The Atlas Decision Ledger tracks decisions in 2 states: open, meaning a decision you still need to make, and settled, meaning one you made along with why. Its headline job is surfacing the open decisions you forgot were pending, oldest first, so a question that was quietly blocking progress actually gets decided. Settling one writes an ADR file into your repository and injects it into future sessions so the agent does not re-argue it. The engine ships today behind the paid tier, and Atlas Pro is not yet on sale.
The decisions that hurt are the open ones
The costly decisions are not the 1 you made last week, they are the ones you needed to make and forgot were open.
The Decision Ledger keeps open decisions visible and aging-aware, ordered oldest first, so a question raised on a call does not stall the work indefinitely. The ordering matters: a digest sorted by recency shows you what you already remember, while one sorted by age shows you what you forgot.
Settled decisions keep their reasons
A decision without its rationale gets re-argued within 2 months; a recorded one does not.
When you settle a decision, Atlas captures the outcome and the reasoning and writes it to decisions/NNNN-slug.md in your repository as an ordinary git-tracked ADR. The file is yours and readable without Atlas. Settling is idempotent, so a decision you confirmed never silently flips on a later pass.
Settled decisions reach the agent, open ones do not
Only settled decisions are injected into session context, and that 1 rule is a deliberate safety property.
Atlas renders settled decisions into the system prompt so the agent works from what you already concluded. Open decisions stay out. That way an unresolved or wrongly captured question stays visible in your digest without silently steering every future session in the repository.
Availability
Decision 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. Its live input is Intake, which reads call transcripts. Capturing decisions from coding sessions is built as a separate engine but is deliberately manual for now. Shipped, the automated work log, is the one record engine available free right now.
Frequently asked questions
- Is Atlas decision tracking 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.
- How is this different from a decision log?
- A log only records decisions you already made. The Decision Ledger also tracks open decisions you still need to make, ages them, and surfaces the oldest first.
- Where do the decisions come from?
- From the decisions Atlas Intake extracts from your call transcripts. Capturing decisions made during a coding session is built as a separate engine but is manual rather than automatic today.
- What happens when I settle a decision?
- Atlas writes an ADR to decisions/NNNN-slug.md in your repository with the outcome and rationale, flips the ledger entry to settled, and starts injecting it into session context.
- Do open decisions affect what the agent does?
- No. Only settled decisions are injected into context. An open or wrongly captured decision stays visible in your digest without steering the agent.
Try Atlas in your terminal
The terminal-native AI coding agent. Free core, single binary.
Install AtlasRelated guides
Atlas with Llama 4 Scout (Ollama): A 10M-Token Window on Local Hardware in 2026
Run Atlas on Llama 4 Scout (Ollama): a 67GB 16-expert MoE with a 10M-token context and image input, free self-hosted. The memory math before you pull 67GB.
Atlas vs Fine.dev: Terminal AI Coding Agents in 2026
Compare Atlas and Fine.dev for terminal AI coding in 2026. Atlas offers a free core TUI with permission-gated tools, while Fine.dev provides asynchronous cloud agents and prebuilt libraries.
Atlas with Qwen3.6 27B: The 2026 Dense Checkpoint, Priced Honestly
Qwen3.6 27B is the dense reasoning model of Alibaba's April 2026 line: 256K tokens (262,144) of context at $0.60 per Mtok input and $3.60 per Mtok output, running inside Atlas.
Atlas with Qwen3.6 Plus: The Stable Million-Token Tier in 2026
Qwen3.6 Plus gives Atlas 1M tokens (1,000,000) of context at $0.50 per Mtok input and $3.00 per Mtok output, matching Qwen3.7 Plus while keeping 3.6 generation behavior.
Atlas with Amazon Nova Pro: Bedrock Billing and a 300K Window (2026)
Amazon Nova Pro runs Atlas at $0.80 / $3.20 per Mtok on a 300,000 token window, billed through your existing AWS account with no new vendor contract.
Atlas with Qwen3-Coder 480B (Ollama): the self-hosted ceiling in 2026
Qwen3-Coder 480B (Ollama) in Atlas: 290GB of weights, roughly 292GB to serve, 256K tokens (262,144) of context. Free (self-hosted), but the hardware is not.
Atlas with GPT-5.3 Codex: Code-Specialized Reasoning in 2026
GPT-5.3 Codex is OpenAI's February 2026 code-specialized reasoning model, $1.75 / $14 per Mtok on a 400K window. Built for the long agentic loops Atlas runs.
Atlas with Qwen2.5 72B (local via Ollama): Air-Gapped Coding in 2026
Run Atlas fully offline on Qwen2.5 72B (local via Ollama) in 2026. Free (self-hosted), about 47 GB at Q4_K_M, and a 32,768 token local context cap.