agentmods.dev

watchfire CLAUDE.md

watchfire CLAUDE.md is an instructions file for coding agents from watchfire-io/watchfire. It costs 1,045 tokens per session, scanned A, original, Apache-2.0.

Project instructions for Watchfire, a system that coordinates coding-agent sessions from project definitions and tasks. They describe the architecture, source-of-truth rules, task completion, merging, and optional GitHub pull requests.

In plain words
What is it for?
Use them when changing Watchfire, checking its architecture, or handling the merge and pull-request process after an agent finishes.
Why use it?
They give agents a consistent reference for how Watchfire is organized and how finished work is handled.

Instructions file

not rated 61 13d ago A scan 1,045 tokens original Apache-2.0
View the source file watchfire-io/watchfire
Install

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.

agentmods
npx agentmods add instructions/watchfire-io/watchfire/claude-md
Clone the repo
git clone --depth 1 https://github.com/watchfire-io/watchfire

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for watchfire CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/watchfire-io/watchfire/claude-md.svg)](https://agentmods.dev/instructions/watchfire-io/watchfire/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/watchfire-io/watchfire/claude-md"><img src="https://agentmods.dev/badge/instructions/watchfire-io/watchfire/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,045 This file is loaded in full into every session.
When invoked 1,045 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.01045 $0.01045
Opus 5 $0.00522 $0.00522
Sonnet 5 $0.00209 $0.00209
Haiku 4.5 $0.00104 $0.00104

Measured 4d ago against content hash b356b643eef6, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

watchfire CLAUDE.md 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 4d 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.

CLAUDE.md · 92 lines

How it starts

The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Watchfire - Development Guide

What is Watchfire?

Watchfire orchestrates coding agents sessions based on specs (project definitions and tasks). Read ARCHITECTURE.md for the full design.

Source of Truth

ARCHITECTURE.md is the single source of truth for:

  • Component responsibilities
  • Data structures
  • Directory layout
  • Build phases
  • Tech stack

If this file and ARCHITECTURE.md conflict, ARCHITECTURE.md wins.

For all decisions, use architecture document as reference. If you need to do something different, please update atchitecture after checking with user.

Task-completion lifecycle (v7.0 Relay)

When an agent finishes a task, the daemon picks one of two merge paths via the on-task-done hook (internal/daemon/agent/taskdone.go:HandleTaskDone):

  • Silent merge (default). git merge --no-ff watchfire/<n> lands the work on the project's default branch and the worktree is removed. This matches the v6.x behaviour and is the path every project gets unless they opt in to auto-PR.
  • GitHub auto-PR. If ~/.watchfire/integrations.yaml enables github.auto_pr for the project (either globally or via project_scopes), the daemon pushes watchfire/<n> to GitHub and opens a PR via gh api with body rendered from the task metadata + v6.0 diff stats. The local merge is suppressed; the user reviews and merges in GitHub. The worktree is still cleaned up after PR creation.

Auto-PR requires the gh CLI on PATH and gh auth status returning 0. If either fails, the daemon logs a single WARN per project lifetime and falls back to silent merge — no task ever strands inside an unmerged worktree. Push failures and gh api errors fall back the same way but log per failure.

To enable, set github.enabled: true in ~/.watchfire/integrations.yaml and optionally restrict to specific projects via github.project_scopes: [<id>...].

Telegram Bridge (v10.0 Torch)

The daemon can be supervised from Telegram: a long-polling bridge (internal/daemon/telegram/ + internal/daemon/telegrambot/) with pairing as the allowlist, live watch-mode relay, and an outbound relay adapter. See ARCHITECTURE.md → "Telegram Bridge (internal/daemon/telegram/) — v10.0 Torch" for the design. Two invariants for anyone touching bridge code: it never calls Resize, and injectSay (the /say verb and plain-text chat forwarding) is the only PTY write — both enforced by internal/daemon/telegram/watch_guard_test.go.

Read the full file on GitHub · 92 lines

Changes

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.

  1. 4d ago First seen · 92 lines · 1,045 tokens per session scan A b356b643eef6

Subscribe to this mod's changes

watchfire CLAUDE.md is an instructions file published in the GitHub repository watchfire-io/watchfire (61 stars, last pushed 13d ago), licensed Apache-2.0. It adds 1,045 tokens to every session, about $0.0052 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.

Orchestrationai-agentsclicoding-agentsgit-workflow
Related

Other instructions, from other repositories

dmux AGENTS.md

Instructions for standardagents/dmux, covering agents.md - dmux maintainer guide, docs map, project overview, important architecture landmarks and native helper.

standardagents/dmux · 1,916 tokens

dmux CLAUDE.md

Instructions for standardagents/dmux, a project described as: A dev agent multiplexer for git worktrees and coding agents.

standardagents/dmux · 3 tokens

workflow AGENTS.md

AGENTS.md instructions for vercel/workflow, covering agent instructions, architecture, core components, workflow execution model and development commands.

vercel/workflow · 8,698 tokens

ephemeral-sandbox CLAUDE.md

Instructions for Ephemeral-AI-Lab/ephemeral-sandbox, covering claude.md, project, engineering practice (required), build & test and sandbox tools.

Ephemeral-AI-Lab/ephemeral-sandbox · 1,203 tokens

matt-skills-with-to-goal CLAUDE.md

Instructions for tt-a1i/matt-skills-with-to-goal: Skills are organized into bucket folders under skills/.

tt-a1i/matt-skills-with-to-goal · 989 tokens

pixir CLAUDE.md

Instructions for Ranvier-Technologies/pixir, covering pixir harness - legacy agent guide, current map and commands.

Ranvier-Technologies/pixir · 246 tokens