Sigilix assistant for Slack

Your team's codebase collaborator.

One assistant for the engineering conversation around your code: grounded answers, repair handoffs, issue triage, PRs, reviews, and CI explanations. Right in Slack.

Create an accountSee the context layerView all tools
#eng-incidents
DatadogApp2:38 AM
Alert: export-service p99 latency > 9s `/v2/export` p99 latency spiked from 340ms to 11.8s after the dashboard launch traffic ramp.
serviceexport-serviceseveritySEV-2windowlast 12m
NK
Nadia K.2:41 AM
@Sigilix pull the traces for `/v2/export` and check whether the billing rate-limit pattern applies here.
SigilixApp2:41 AM
Pulled APM traces since 02:15 UTC. The slow traces all fan into the export builder after auth; billing stays healthy because it already gates by org before the heavy payload work starts.
reposigilix/apiprior artPR #2719, ADR-0031sourceredis-cell sliding window
app/middleware/rateLimit.ts
const { success } = await ratelimit.limit(`billing:${orgId}`, {
  window: "1m",
  max: 200,
});
if (!success) return tooMany(429);
MJ
Marcus J.2:43 AM
Would the same ceiling hold? export payloads are a lot heavier than billing.
Message #eng-incidents
BIUS</>

Connectors

Bring the stack into the conversation.

The assistant reaches across the places your team already leaves evidence, then keeps the answer in the Slack thread where the decision is happening.

Slack
GitHub
Linear
Jira
Notion
CI logs
ADRs
Runbooks

Context

Answers that show why they know.

Code

Repository structure, nearby files, changed lines, merged PRs, and the conventions your reviews already established.

Work

Linear and Jira issues, ownership signals, active branches, review status, and the next gate blocking the task.

Decisions

Slack threads, ADRs, prior adjudications, and the bounded memory that explains why the team chose a path.

Actions

Useful because it can act.

Slack is the interface. The product is the grounded handoff from question to evidence, decision, and repair.

Answer from the repo

@Sigilix any question and get an answer tied to file:line, merged PRs, and ADRs instead of a plausible detached summary.

Open the PR

When the thread lands on a concrete approach, Sigilix turns the decision into a branch, a pull request, and a regression test shaped like the existing suite.

File the issue

A messy thread becomes a clean Linear or Jira ticket with title, severity, likely owner, linked evidence, and missing proof.

Explain failing CI

Forward a red build and Sigilix reads job logs against the diff, then replies with the failure path and likely fix.

Summarize a PR

Ask for the verdict, effort estimate, risk-ranked files, gates passed, and what still needs review.

Run a review

Trigger the full review loop from Slack, post the verdict to GitHub, and keep the thread updated with the useful summary.

Carry team memory

Conventions, prior verdicts, and adjudicated decisions travel with the thread so repeated questions do not restart from zero.

Stay scoped

Access follows your connected app permissions and workspace boundaries. The assistant should show what it used and what it could not see.

Talk to it

@Sigilix <question>grounded answer
@Sigilix why is CI red?log explanation
@Sigilix open a PRbranch + test + PR
@Sigilix file thistriaged issue
@Sigilix review #2847full review loop
@Sigilix summarize #2847verdict + risk

Workflow

From Slack question to shipped repair.

01

Ask in the thread.

Mention Sigilix where the decision is already happening. It answers with the files, PRs, docs, and missing proof that shaped the answer.

02

Steer with the team.

People can redirect the work in Slack: lower a rate limit, preserve an env var, assign an owner, or ask for a narrower PR.

03

Turn the decision into work.

When the next step is concrete, Sigilix opens the PR, files the Linear or Jira issue, runs a review, or explains the failing build.

Put the reviewer, triager, and repair loop in the place your team already makes decisions.

Create an account