Back to AI Pods
Techvify Forge Insights

See how your team really builds.

Measure how you deliver — phase adherence from agent telemetry, plus DORA delivery metrics and cost visibility, fed back into PLAN.

Forge Insights is the measure half of the Forge pair — where Forge DevOps does the delivery, Insights observes how it went. One command bootstraps agent telemetry; then it scores how well each developer and task implemented the six phases from real session events, computes the four DORA delivery metrics from your CI and delivery controller, and turns live cost data into a showback — routing the gaps back into PLAN. Built on the telemetry stack already in place.

Setup

Setup steps live in the guidebook.

The guidebook carries the current install and configuration steps for every kit, kept in one place so they stay accurate. Open to Techvify users and invited guests.

What’s inside

  • One-command setup

    One command bootstraps the telemetry — it installs the OTel configuration plus a per-task session hook into your agent settings, so sessions start flowing to the metrics and log hub. No manual rollout script to hunt down.

  • Phase adherence, from real activity

    The reports read your session events — skill activations, command usage, subagents, tool decisions — and an analysis agent maps them onto the six phases to show what was followed vs skipped.

  • Per-task, by branch

    The setup hook tags every session with a task id (your tracker’s key or the branch slug) and repo, so you can ask "did PROJ-456 go through TEST before DEPLOY?" — not just team-wide aggregates.

  • DORA delivery metrics

    Computes the four DORA metrics — deployment frequency, lead time for change, change-failure rate, time to restore — from your CI and delivery controller, classifies the team Elite/High/Medium/Low per metric, and names the one bottleneck to fix next. States data gaps honestly rather than guessing.

  • FinOps cost showback

    Turns the live cost data already flowing into your metrics store into a showback — dollars per namespace or cluster and the biggest idle or over-provisioned waste, ranked. Showback only; it never provisions or deletes.

  • Closes the loop into PLAN

    Each process gap is drafted as a backlog item with acceptance criteria — a spec stub or an issue for your tracker — and your team decides which ones to file into PLAN.

  • What the telemetry records

    Prompt text is never sent. What is sent is activity: who ran which skill, command or tool, and when — including the tool arguments, which for an edit can contain part of the code being changed. That is what lets each phase be traced to its task.

  • Built on what already runs

    Reuses the telemetry stack already in place and its three dashboards (Metrics, Logs, Per-Task Insights) — Forge Insights adds the AI analysis layer, not a parallel pipeline.

  • Self-service rollout diagnostics

    A doctor command checks your local install against the latest setup and the collector endpoint; a verify command confirms your metrics and logs actually reached the hub with clean task and repo enrichment — so a broken rollout self-triages instead of going machine-by-machine.

What a report looks like

Phase adherence, from real activity.

The report pulls a dev or team’s agent telemetry over a window and an analysis agent maps the session events onto the six phases — showing what was exercised, what wasn’t observed, and what to feed back into PLAN. “Not observed” means no telemetry, not an accusation.

Illustrative output — representative activity, not a real report.

See a full adherence report

The complete report — phase-by-phase activity, what was never observed, and the recommendations it routes back into PLAN — is open to Techvify users and invited guests.