Skip to main content
A single-pass reviewer makes the same mistakes on every PR. Tell it once that your codebase uses integer cents and it forgets by the next review. Sigilix’s review memory is built so guidance you give once shapes every review afterward — and so you can see exactly when a learned rule was applied. There are two layers. The headline is Conversational Learnings: rules you teach in plain language. Underneath sits a quieter statistical accept/dismiss corpus that nudges category-level flag-worthiness from how your team engages with findings.

​
Conversational Learnings

Reply to a Sigilix finding the way you’d reply to a human reviewer:
We use integer cents here, not floats — this is intentional.
Sigilix records that as a durable rule for this repo, applies it judiciously in later reviews, and attributes it inline when it does. The next time a finding would touch the same ground, you see why it was shaped:
Applied because of a learned rule: “we use integer cents here.”
A “Learned something new” footer confirms capture at the time you teach it.

​
Teaching and forgetting

Two explicit commands, plus implicit capture from natural replies:
ActionHow
Teach a rule explicitly@sigilix remember we use integer cents, never floats, for money
Forget a rule@sigilix forget <the rule>
Teach implicitlyReply to a finding in plain language (“this is intentional, we always do X”) — Sigilix extracts the rule
Learnings are captured across review surfaces — replying on a finding thread, an @sigilix comment, or a remember command all feed the same store.

​
How a learning shapes a review

A learning is guidance the reviewer reasons with, not a hard mute. When a later review encounters relevant code:
  • Sigilix injects the learned rules into the specialists’ context before they run.
  • A finding that contradicts a learning is reconsidered — often suppressed, sometimes re-framed — rather than re-posted blindly.
  • When a learning changes the outcome, the inline attribution makes it visible, so the memory is never a silent black box.

​
What is protected from learnings

Learnings make reviews quieter where you’ve earned it, but they do not let the repo silence its own safety net. Critical and security-class findings are carved out: a learning cannot suppress a genuine Critical or a security finding. The reviewer still surfaces those regardless of how much guidance has accumulated.

​
Privacy and scope

  • Learnings are scoped per repo. They are not shared across customers or repos.
  • You teach team-level rules (“we use integer cents”), not models of individual reviewers.
  • Forgetting a rule with @sigilix forget removes it from the active set.
Conversational Learnings is the per-customer flywheel: every correction you give compounds into a reviewer that fits your codebase. See Conversational Learnings for the full lifecycle, and Commands for remember / forget.

​
Statistical accept/dismiss corpus (secondary)

Beneath the explicit Learnings layer, Sigilix also watches how your team engages with findings and uses that aggregate signal to calibrate category-level flag-worthiness. This is the older mechanism — useful, but it operates on patterns, not on rules you can state.

​
What’s a feedback signal?

SignalInterpretation
Finding’s suggested patch was committedStrong accept
Reply like “Thanks, fixed” / “Good catch”Accept
Reply like “Not applicable” / “False positive” / 👎 reactionDismiss
PR merged with the finding’s underlying code unchangedSoft dismiss
Reviewer resolved the finding’s threadSoft accept
These are noisy per-finding. The corpus aggregates them across many PRs to find category-level patterns — categories the team consistently dismisses, categories they consistently act on.

​
How the corpus shapes future reviews

It does not silence findings outright. It nudges the flag-worthiness floor per category, per repo: 
Category dismissed 8/10 times on this repo
  → floor raises (only stronger versions surface)

Category acted on 8/10 times on this repo
  → floor stays low (any signal surfaces aggressively)
The corpus is scoped per repo + per category, never per file or per user. Sigilix stores aggregate counts, not the underlying findings or the code they referred to.

​
Carve-outs (same as Learnings)

ConcernReality
”Will it stop catching real bugs?”The floor adjustment is bounded. A Critical or security finding always surfaces regardless of corpus state.
”Will it remember a specific bug?”No. The corpus stores aggregate signals, not findings.
”Can I reset it?”Set { "reviewMemory": { "enabled": false } } to ignore the corpus for a review, then re-enable. A hard reset is a support request.

​
Interaction with other features

FeatureHow they relate
Conversational LearningsThe headline layer. Explicit, repo-scoped, attributed inline. Takes precedence as stated guidance.
profile: assertiveLowers the flag-worthiness floor globally. The statistical corpus still applies on top.
deterministicChecksUnaffected. Memory only adjusts LLM-specialist findings, never regex matches.
Secret scanning, AST rulesUnaffected. Perfect-provenance findings are never down-weighted by memory.
rules.<role>Unaffected. Explicit rules you set are honored.
The general rule: memory adjusts LLM judgment calls, never deterministic findings or explicit safety carve-outs.

Conversational Learnings

Teach a rule in plain language; see it applied and attributed inline.

Confidence Scoring

Proof-tier receipts and the grounding gate that govern what posts.