Step 1
The PR
A developer opens a PR that modifies a protected file.
- Title: Upgrade RDS instance to r6g.xlarge
- File touched: infra/terraform/rds.tf
- Policy match: infra/**
For engineering teams shipping critical changes on GitHub
Install a GitHub App. Mark which paths matter. When a PR touches them without a linked decision, merge is blocked. Setup takes minutes, and the decision record remains permanent.
Product demo coming soon
Every team has stories like these. The why behind critical changes disappears the moment the PR is merged.
Invisible cost
A Terraform change doubles infra cost, and six months later nobody can explain why it was approved.
Buried context
A critical architecture decision lived in Slack, then got buried before the next team needed it.
Missing accountability
A postmortem asks who approved a risky auth change, but the PR history only shows what changed, not why.
This is the exact workflow your team sees when critical code changes are proposed.
Step 1
A developer opens a PR that modifies a protected file.
Step 2
dlogs check fails because no active Decision ID is linked.
Step 3
The developer writes a quick decision record.
Step 4
Decision is linked, check turns green, PR is mergeable.
Step 1
Install the dlogs GitHub App and set policy paths like infra/, db/, or auth/ where mistakes are expensive.
Step 2
When a PR touches a protected path without an active Decision ID, status checks fail and the PR cannot merge.
Step 3
Capture problem, rationale, and tradeoffs in under a minute, attach the Decision ID, and move forward with an auditable trail.
Your team makes decisions in Slack every day. dlogs meets you there — retrieve context during incidents and capture decisions before the thread gets buried.
Something changed in token rotation recently.
@dlogs auth token rotation
DEC-2026-0042 · ACTIVE
Set token rotation to 24h window
Constraint: must not exceed Redis connection pool limit
Supersedes: DEC-2025-0018
View full decision →
Found it. Pool config was never updated after DEC-2026-0119. Fixing now.
OK so we're going with Redis cluster over Memcached. Write-through handles our consistency needs. We accept higher memory cost.
Agreed. Let's record this before it gets buried.
/dlogs record
Decision form opened · AI pre-filled from thread context
Review and edit in the modal, then submit.
Recorded · DEC-2026-0089
Adopt Redis cluster with write-through caching
By @alex.kim · Linked to this thread
View full decision →
Ask a question about any function. dlogs matches it to the exact decision that governs that code, including constraints, tradeoffs, and the full supersede chain.
24h window balances token freshness against Redis connection pool limits. 1h window (DEC-2025-0018) was rejected for causing latency spikes under load.
Constraint: Redis pool max connections
Supersedes: DEC-2025-0018
View full decision →
Decisions are never edited or deleted. New facts supersede old ones; history stays intact.
Tamper-evident, hash-chained records and derived status prevent silent drift from truth.
AI may suggest topics and gaps. It does not own decisions; your team remains accountable author.
GitHub is the first enforcement surface, not the boundary of the product. dlogs is your long-term decision governance layer.
Six months after a decision was captured, it pays for itself when the next team needs to understand why.
Initial RDS sizing for launch
@vamshi · Sep 2025
Upgrade RDS to r6g.xlarge
@priya · Jan 2026
Add read replicas for auth DB
@alex · Mar 2026
Postmortem finding
DEC-2026-0119 added replicas without adjusting pool limits. The binding constraint from DEC-2026-0042 was still in effect. Root cause identified in minutes using the decision chain.
Yes for enforcement in v1. GitHub is our first capture/enforcement surface, not the product boundary.
No. AI may suggest decision-worthy moments or topics; humans author every decision entry.
Most teams can connect GitHub, set their first criticality rule, and run enforcement in about 10 minutes.
Enough context for future teammates to understand the tradeoff: the problem, alternatives considered, final rationale, constraints, and who approved it.
Yes. Type @dlogs in Slack to query decisions instantly, or /dlogs record to capture a new decision with AI-assisted form filling — without leaving the conversation. For your editor, dlogs exposes an MCP server for Cursor and IDE integration.
They become queryable by file, path, function, author, or keyword. Every decision is permanent and linked to the code it governs.
dlogs is built to become one goldmine for code, product, and business decisions with seamless capture across multiple surfaces.
Install free, set your critical paths, and make every high-impact change explainable months later.