop and tell the user:No task frontiers found. Run
/forge planfirst to decompose specs into tasks. -
Check for Ralph Loop conflict: if
.claude/ralph-loop.local.mdexists, stop and tell the user:Ralph Loop is active. Run
/cancel-ralphfirst — only one loop plugin should be active at a time. -
Verify Node 18+ is on PATH (
node --version). If not, stop and tell the user:/forge watchneeds Node.js 18+ for the TUI dashboard. Install Node or use/forge executefor the plain runner.
Parse Arguments
Accepts all /forge execute flags plus TUI-specific flags:
| Flag | Default | Description |
|---|---|---|
--autonomy full|gated|supervised |
from config | Inherited from /forge execute |
--max-iterations N |
from config | Inherited from /forge execute |
--token-budget N |
from config | Inherited from /forge execute |
--depth quick|standard|thorough |
from config | Inherited from /forge execute |
--filter NAME |
(all specs) | Inherited from /forge execute |
--max-restarts N |
10 | Max Claude restart attempts (TUI runner) |
--base-delay N |
3 | Base backoff delay in seconds |
--transcript-lines N |
50 | Transcript ring buffer size |
--no-fallback |
off | Do not fall back to plain runner on TUI self-abort |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 119 lines · 11 tokens per session scan A 729213599d81
watch is a command published in the GitHub repository LucasDuys/forge (55 stars, last pushed 1mo ago), licensed MIT. It adds 11 tokens to every session and 1,491 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
audit-system
Run a comprehensive audit of an existing design system for consistency, completeness, and accessibility.
build-presentation
Build a design presentation end to end — audience framing, narrative structure, and supporting rationale.
discover
Run a full user research cycle — persona creation, empathy mapping, and journey mapping for a product or feature.
interview
Prepare an interview script or summarize an interview transcript into structured insights.
deploy
Deploy the application to production.
eg-new-feature
Build a new feature using the elephant/goldfish workflow — design doc, goldfish design check, implement, review, validate.