Sentinel — Competitor Analysis

Competitor analysis — Sentinel by Aftergraph (governance-first)

Wedge guardrail: Sentinel turns pull requests into merge-ready verdicts. First wedge: CLI review on exact HEAD → SHIP / DO NOT SHIP verdict with cited evidence → GitHub App.

Evidence status: vendor-doc quotes below marked [VERIFY] must be confirmed against first-party docs before any public publish. No claim here has been checked against live vendor pages.

Categories (structural, not “us vs everyone”)

AI PR reviewers — CodeRabbit, Greptile, Copilot code review. Direct layer.
Rules-based quality gates — SonarQube/SonarCloud, CodeScene. Adjacent: compliance buyers.
General coding agents with review features — Claude Code, Codex, Jules. Suppliers/partners, not competitors.

Per competitor (governance → distribution → lock-in)

CodeRabbit

Governance: Probabilistic comments on diffs; no exact-head verdict contract; no tamper-evident audit chain. [VERIFY] against docs.
Distribution: GitHub App marketplace — strongest in category.
Lock-in: Review history lives in their cloud; verdicts not exportable as evidence.
One word they cannot own: “verified”. Their business model is comment volume per PR; a hard exact-head guarantee would shrink their output and slow their UX.
Tag: moat (their model contradicts verification guarantees).

Greptile

Governance: Custom rules + AI summaries; verdict authority unclear. [VERIFY].
Distribution: GitHub App, devtool-led growth.
Lock-in: Custom rule packs locked to platform.
One word they cannot own: “evidence”. Rules-as-product means findings cite rules, not the verified commit.
Tag: moat-leaning (custom-rules model resists per-commit evidence chains).

Copilot code review (GitHub)

Governance: Suggestions inside the GitHub permission model; inherits repo roles, no independent verdict layer. [VERIFY].
Distribution: Bundled with Copilot seats — unbeatable reach.
Lock-in: GitHub + Microsoft model stack.
One word they cannot own: “independent”. They cannot credibly overrule the platform that hosts them.
Tag: moat (bundling contradicts independence — our wedge for regulated buyers).

SonarQube / SonarCloud

Governance: Deterministic rules, quality gates that block merges — the closest to verdicts, but rules-only, no semantic review. Strong compliance story.
Distribution: Self-host + cloud; entrenched in enterprise.
Lock-in: Quality profiles and history in their format.
One word they cannot own: “semantic”. Rules engines cannot judge intent, only patterns.
Tag: feature gap for them, partner surface for us (Sentinel verdicts + Sonar gates compose; do not compete head-on before distribution exists).

CodeScene

Governance: Behavioral code analysis (hotspots, coupling) — evidence-flavored metrics, advisory not verdicts. [VERIFY].
Distribution: Niche devtool channel.
Lock-in: Historical analysis data.
One word they cannot own: “merge-decision”. Advisory by design.
Tag: feature (complementary; possible integration, not competition).

Consequences for us (6-12 month window)

If GitHub ships exact-head verdict semantics into Copilot review, our independence moat compresses to regulated/self-hosted buyers only. Mitigation: ship the self-host CLI first ( portable, no platform permission needed) so the product survives any single platform’s move. Revisit this section quarterly.

Dropped deliberately

Pricing tables (stale in months), feature checklists (category placement instead), “better at X” without cited benchmarks.

Source: Aftergraph/sentinel@beececfb4885d8055cc6694e8979b47759a212a4 docs/competitor-analysis.md.md — verbatim, pinned, no reinterpretation.

Was this page helpful?