What is DevSpecs for?
DevSpecs keeps the durable work thread attached to the repo so humans and agents can recover intent, evidence, and handoffs before the next edit.
Local intent trail
DevSpecs keeps the durable thread of AI coding work attached to your repo, so you can let an agent run, come back later, and understand why the changes exist and what happens next.
ds recentLocal-first. No cloud sync. No LLM calls. Works with your existing coding agents.
Git
what the agent changed
fastapi/openapi/docs.py
tests/test_swagger_ui_redirect.pyDevSpecs
why it exists, what to review next
Thread: Swagger OAuth redirect regression
State: tests passed, review requiredContext recovery
Agents and developers already use Git and rg to inspect recent work. DevSpecs adds the local thread: why the change exists, what state it is in, and where to continue.
Q: Can we continue from here?
Git + rg
Install
Start with ds recent when you came back cold. Use ds task when the goal is already clear. On a large first run, DevSpecs shows indexing progress while keeping quiet and JSON output clean.
brew install devspecs-com/tap/devspecsRecover the active thread, lifecycle state, and likely next context before reviewing or continuing the work.
Open reviewed workflowTry on FastAPI
After installing ds, use FastAPI to feel the thread recovery and task handoff loop without touching your own code first.
DevSpecs writes local task files and a local SQLite index. Delete the checkout when you are done.
See reviewed FastAPI workflowsHow it works
DevSpecs reads repo files, docs, specs, plans, commits, and task records. It keeps evidence, decisions, checkpoints, and bounded handoffs connected so the next human or agent can recover the thread.
Intent
Repo
DevSpecs
DevSpecs keeps the durable work thread attached to the repo so humans and agents can recover intent, evidence, and handoffs before the next edit.
Keep using them. Git and rg show source evidence; DevSpecs adds lifecycle state, decisions, superseded work, and the next gate.
No. The CLI is local-first and boring on purpose: it indexes local artifacts and emits context your existing agent can use.
No. Keep Jira, Linear, GitHub Issues, or Notion. DevSpecs tracks the local AI work layer underneath.
Partly, but broader. DevSpecs can create lightweight task specs with packed context, gates, iteration slices, and checkpoints. It also works as a local navigation layer for the intent already scattered through your repo.