FAQ

Questions, answered concretely

Concrete and checkable beats superlatives. If an answer here ever stops being true, that is a bug.

What is sitrep?
sitrep is an AI project status dashboard: a state layer on top of the markdown your AI agents already generate. It reads the plans, decisions and progress your agents already write - no new format, no import - and shows what moved, what is stalled and what is waiting on you, across every project.
What is an AI project status dashboard?
A tool that turns the markdown files AI coding agents produce - plans, analyses, decision logs, progress notes - into a glanceable overview of project state. Instead of re-reading documents to work out where things stand, you see extracted status, phases, open decisions and staleness across all your projects at once.
Which AI coding agents does sitrep work with?
sitrep reads Claude Code, Cursor and Codex markdown across all your projects - and any other markdown, agent-written or human-written. It is agent-agnostic by construction: it parses conventions in the files themselves, not any tool-specific format.
Do I need to add frontmatter or restructure my files?
No. Zero adoption cost is a core promise: no new format, no frontmatter requirement, no import, no vault, no restructuring. You point sitrep at folders and it reads what is already there. Extraction vocabularies are editable settings, never demands on your files.
How does sitrep know the status of a project?
It extracts the conventions already common in agent-written docs: dated status lines like "**Status (2026-08-08):** waiting on review", phase headings like "## Phase 2 (SHIPPED)", decision tables with a verdict column, and due-date lines. It deliberately ignores misleading signals - checkboxes (QA sign-off, not progress) and git history alone (in real corpora much of the in-progress work is gitignored or untracked).
Is sitrep a markdown viewer or editor?
Neither. sitrep renders markdown at full fidelity - GFM, highlighted code, Mermaid, KaTeX, callouts - but rendering is the floor, not the product. It is not an editor at all: it will never restructure, rewrite or manage your files.
Does sitrep write to my files?
Only through one narrow, explicit surface: three convention-shaped operations - set a status, set a due date, mark a decision superseded - each writing a single line in the document's own convention, behind a preview and a content-hash guard. No other code path writes to your documents, ever. Sorting, filtering and saved views are local view state and never touch the file.
Is my data uploaded anywhere?
No. sitrep is local-first: fully functional forever with no account and no network, and no telemetry by default. Your files never leave your machine. A future optional sync layer will be end-to-end encrypted with user-held keys - the relay cannot read the data it carries.
How is sitrep different from Obsidian or a second-brain app?
Obsidian is a knowledge base you author into; sitrep is a status dashboard over files you never authored for it. There are no vaults, no backlink graphs and no embeddings - search stays scoped to jumping to the decision or doc you need. If you want a second brain, use Obsidian; sitrep answers a different question: where do my projects stand?
Does sitrep replace my task manager?
No, and it refuses to become one. A task manager makes you author its format; authoring structured state is the exact friction sitrep removes. It surfaces the state your docs already contain, including the decisions still waiting on you.
How does sitrep know a document is stale?
A doc claims work is happening when it carries an active dated status or an unfinished phase list. sitrep asks four questions, strongest first: has the whole project been silent over two months (dead claim)? Is this a plan whose own linked step docs moved two weeks past it (plan adrift - the classic agent failure, where a session updates the six step documents and never touches the master plan)? Is the project moving while this doc sat untouched over a month (left behind)? Is the claim's own date over a week old (stale claim)? The badge shows the number that tripped it - "left behind 40d" means your repo moved 40 days past the plan - and it is computed fresh at read time, never stored.
Won't staleness flags just nag me?
The thresholds were measured, not guessed: every flag was read against the real development corpus before shipping, and the intuitive two-week threshold would have flagged half of everything - so the drift line sits at thirty days, where the reviewed false-positive rate was zero. When sitrep is unsure it stays quiet, because a false "this doc lies" is worse than a missed one. And the thresholds are settings with global, workspace and project granularity - a fast project can call three days stale while a quarterly one is fine at ninety.
How do I keep track of all the markdown files Claude Code creates?
Stop tracking files and track state instead. sitrep points at the folders your projects already live in and parses the plans, analyses and decision docs Claude Code writes into one cross-project overview - status, phases, open decisions, due dates - so the pile of markdown becomes a glanceable inbox instead of a reading list.
How do I resume a Claude Code project after weeks away?
Open sitrep instead of your memory: the project's docs are parsed into what moved last, what the stated status is, which decisions are still open and what has gone stale. Re-entry becomes triage - read the one or two documents that matter - rather than re-deriving where things stand from scratch.
How do I see what my AI agents did across all my repos?
That is sitrep's core surface: an email-client-style overview across every project - newest activity first, pending items surfaced. It is agent-agnostic, so sessions run in Claude Code, Cursor, Codex or anything else that writes markdown all land in the same inbox, whichever machine or tool produced them.
Can sitrep stop me re-explaining context to Claude every session?
Partly, and honestly: your CLAUDE.md still briefs the agent. What sitrep adds is an MCP server exposing parsed state - status, open decisions, due items, search - so a cold session can query where things stand instead of re-reading every doc. And it briefs you, which CLAUDE.md never did.
What platforms does sitrep run on?
macOS first, Windows second, Linux best-effort. It is a desktop app - local-first by design, not a web service.
When can I use it?
sitrep is in active development and launching first on macOS. This site is where availability will be announced, and the pages here describe only what the product actually does - claims are measured before they are made.
Can my agents query sitrep?
That is the plan: an MCP server exposing the parsed state - status, rulings, due dates, search - read-only first, so agents can consult project state in their own loop. Any writes would go through the same single writing surface as the UI.