Agent-native
The flag is the control plane for agent-shipped code.
A Model Context Protocol server exposes flags, readouts, metrics, logs, and traces as tools, so an agent can create a flag, ramp it, read the lift, pull trace evidence into a rollback rationale, and clean up dead flags — in the tools your team already lives in.
FreeIncluded from the Free tier.
foredeck-mcp12 toolsEvery mutation attributable — human or agent.
What you get
Why Foredeck for agent tools (mcp).
Flags as tools
create_flag, set_rollout, attach_metric, get_readout, kill_flag — the single-object model, one tool per operation.
Evidence for decisions
search_traces, get_trace, service_map, and log search let an agent quote the exact span or line behind a rollback.
Attributable mutations
Every change records the actor (human vs. agent token) and a rationale, surfaced in flag history.
Works in your tools
Runs under Claude Code, Cursor, and any MCP-capable agent — no bespoke integration.
How it works
Three steps, one pipeline.
Add the Foredeck MCP server to your agent.
Create and ramp flags, read metrics, pull trace/log evidence.
Every action is attributed and auditable in flag history.
Better together
Works with the rest of the platform.
Feature Flags
FreeShip every change behind a flag — free and unlimited.
Explore Feature Flags →Guardrails & Auto-Rollback
ProCatch a regressing release before it spreads — automatically.
Explore Guardrails & Auto-Rollback →Tracing
FreeFollow one call across every service — spans stamped with the variation.
Explore Tracing →Logs
FreeSearch logs in release context — OTLP-native, correlated to flags.
Explore Logs →One object, one pipeline. The same rollout you ship is automatically measured, guarded, and — when it breaks — traceable, because every signal is joined to the flag and variation the unit was in.
Get in