// DETECTIVE WORKFLOW INDUCTION MANUAL

HOW TO INTEGRATE CAPI IN 30 SECONDS

**No cloning required.** Capi is distributed as a standalone CLI and standard Git pre-commit hook. Install it directly into your existing microservices or CI/CD pipelines with a single command.

ENGINE: COGNEE OPEN SOURCE (Local FastEmbed + SQLite)

CLOUD READY: COGNEE CLOUD API (Tenant Connected)

// STEP 1 OF 3: ONE-LINE INSTALLATION & GIT HOOKS

CHOOSE YOUR INSTALLATION METHOD

ZERO REPO CLONING NEEDED

You do **not** need to clone the Capi codebase into your workspace. Select how your team manages developer tooling:

bash — run inside any of your existing project repositories

# Install globally or in your venv directly from GitHub:

$ pip install git+https://github.com/Akarshkushwaha/Capi.git

# Enable the guardrail hook in your current Git repo:

$ capi install-hook

✓ Guardrail hook deployed to .git/hooks/pre-commit

// INTERACTIVE TEST: SIMULATE A DANGEROUS COMMIT IN YOUR REPO

// STEP 2 OF 3: CODEBASE & GIT HISTORY INGESTION

CONNECT YOUR REAL REPOSITORIES & PULL REQUESTS

Once installed, run `capi ingest` in any project folder. Capi’s GitPython engine parses historical commit logs, extracts variable declarations, and constructs a causal graph in Cognee linking engineers to every variable change.

bash — terminal

# 1. Scan your current Git project commits and .env files into Cognee graph memory:

$ capi ingest . --service payments-api

# 2. Extract GitHub Pull Request review reasoning & architecture discussions:

$ capi ingest-prs --repo org/my-service --token $GITHUB_TOKEN

// STEP 3 OF 3: SELF-IMPROVING FEEDBACK LOOPS (`cognee.improve`)

AUTOMATE CI/CD AUDITS & SLACK OUTAGE REWARDS

CONTINUOUS LEARNING

Capi gets smarter every time your team ships code. Integrate our feedback commands into your CI/CD deployment scripts or Slack bot incident workflows. When a deployment is clean, Capi reduces the variable's Danger Score. When an outage occurs, Capi immediately penalizes the score so no engineer repeats the mistake.

POSITIVE FEEDBACK (SAFE DEPLOYMENT)

Run after a successful Kubernetes or Docker deployment in your CI/CD script:

$ capi safe DB_POOL_SIZE --service payments-api

# Result: Danger score reduced (-10) in Cognee memory.

NEGATIVE FEEDBACK (INCIDENT POST-MORTEM)

Run when a variable causes a P1/P2 outage or via Slack emergency webhook:

$ capi incident DB_POOL_SIZE --service payments-api --severity P1

# Result: Danger score penalized (+20) in Cognee memory.

2-WAY SLACK & WEBHOOK INTEGRATION (NO PROPRIETARY APP REQUIRED)

Because Capi is built on open REST standards, you don't need a heavy Slack app installed. Any Slack slash command, incident bot (PagerDuty / Rootly / Jira), or CI/CD script can connect via standard HTTP webhooks:

1. Inbound: Slack Bot → Capi (/incident)

When an outage is reported in Slack #incidents, trigger Capi's negative feedback loop:

$ curl -X POST https://capi-backend.onrender.com/incident \
-H "Content-Type: application/json" \
-d '{"key":"DB_POOL_SIZE","service":"payments-api","notes":"Reported via Slack","severity":"P1"}'

2. Outbound: CI/CD Guardrail → Slack (#deployments)

When Capi blocks a dangerous PR, send an alert to your Slack channel:

$ curl -X POST -H "Content-type: application/json" \
--data '{"text":"🚨 [Capi Guardrail Blocked PR]: DB_POOL_SIZE (20) breaches safe boundary (5 ≤ value ≤ 15). Caused Outage INC-47."}' \
$SLACK_WEBHOOK_URL

READY TO BECOME A CONFIG ARCHAEOLOGY DETECTIVE?

Start your first investigation right now on Mission Control or audit your team's staged configuration changes in seconds.

OPEN MISSION CONTROL RUN PRE-CHANGE SAFETY CHECK