How Campfyrr works: briefs, assumptions and handoffs
By Soumen Sarkar · October 2026
Campfyrr carries context between the people and AI agents on a team. It briefs each agent before it changes an area of the code, records assumptions and decisions as they're made, leaves a handoff whenever work stops, attaches the reasons to every change, and shows people one Activity page per project.
Campfyrr is built on VibeFlock, a way of working where nobody waits. This article explains, plainly, what happens while your agents work, what your team sees, and what stays private.
What happens when an agent starts work?
It starts with a brief. When a Claude Code session begins, Campfyrr gives the agent:
- Work to pick up: a handoff it's continuing, or its own unfinished work.
- What's waiting on the person it works for.
- What changed since its last session, and the decisions and rules that stand.
The agent knows where things are before it writes a line.
What happens when an agent touches the code?
The first time a session reads or edits a file in an area of the code, Campfyrr adds that area's brief to the agent's context:
- Being changed right now: who's working there, what for, and the plan.
- Decided here: the decisions and rules that apply, with their reasons.
- Changed this week: what landed in the area, and why.
- Tried and dropped: approaches that didn't work, so they aren't tried again.
- Open assumptions: what others are currently assuming about this area.
Areas are made automatically from your folders, with names people use ("Payments · lib/billing"). The team can rename them. When the agent edits, teammates see that the area is being changed.
What happens when an agent isn't sure?
It states an assumption and keeps going, using the state_assumption tool:
- Campfyrr checks the team's record first. If the answer is already written down, it comes straight back.
- If not, the assumption goes to the right person: whoever made the closest decision, or last worked in the area.
- That person confirms or corrects it in one click, from a link to the assumption's own page. Their answer reaches the agent at its next turn.
What happens when work stops, lands or depends on other work?
- Stops unfinished: when the context is summarised, the session pauses or it ends, Campfyrr writes a handoff from the session: what's done, what's left, the next step, open assumptions, and what to watch out for. It opens the person's next session under "Pick up where it was left", and a teammate can take it over by pasting it into their first prompt.
- Depends on other work: wait_on links the two. When the other work lands, the waiting work is told what landed and how it differs from the plan.
- Makes a decision: log_decision records it with its reason, and Campfyrr checks it against the decisions that stand. If it contradicts one, the newer call stands, the earlier one is marked replaced, and both agents hear it at their next turn.
- Commits or opens a pull request: Campfyrr writes the change's reasons (what it's for, the calls made, what was tried and dropped, open assumptions) and asks the agent to link them in the pull request.
What does the team see?
One Activity page per project: a read-only feed of updates, each written as what happened and why it matters, with filters for type, who, which area and since when (including since your last visit):
- Decisions: calls made, calls that replaced earlier ones, and open assumptions, each with an "Answer it" link.
- Actions: finished work and pages written.
- Status updates: work in progress with its goal, and work paused partway with its next step.
People react to updates. On top, a "What Campfyrr caught for your team this week" card counts the catches, and a "Where are we?" link gives a plain-words view of the project. Each person's day is also told in plain words, with no standup to write.
How do I connect it, and what stays private?
- Claude: add Campfyrr as a connector, sign in and pick your team.
- Claude Code: one command from the Connect page installs the hooks that do all of the above automatically.
- Codex, Cursor and other MCP clients: connect with a token and use the same tools.
A team's work is visible to that team only. Private pages are never shown to anyone else, never feed a brief, and never appear on the Activity page.
In short
Campfyrr makes context travel with the work. Agents are briefed before they change code, state assumptions instead of stopping, leave handoffs when they stop, and attach their reasons to every change. People see the work told in plain words. Nobody waits.
Add Campfyrr to Claude at https://campfyrr.vercel.app/connect. It takes two minutes. It's free during early access, then $9 per developer a month for teams. Read the seven principles at https://campfyrr.vercel.app/vibeflock.