beacon

A set of rules for organizing project documentation, including where each kind of document belongs and how files are named.

In plain words
What is it for?
It helps coding agents create, update, archive, and classify plans, decision records, backlogs, evaluations, and operational guides consistently.
Why use it?
It prevents duplicate, misplaced, or hard-to-find documents and keeps decisions and project history available after a chat ends.

Cursor rule for Cursor

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 rules/juliocbm/beacon-docs/beacon
Clone the repo
git clone --depth 1 https://github.com/Juliocbm/beacon-docs

Made for: Cursor.

Per session 1,058 This file is loaded in full into every session.
When invoked 1,058 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 $0.01058 $0.01058
Opus 5 $0.00529 $0.00529
Sonnet 5 $0.00212 $0.00212
Haiku 4.5 $0.00106 $0.00106

Measured 2d ago against content hash 0872c9f30e27, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

beacon 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.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

.cursor/rules/beacon.mdc · 52 lines

How it starts

The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Universal rules

  1. One doc = one category. Never duplicate across folders.
  2. Status via folder, never filename. Closed plan → _archive/, never *_DONE.md or *_v2.md.
  3. kebab-case for all filenames. No camelCase, no snake_case, no spaces.
  4. README required in every category folder.
  5. ADRs are append-only. Supersede with a new ADR; never edit an accepted one.
  6. Evaluations are immutable snapshots. To revise, create a new dated file.
  7. Plans archive to _archive/ when done. Move them, never rename.
  8. Generated AI files (this one) must match docs/_meta/convention.md. Run beacon sync after editing the convention.
  9. Don't create folders outside the convention. Run beacon enable <addon> or propose an ADR.

Persistence rule: Decisions live in ADRs, multi-step work lives in plans, deferred work lives in backlog, retrospection lives in evals. Chat memory is session-scoped and lost when the conversation ends. If you find yourself relying on chat memory to track a decision, plan, or follow-up — write the document instead.

Project-specific rules

  • operations/*.guide.md — deploy guides, admin runbooks, troubleshooting. Post-mortems go in evaluations/.

Workflow triggers — when to create documents

These are the conversational signals that should prompt you to create or update Beacon documents. Treat them as imperatives, not suggestions.

  • Design decision made → write an ADR before implementing. Decisions about architecture, dependencies, naming conventions, or future flexibility belong in adr/. Run beacon new adr <slug>. Draft the ADR inline; don't ask permission to write it.
  • Multi-step work agreed (3+ distinct actions) → write a plan with checkbox steps. Run beacon new plan <slug> and write bite-sized steps. Check off each step in the same commit that completes it. Don't keep multi-step work only in chat memory.
  • Scope deferred → capture it as a backlog item immediately. When the user says "let's do that later", "out of scope for now", "maybe v0.5", or similar, run beacon new todo <slug> and write the why plus acceptance criteria. Don't let deferred work live only in conversation.
  • Release shipped or milestone reached → write a dated retrospective eval. Run beacon new eval <slug>-retrospective and capture what worked, what didn't, what you'd do differently. Even a 10-line eval is better than zero.
  • Approach explained twice → write a pattern doc. If you find yourself explaining the same technical approach across multiple files or sessions, run beacon new pattern <slug> and capture it in reference/.
  • System structure changed → update or add an architecture doc. New module, new layer, new external dependency that affects how the system is shaped — that goes in architecture/.

Read the full file on GitHub · 52 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. 2d ago First seen · 52 lines · 1,058 tokens per session scan A 0872c9f30e27

Subscribe to this mod's changes

beacon is a cursor rule published in the GitHub repository Juliocbm/beacon-docs (6 stars, last pushed 3mo ago), licensed MIT. It adds 1,058 tokens to every session, about $0.0053 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 2026-08-31.