API operational·MCP transport streamable-http·Version v1.0·Private early access — now onboarding
Repo Memory & Governance API

One source of truth
for how your repo
is allowed to change.

LaserOwl gives every contributor — human or AI — the full context they need before touching your codebase. Ownership, constraints, surfaces, and risk. Declared once. Enforced everywhere. Build on it or query it directly.

Request AccessSee Use Cases →
api.laserowl.dev/v1
# Any tool — LLM agent or internal app — queries LaserOwl GET /v1/files/payments/retry-logic.tsAuthorization: Bearer cf_live_•••••••• 200 OK ───────────────────────────────── "owner": "payments-team""stability": "extended""rework_rate": 49.3%"surfaces": ["checkout-api", "ledger-api"]"constraints": [{  "rule": "PCI compliance review required"}]"audit": { "agent": "claude-mcp", "checked": true } logged · constraint checked · owner notified
4
Context dimensions
1
API call. Full context.
0
Rebuilt from scratch
§ 01The problemLLMs and internal tools both contribute blind without LaserOwl
Without LaserOwl
✗The agent sees the file. That's it.
✗No ownership context
✗No declared constraints
✗No surface awareness
✗No churn or stability signals
✗No audit trail
Contributes blind. Breaks things it couldn't have known about. Leaves no trace.
With LaserOwl API or MCP
✓The agent sees the file
✓Knows the owner and team
✓Checks constraints before writing
✓Knows which surfaces are affected
✓Understands risk — churn, stability, rework rate
✓Every action logged with full context
Contributes correctly. Every constraint checked. Every interaction provable.
§ 02What LaserOwl knowsFour dimensions of repo memory, queryable via API + MCP
01
Ownership

Inferred from real contribution history. Confirmed and declared by your team. Always current — not a stale CODEOWNERS file.

Inferred + declared
02
Constraints

Declare rules once — PCI review, load testing, security sign-off. Any agent or tool querying LaserOwl inherits them automatically.

Declared + enforced
03
Surfaces

Map which files feed which APIs, services, and external dependencies. Know the blast radius before anything is touched.

Declared + queryable
04
Lifecycle

Every file's full history — churn, stability mode, rework rate, coordination overhead. Risk signals, not just current state.

Scanned + tracked
§ 03Use casesLLM agents · Internal apps · Governance tooling
Use case 01 — LLM agents

Claude with full repo context, not just the open file.

Without LaserOwl, an LLM sees the file it's editing. With LaserOwl MCP connected, it sees the ownership, constraints, surface dependencies, and risk signals before writing a single line. It contributes like someone who actually knows the codebase.

What Claude knows via LaserOwl MCP
# Developer asks Claude to refactor payments/retry-logic.ts # Claude queries LaserOwl before writing:"owner": "payments-team" → will notify"stability": "extended" → high risk file"surfaces": ["checkout-api"] → downstream impact"constraints": PCI review required → flagged # Claude responds with full awareness, not blind confidence
✓Claude flags the PCI constraint, notifies the payments team, and proceeds with the right context — before a line is written, not after a PR is rejected.
§ 04How it worksConnect once. REST API + MCP. Works for humans and AI alike.
LaserOwl sits between your contributors — human or AI — and your codebase. Before anything is written, context is checked. Every interaction is logged.
1
Connect your repo

Install the GitHub app. LaserOwl scans your full repo history and builds the memory layer — ownership signals, lifecycle data, risk scores. No configuration needed to start.

2
Declare your constraints and surfaces

Add the rules your team already knows but has never encoded. PCI review on payments. Load test required on core services. Ownership confirmed by team leads.

POST /v1/constraints"file": "payments/processor.ts""rule": "PCI compliance review""enforced_by": "payments-lead"
3
Connect your tools — LLM agents or internal apps

Point Claude or any MCP-compatible agent at mcp.laserowl.dev/v1. Or call the REST API from any internal tool, PR automation, or review workflow.

4
Every interaction logged automatically

No setup required. LaserOwl logs every query — what was checked, what constraints applied, what agent or tool acted. Your audit trail is structural, not aspirational.

§ 05Audit trailEvery AI touch. Logged. Provable.
100%
of AI
touches
logged.

Mandate LaserOwl MCP for internal AI tooling and you get a complete record of every AI contribution — which agent, which file, which constraints were checked. Not a log you have to build. A log that exists because LaserOwl sat in the middle.

Get Early Access
TimeActionAgentStatus
09:23:11Zpayments/retry-logic.ts — constraint checkclaude-mcppassed
09:23:14Zpayments/processor.ts — PCI rule hitclaude-mcpflagged
09:41:02Zauth/session.ts — owner notifiedcursorpassed
10:15:33Zbilling/invoices.ts — constraint checkclaude-mcppassed
10:22:07Zcore/router.ts — no constraintscopilotpassed
10:38:44Zinfra/deploy.ts — owner checkclaude-mcppassed
§ 06API referenceapi.laserowl.dev/v1 · Bearer auth · MCP at mcp.laserowl.dev/v1
Files — lifecycle, risk signals, cochange graph
GET/v1/filesList all tracked files
GET/v1/files/{path}Full record — owner, constraints, surfaces, risk signals
GET/v1/files/{path}/lifecycleFull lifecycle history
GET/v1/files/{path}/cochangeFiles that historically change together

Set the rules once.
Enforce them everywhere.

LaserOwl is in private early access. We're onboarding platform engineering teams now. Bring your repo — we'll have it running in a day.

Request Access →

Works with Claude, Cursor, Copilot, and any MCP-compatible agent. REST API + MCP. Deployed on Railway. GitHub app included. Your data stays in your stack.

hi@laserowl.dev
laserowl.dev