The Atlas Docs Ledger reconciles each document against the codebase and grades it into 1 of 4 verdicts: looks shipped, forgotten intent, orphaned, or stale. It nudges you when a spec you wrote was never built, when a shipped doc points at code that no longer exists, or when documentation has drifted from what it describes. The routine pass contains no LLM calls, so it costs nothing to run. The engine ships today behind the paid tier, and Atlas Pro is not yet on sale.
Documentation that holds itself accountable
Most documentation rots quietly, in 1 of 2 directions: a spec written early is forgotten, or a shipped doc keeps describing behaviour that changed.
The Docs Ledger treats the status you declare in a document as a claim and the actual code as reality. When the two disagree, that gap becomes a verdict rather than a vague worry: a planned feature never built is forgotten intent, a doc describing deleted code is orphaned, and a shipped doc whose code has churned is stale.
Zero tokens for the routine pass
The everyday reconcile runs with 0 LLM calls, using a 2-tier deterministic cascade.
Tier 1 is an explicit-target check: does the code this document names actually exist. Tier 2 is a convention-by-location and keyword match. Both are deterministic, so Atlas can run the check on every session without metering you. A deeper LLM judgement exists for prose-only cases, and it runs only when you ask with --deep.
An audit trail under every file
Knowing when a document first appeared is the basis for any staleness signal, so Atlas reads it from git rather than guessing.
Creation dates come from git history for committed files, with a first-seen fallback for drafts that are not committed yet. Those dates are cached against the file's modification time, so an unchanged document is never re-read. That trail is what lets Atlas tell you how long a planned spec has been sitting unbuilt and rank what actually needs attention.
Availability
Docs Ledger is 1 of the 8 record engines in Atlas, and it sits behind the paid tier.
The engine is built and running in the product, gated to Atlas Pro. Pro has no checkout yet, so it is not something you can buy today. Unlike most of the record engines it needs no meeting transcripts, since it reads only your docs and your code. Shipped, the automated work log, is the one record engine available free right now.
Frequently asked questions
- Is the Atlas documentation audit trail available now?
- The engine is built and running in Atlas, but it sits behind the paid tier and Atlas Pro is not yet on sale, so you cannot purchase access today.
- Where do the creation dates come from?
- From git history for committed files, with a first-seen fallback for drafts that have not been committed yet. Dates are cached by file path and modification time.
- What kinds of documentation problems does it catch?
- Four: docs that look shipped, planned specs never built (forgotten intent), docs whose code was deleted (orphaned), and docs that have drifted from the code they describe (stale).
- Does it use an LLM?
- Not for the routine pass, which is fully deterministic and spends no tokens. A deeper LLM judgement runs only when you pass --deep.
- Will it change my files automatically?
- No. It surfaces and suggests. You confirm before any document is updated.
Try Atlas in your terminal
The terminal-native AI coding agent. Free core, single binary.
Install AtlasRelated guides
Atlas for Deno: Terminal-Native AI Coding in 2026
Atlas is a terminal-native AI coding agent for Deno. Tighten --allow-all down to real permission scopes, add JSR deps, and run deno test and deno check in 2026.
Atlas with Llama 3.3 70B (local via Ollama) in 2026: Dense 70B on Your Own Hardware
Llama 3.3 70B (local via Ollama) drives Atlas at 128K tokens (131,072) and Free (self-hosted) pricing, with a $0.59 / $0.79 per Mtok Groq fallback on identical weights.
Atlas with GPT-5.1 Codex Max: The Cheapest Codex Tier in 2026
GPT-5.1 Codex Max still bills at the November 2025 rate of $1.25 / $10 per Mtok on a 400K window. The lowest-cost entry into OpenAI's Codex post-training for Atlas.
Atlas with Qwen3.6 Max Preview: The Top 3.6 Tier, Reviewed in 2026
Qwen3.6 Max Preview is Alibaba's top Qwen3.6 tier at $1.30 per Mtok input and $7.80 per Mtok output, with 256K tokens (262,144) of context. What it is worth inside Atlas.
Atlas with GLM-4.5-Air: The 106B Self-Hostable Cheap Slot in 2026
GLM-4.5-Air drives Atlas at $0.20 per Mtok input and $1.10 per Mtok output on a 128K tokens (131,072) window. A 106B total / 12B active MIT-licensed MoE.
Extract a Shared Helper from Duplicated Code with Atlas (2026 Workflow)
How to extract a shared helper from duplicated code with Atlas in 2026: codebase_search finds the copies by meaning, write creates the module, apply_patch swaps each call.
Atlas with Gemini 3.5 Flash: The May 2026 Latency Tier That Keeps 1M Context
Gemini 3.5 Flash in Atlas: the May 2026 release keeps the full 1,048,576 token window at $1.50 / $9 per Mtok, with reasoning and tool calling enabled.
Atlas for Erlang in 2026
Atlas is a terminal-native AI coding agent for Erlang/OTP in 2026. Run it in an app with a rebar.config, map supervisors and gen_server modules, review every diff.